Downdetector\u003C/a> showing over 15,000 user complaints. Amazon's own services, like the Alexa voice assistant, are also reported to be unresponsive.\u003C/li>\u003Cli class=\"content-block-list-item\">The outage has impacted a vast array of major companies, highlighting the internet's deep reliance on AWS.\u003C/li>\u003Cli class=\"content-block-list-item\">Prominent services experiencing downtime or performance issues include \u003Cb>Perplexity, Coinbase, Signal, Airtable, Canva, Snapchat, Fortnite,\u003C/b> and \u003Cb>Zoom\u003C/b>.\u003C/li>\u003C/ul>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">\u003Ca href=\"/tag/news\" target=\"_blank\" rel=\"noopener noreferrer ugc\">#news\u003C/a>\u003Ca href=\"/tag/#aws\" target=\"_blank\" rel=\"noopener noreferrer ugc\"> #aws\u003C/a>\u003C/p>\u003C/div>\u003C/div>",{"time":417,"blocks":418,"version":123},1760951029847,[419,424,428,432,436,453],{"id":420,"type":19,"data":421},"img_aws_main",{"caption":63,"withBorder":22,"withBackground":22,"stretched":22,"file":422},{"url":423},"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/aa4ddfc6-7097-4fe6-9b82-fd796b569a0d/1184.webp",{"id":425,"type":14,"data":426},"intro_paragraph",{"text":427},"Amazon Web Services (AWS), a cornerstone of the modern internet, is experiencing a significant outage that originated in its US-EAST-1 region. The issue has cascaded globally, affecting thousands of websites and online services that rely on Amazon's cloud infrastructure.",{"id":429,"type":60,"data":430},"quote_aws_status",{"quoteText":431,"authorText":63},"The official \u003Ca href=\"https://health.aws.amazon.com/health/status\">AWS health status page\u003C/a> confirms that several services are facing \"operational issues,\" and the company is actively working to resolve them.",{"id":433,"type":27,"data":434},"h3_impact_title",{"text":435,"level":30},"The Ripple Effect",{"id":437,"type":37,"data":438},"list_impact_details",{"style":39,"meta":439,"items":440},{},[441,445,449],{"content":442,"meta":443,"items":444},"Reports of service failures have skyrocketed, with \u003Ca href=\"https://downdetector.com/status/amazon/\">Downdetector\u003C/a> showing over 15,000 user complaints. Amazon's own services, like the Alexa voice assistant, are also reported to be unresponsive.",{},[],{"content":446,"meta":447,"items":448},"The outage has impacted a vast array of major companies, highlighting the internet's deep reliance on AWS.",{},[],{"content":450,"meta":451,"items":452},"Prominent services experiencing downtime or performance issues include \u003Cb>Perplexity, Coinbase, Signal, Airtable, Canva, Snapchat, Fortnite,\u003C/b> and \u003Cb>Zoom\u003C/b>.",{},[],{"id":454,"type":14,"data":455},"rnk-txUUBR",{"text":456},"#news #aws",{"authorId":458,"name":459,"avatar":460,"username":461},"Rv94a6RTPrfQnwzLZ2ofJwtBCgU2","David M.","https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/0aa56e19-853e-49cc-b06b-0f9268e9a39b/1184.webp","id354379078763","SERVICES",[464],{"emojiId":214,"count":137,"reactedByUser":22},{"comments":136,"favorites":141,"views":466,"hits":467},28,54,"2025-10-20T09:02:16.841","2025-10-20T14:44:18.725",{"postId":471,"title":472,"slug":473,"preview":474,"html":475,"content":476,"author":612,"authorUsername":616,"topic":617,"status":138,"reactions":618,"counters":623,"createdAt":626,"updatedAt":627,"publishedPostId":145,"favorite":22},"c601e889-04a0-4f6e-9dde-fbc36d32e2d1","Career Tips for Beginner Programmers","296706-career-tips-for-beginner-programmers","\u003Cdiv class=\"block-wrapper--media\">\u003Cfigure class=\"block-content content-block-image\">\u003Cimg src=\"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/09b7e7f0-c4fc-4adb-93a6-76ead8c8e2a5/1184.webp\" alt=\"Image\" class=\"rounded-xl medium-zoom-image\" loading=\"lazy\">\u003C/figure>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">Remember, we operate in a capitalist system. This means money is the primary driver. Don't be ashamed to discuss your salary, bonuses, or compensation for overtime. It’s not a 'dirty' topic; it's as fundamental as any other term of your employment.\u003C/p>\u003C/div>","\u003Cdiv class=\"content-wrapper break-words\">\u003Cdiv class=\"block-wrapper--media\">\u003Cfigure class=\"block-content content-block-image\">\u003Cimg src=\"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/09b7e7f0-c4fc-4adb-93a6-76ead8c8e2a5/1184.webp\" alt=\"Image\" class=\"rounded-xl medium-zoom-image\" loading=\"lazy\">\u003C/figure>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch2 class=\"block-content\">Embrace the Rules of the Game\u003C/h2>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">Remember, we operate in a capitalist system. This means money is the primary driver. Don't be ashamed to discuss your salary, bonuses, or compensation for overtime. It’s not a 'dirty' topic; it's as fundamental as any other term of your employment.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch2 class=\"block-content\">Understand Your Employer's Motivation\u003C/h2>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">Your employer doesn't \u003Ci>love\u003C/i> or \u003Ci>respect\u003C/i> you in a personal sense; you are there to help them generate revenue. Perks like a fancy office, a gym membership, and free snacks are designed to boost morale and retention while keeping salaries competitive, not excessive. Don't fall for vague promises about \u003Ci>professional growth\u003C/i> or a \u003Ci>family-like team\u003C/i>. Instead, focus on two things:\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cul class=\"block-content content-block-list-unordered\">\u003Cli class=\"content-block-list-item\">Your genuine interest in the work\u003C/li>\u003Cli class=\"content-block-list-item\">Your base salary\u003C/li>\u003C/ul>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">Everything else is secondary. And if you encounter genuinely unpleasant colleagues, remember you can always find a better environment elsewhere.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch2 class=\"block-content\">Trust the Contract, Not the Promises\u003C/h2>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">Performance reviews, salary adjustments, and promises of advancement are often just talk. Verbal commitments are free and non-binding. The company's immediate goal is to get you in the door; the office perks are there to keep you from leaving.\u003C/p>\u003C/div>\u003Cfigure class=\"block-wrapper block-wrapper--media\">\u003Cdiv class=\"block-wrapper__content\">\u003Cblockquote class=\"block-quote block-quote--text-small\">\u003Cdiv class=\"block-quote__line\">\u003C/div>\u003Cdiv class=\"block-quote__text\">The fastest way to increase your income is almost always by switching jobs\u003C/div>\u003C/blockquote>\u003C/div>\u003C/figure>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">This isn't to say employers are deceptive if they don't offer annual raises or shower you with bonuses. Those things are never guaranteed, no matter what was said. The only document that matters is your employment contract — read it carefully.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch2 class=\"block-content\">Corporations Aren't Evil, They're Machines\u003C/h2>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">This doesn't mean corporations are evil. They are simply machines designed to achieve their primary goal: making money. Paying a programmer more doesn't magically make them smarter or more productive. From a purely business perspective, increasing salaries reduces profits. If you can help that machine run better and generate more revenue, you might get a share of the profits. But that path often leads toward management and finance — a world where you're laying off 500 people instead of writing C++.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch2 class=\"block-content\">True Professionalism is Meeting Deadlines\u003C/h2>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">Professionalism in software development isn't about writing perfect code; it's about delivering a functional product on time. Your boss has made a commitment to a client, and that client needs the result by a specific date to stay competitive. A professional delivers on schedule. All other qualities are secondary.\u003C/p>\u003C/div>\u003Cfigure class=\"block-wrapper block-wrapper--media\">\u003Cdiv class=\"block-wrapper__content\">\u003Cblockquote class=\"block-quote block-quote--text-small\">\u003Cdiv class=\"block-quote__line\">\u003C/div>\u003Cdiv class=\"block-quote__text\">A flawed result delivered on time is more valuable than a perfect one delivered late\u003C/div>\u003C/blockquote>\u003C/div>\u003C/figure>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">An \u003Ci>artisan\u003C/i> who delivers pristine code after the deadline is just an artist, not a professional. A true pro knows how to version their work, save improvements for the next iteration, and stick to the agreed-upon plan, even when they see a 'better' way.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch2 class=\"block-content\">Appreciate the Magic of Instant Creation\u003C/h2>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">One of the best psychological perks of programming is the incredibly short feedback loop between an idea and its execution. Unlike building a skyscraper or designing a circuit board, you can have a working result in minutes. A quick recompile, and your vision comes to life. Cherish this unique power — few other fields offer such immediate gratification.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch2 class=\"block-content\">The Myth of the Management Career Path\u003C/h2>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">Don't fall into the trap of thinking that the only career path for a developer leads to management. Becoming a team lead often means endless meetings and a slow decay of your technical skills. The world remembers great developers like Linus Torvalds, Fabrice Bellard and many others as not great team leads. While we remember figures like Steve Jobs and Bill Gates, they were visionary founders, not middle managers.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch2 class=\"block-content\">Stay Curious: The Power of Pet Projects\u003C/h2>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">If you've chosen to remain a developer, it's likely because you genuinely love programming. Embrace that passion! Build personal projects that truly interest you, not just what's trending. The goal is to learn and understand things more deeply. Most importantly: \u003Cb>finish them\u003C/b>. A pet project should end with a working result, whether it's a private tool or a public GitHub repo. An unfinished project is a disservice to your own time and effort.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Chr class=\"block-content content-block-delimiter\">\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">There are no shortcuts to genius. Great work isn't coded overnight in a caffeine-fueled haze. The real superpower is the ability to break down a massive problem into small, manageable steps and execute them methodically.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch2 class=\"block-content\">Reinvent the Wheel. Relentlessly.\u003C/h2>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">Ignore the naysayers who cry, \u003Ci>You're just reinventing the wheel!\u003C/i> These critics rarely create anything themselves. Learn to distinguish between constructive feedback and baseless negativity. Constructive feedback comes from people who have built things. The phrase 'show me what you've built first' is not just trolling; it’s a valid filter.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">The dirty secret is that most successful startups are just better-built wheels. A successful new business is often an old one that saves the customer time or money. Building something in a field with no competitors is often a sign that there's no demand. So, embrace the wheel! Build your own programming languages, encryption algorithms, and game engines. The market, not armchair critics, will ultimately decide what is valuable. The world has enough commentators; it needs more creators.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch2 class=\"block-content\">Never Stop Building\u003C/h2>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">Even if you do end up in management, never stop building. Your side projects might evolve into business ventures, but the principle remains the same: always be creating. Your employer is not responsible for your personal development. You owe it to yourself to find your own path and keep growing.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">Good luck.\u003C/p>\u003C/div>\u003C/div>",{"time":477,"blocks":478,"version":123},1760681953391,[479,484,488,492,496,500,513,517,521,525,529,533,537,541,545,549,553,557,561,565,569,573,577,581,584,588,592,596,600,604,608],{"id":480,"type":19,"data":481},"4iG937-Dnr",{"caption":63,"withBorder":22,"withBackground":22,"stretched":22,"file":482},{"url":483},"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/09b7e7f0-c4fc-4adb-93a6-76ead8c8e2a5/1184.webp",{"id":485,"type":27,"data":486},"header_1",{"text":487,"level":137},"Embrace the Rules of the Game",{"id":489,"type":14,"data":490},"rules_para_1",{"text":491},"Remember, we operate in a capitalist system. This means money is the primary driver. Don't be ashamed to discuss your salary, bonuses, or compensation for overtime. It’s not a 'dirty' topic; it's as fundamental as any other term of your employment.",{"id":493,"type":27,"data":494},"header_2",{"text":495,"level":137},"Understand Your Employer's Motivation",{"id":497,"type":14,"data":498},"motivation_para_1",{"text":499},"Your employer doesn't \u003Ci>love\u003C/i> or \u003Ci>respect\u003C/i> you in a personal sense; you are there to help them generate revenue. Perks like a fancy office, a gym membership, and free snacks are designed to boost morale and retention while keeping salaries competitive, not excessive. Don't fall for vague promises about \u003Ci>professional growth\u003C/i> or a \u003Ci>family-like team\u003C/i>. Instead, focus on two things:",{"id":501,"type":37,"data":502},"motivation_list_1",{"style":39,"meta":503,"items":504},{},[505,509],{"content":506,"meta":507,"items":508},"Your genuine interest in the work",{},[],{"content":510,"meta":511,"items":512},"Your base salary",{},[],{"id":514,"type":14,"data":515},"motivation_para_2",{"text":516},"Everything else is secondary. And if you encounter genuinely unpleasant colleagues, remember you can always find a better environment elsewhere.",{"id":518,"type":27,"data":519},"header_3",{"text":520,"level":137},"Trust the Contract, Not the Promises",{"id":522,"type":14,"data":523},"contract_para_1",{"text":524},"Performance reviews, salary adjustments, and promises of advancement are often just talk. Verbal commitments are free and non-binding. The company's immediate goal is to get you in the door; the office perks are there to keep you from leaving.",{"id":526,"type":60,"data":527},"contract_quote_1",{"quoteText":528,"authorText":63},"The fastest way to increase your income is almost always by switching jobs",{"id":530,"type":14,"data":531},"contract_para_2",{"text":532},"This isn't to say employers are deceptive if they don't offer annual raises or shower you with bonuses. Those things are never guaranteed, no matter what was said. The only document that matters is your employment contract — read it carefully.",{"id":534,"type":27,"data":535},"header_4",{"text":536,"level":137},"Corporations Aren't Evil, They're Machines",{"id":538,"type":14,"data":539},"machines_para_1",{"text":540},"This doesn't mean corporations are evil. They are simply machines designed to achieve their primary goal: making money. Paying a programmer more doesn't magically make them smarter or more productive. From a purely business perspective, increasing salaries reduces profits. If you can help that machine run better and generate more revenue, you might get a share of the profits. But that path often leads toward management and finance — a world where you're laying off 500 people instead of writing C++.",{"id":542,"type":27,"data":543},"header_5",{"text":544,"level":137},"True Professionalism is Meeting Deadlines",{"id":546,"type":14,"data":547},"deadlines_para_1",{"text":548},"Professionalism in software development isn't about writing perfect code; it's about delivering a functional product on time. Your boss has made a commitment to a client, and that client needs the result by a specific date to stay competitive. A professional delivers on schedule. All other qualities are secondary.",{"id":550,"type":60,"data":551},"deadlines_quote_1",{"quoteText":552,"authorText":63},"A flawed result delivered on time is more valuable than a perfect one delivered late",{"id":554,"type":14,"data":555},"deadlines_para_2",{"text":556},"An \u003Ci>artisan\u003C/i> who delivers pristine code after the deadline is just an artist, not a professional. A true pro knows how to version their work, save improvements for the next iteration, and stick to the agreed-upon plan, even when they see a 'better' way.",{"id":558,"type":27,"data":559},"header_6",{"text":560,"level":137},"Appreciate the Magic of Instant Creation",{"id":562,"type":14,"data":563},"magic_para_1",{"text":564},"One of the best psychological perks of programming is the incredibly short feedback loop between an idea and its execution. Unlike building a skyscraper or designing a circuit board, you can have a working result in minutes. A quick recompile, and your vision comes to life. Cherish this unique power — few other fields offer such immediate gratification.",{"id":566,"type":27,"data":567},"header_7",{"text":568,"level":137},"The Myth of the Management Career Path",{"id":570,"type":14,"data":571},"management_para_1",{"text":572},"Don't fall into the trap of thinking that the only career path for a developer leads to management. Becoming a team lead often means endless meetings and a slow decay of your technical skills. The world remembers great developers like Linus Torvalds, Fabrice Bellard and many others as not great team leads. While we remember figures like Steve Jobs and Bill Gates, they were visionary founders, not middle managers.",{"id":574,"type":27,"data":575},"header_8",{"text":576,"level":137},"Stay Curious: The Power of Pet Projects",{"id":578,"type":14,"data":579},"pets_para_1",{"text":580},"If you've chosen to remain a developer, it's likely because you genuinely love programming. Embrace that passion! Build personal projects that truly interest you, not just what's trending. The goal is to learn and understand things more deeply. Most importantly: \u003Cb>finish them\u003C/b>. A pet project should end with a working result, whether it's a private tool or a public GitHub repo. An unfinished project is a disservice to your own time and effort.",{"id":582,"type":273,"data":583},"delimiter_1",{},{"id":585,"type":14,"data":586},"short_thought_1",{"text":587},"There are no shortcuts to genius. Great work isn't coded overnight in a caffeine-fueled haze. The real superpower is the ability to break down a massive problem into small, manageable steps and execute them methodically.",{"id":589,"type":27,"data":590},"header_9",{"text":591,"level":137},"Reinvent the Wheel. Relentlessly.",{"id":593,"type":14,"data":594},"wheel_para_1",{"text":595},"Ignore the naysayers who cry, \u003Ci>You're just reinventing the wheel!\u003C/i> These critics rarely create anything themselves. Learn to distinguish between constructive feedback and baseless negativity. Constructive feedback comes from people who have built things. The phrase 'show me what you've built first' is not just trolling; it’s a valid filter.",{"id":597,"type":14,"data":598},"wheel_para_2",{"text":599},"The dirty secret is that most successful startups are just better-built wheels. A successful new business is often an old one that saves the customer time or money. Building something in a field with no competitors is often a sign that there's no demand. So, embrace the wheel! Build your own programming languages, encryption algorithms, and game engines. The market, not armchair critics, will ultimately decide what is valuable. The world has enough commentators; it needs more creators.",{"id":601,"type":27,"data":602},"header_10",{"text":603,"level":137},"Never Stop Building",{"id":605,"type":14,"data":606},"final_para_1",{"text":607},"Even if you do end up in management, never stop building. Your side projects might evolve into business ventures, but the principle remains the same: always be creating. Your employer is not responsible for your personal development. You owe it to yourself to find your own path and keep growing.",{"id":609,"type":14,"data":610},"TI4-WsN5lM",{"text":611},"Good luck.",{"authorId":613,"name":614,"avatar":615,"username":616},"xLAuHnTP0HTqGmxrA8LDZuGKvt33","Thomas","https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/6a600019-bf7d-449f-b8a8-bf545d62ae35/1184.webp","id131567261828","CAREER",[619,620,621],{"emojiId":141,"count":141,"reactedByUser":22},{"emojiId":30,"count":30,"reactedByUser":22},{"emojiId":622,"count":141,"reactedByUser":22},12,{"comments":141,"favorites":141,"views":624,"hits":625},23,207,"2025-10-17T06:09:29.073","2025-10-18T08:39:58.353",{"postId":629,"title":630,"slug":631,"preview":632,"html":633,"content":634,"author":857,"authorUsername":131,"topic":132,"status":138,"reactions":858,"counters":861,"createdAt":863,"updatedAt":864,"publishedPostId":145,"favorite":22},"7742aec7-9984-4bfc-8314-2ca36590d7c6","Review of Ryzen 5 9500F: The Cheapest Zen 5","825043-review-of-ryzen-5-9500f-the-cheapest-zen-5","\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">The AM5 platform has officially gone mainstream. AMD has successfully populated the market not only with high-end CPUs for enthusiasts but also with affordable motherboards and processors starting around $120. However, budget-conscious builders have been limited to previous-generation Zen 4 chips like the Ryzen 5 7500F. While effective, these 2022-era CPUs are starting to show their age.\u003C/p>\u003C/div>","\u003Cdiv class=\"content-wrapper break-words\">\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">The AM5 platform has officially gone mainstream. AMD has successfully populated the market not only with high-end CPUs for enthusiasts but also with affordable motherboards and processors starting around $120. However, budget-conscious builders have been limited to previous-generation Zen 4 chips like the Ryzen 5 7500F. While effective, these 2022-era CPUs are starting to show their age.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">Now, AMD is shaking things up by bringing its latest architecture to the budget segment. The new \u003Cb>Ryzen 5 9500F\u003C/b> is the successor to the popular 7500F, built on the cutting-edge Zen 5 architecture. This upgrade promises a significant performance boost where it matters most for gamers.\u003C/p>\u003C/div>\u003Cfigure class=\"block-wrapper block-wrapper--media\">\u003Cdiv class=\"block-wrapper__content\">\u003Cblockquote class=\"block-quote block-quote--text-small\">\u003Cdiv class=\"block-quote__line\">\u003C/div>\u003Cdiv class=\"block-quote__text\">AMD claims the Zen 5 architecture delivers a 16% improvement in instructions per clock (IPC) over Zen 4, translating to a 9-11% frame rate increase in esports and CPU-intensive AAA titles.\u003C/div>\u003C/blockquote>\u003C/div>\u003C/figure>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">Interestingly, the Ryzen 5 9500F isn't about making AM5 systems cheaper. It's priced slightly higher than its predecessor. Instead, AMD's goal is to elevate the performance of budget AM5 builds, making them a more compelling alternative to Intel's popular Core i5 series, which remains a strong competitor even after the launch of the Arrow Lake family.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">In this review, we'll determine if AMD has crafted a worthy rival to the Core i5-14400F and i5-14600KF, and whether the new Ryzen 5 9500F has what it takes to become the next go-to processor for budget-savvy builders.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch2 class=\"block-content\">Ryzen 5 9500F: Key Features\u003C/h2>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">The Ryzen 5 9500F is a straightforward six-core, twelve-thread processor based on the Zen 5 architecture, featuring a 32 MB L3 cache. Its specifications are very similar to its predecessor, the Ryzen 5 7500F, with the primary difference being the upgraded core architecture. However, there are a few key distinctions:\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cul class=\"block-content content-block-list-unordered\">\u003Cli class=\"content-block-list-item\">\u003Cb>No Integrated Graphics:\u003C/b> The F suffix indicates the lack of an integrated GPU, which is a non-issue for gamers who will be using a discrete graphics card.\u003C/li>\u003Cli class=\"content-block-list-item\">\u003Cb>Clock Speeds:\u003C/b> The 9500F has a maximum boost clock of 5.0 GHz, which is 400 MHz lower than the Ryzen 5 9600X. While Precision Boost Overdrive (PBO) can add up to 200 MHz, it won't reach the levels of the top-tier 9600X.\u003C/li>\u003Cli class=\"content-block-list-item\">\u003Cb>Power Profile:\u003C/b> Built on TSMC's 4nm process, the Zen 5 cores have a different power-to-frequency curve. Within the same 65W TDP (up to 88W peak), the 9500F runs at slightly lower all-core frequencies (4.8 GHz) compared to the 7500F (4.9 GHz) under heavy load.\u003C/li>\u003C/ul>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">Despite the slightly lower all-core clock, the architectural improvements of Zen 5 are expected to more than compensate for this difference. Thermals are also well-managed. Thanks to a conservative power limit and a high-quality soldered thermal interface, the 9500F runs cool, peaking at just 67°C in our tests with a liquid cooler, compared to 72°C for the 7500F. Even the included Wraith Stealth stock cooler should be sufficient for stock operation.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">For enthusiasts, the 9500F remains fully unlocked. By enabling PBO, lifting power limits, and using Curve Optimizer, it's possible to push the CPU to a stable 5.25 GHz on all cores. However, this transforms it into a much hotter chip, drawing up to 125W and requiring a high-performance cooling solution.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch2 class=\"block-content\">Test Methodology\u003C/h2>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">We tested the Ryzen 5 9500F against a full lineup of its direct competitors in the sub-$200 price category, as well as its slightly more expensive siblings, to provide a complete performance picture.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch3 class=\"block-content\">Processors Tested\u003C/h3>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cul class=\"block-content content-block-list-unordered\">\u003Cli class=\"content-block-list-item\">AMD Ryzen 5 9600X (6-core, Zen 5)\u003C/li>\u003Cli class=\"content-block-list-item\">\u003Cb>AMD Ryzen 5 9500F (6-core, Zen 5)\u003C/b>\u003C/li>\u003Cli class=\"content-block-list-item\">AMD Ryzen 5 7600X (6-core, Zen 4)\u003C/li>\u003Cli class=\"content-block-list-item\">AMD Ryzen 5 7500F (6-core, Zen 4)\u003C/li>\u003Cli class=\"content-block-list-item\">Intel Core Ultra 5 225 (6P+4E, Arrow Lake)\u003C/li>\u003Cli class=\"content-block-list-item\">Intel Core i5-14600K (6P+8E, Raptor Lake Refresh)\u003C/li>\u003Cli class=\"content-block-list-item\">Intel Core i5-14400 (6P+4E, Alder Lake)\u003C/li>\u003C/ul>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch3 class=\"block-content\">Test Bench Hardware\u003C/h3>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cul class=\"block-content content-block-list-unordered\">\u003Cli class=\"content-block-list-item\">\u003Cb>CPU Cooler:\u003C/b> Custom Liquid Cooling Loop (EKWB components)\u003C/li>\u003Cli class=\"content-block-list-item\">\u003Cb>Motherboards:\u003C/b> MSI MEG Z890 Unity-X (LGA1851), ASUS ROG Maximus Z790 Apex (LGA1700), MSI MPG X670E Carbon WiFi (AM5)\u003C/li>\u003Cli class=\"content-block-list-item\">\u003Cb>Memory:\u003C/b> 32GB (2x16GB) G.Skill Ripjaws S5 DDR5-6400 (run at DDR5-6000 for AM5)\u003C/li>\u003Cli class=\"content-block-list-item\">\u003Cb>Graphics Card:\u003C/b> Palit GeForce RTX 5090 GameRock\u003C/li>\u003Cli class=\"content-block-list-item\">\u003Cb>Storage:\u003C/b> 2TB Intel SSD 760p\u003C/li>\u003Cli class=\"content-block-list-item\">\u003Cb>Power Supply:\u003C/b> 1200W Deepcool PX1200G (80+ Gold)\u003C/li>\u003C/ul>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch2 class=\"block-content\">Application Performance\u003C/h2>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">In productivity and content creation tasks, the Ryzen 5 9500F fits perfectly into AMD's lineup. On average, it delivers a \u003Cb>12% performance increase\u003C/b> over its predecessor, the 7500F. This is enough to also surpass the last-gen Ryzen 5 7600X by a comfortable 5-6% margin.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">When pitted against Intel, the 9500F shows its strength. It's about 10% faster than the Core i5-14400 and also edges out the newer Core Ultra 5 225 in most workloads. The only competitor that remains out of reach is the Core i5-14600K, whose 14-core design provides a 20-25% advantage in heavily multi-threaded applications. However, the i5-14600K also sits in a higher price bracket.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch2 class=\"block-content\">Gaming Performance (1080p)\u003C/h2>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">In gaming, the results are consistent. The Ryzen 5 9500F is a clear upgrade over the entire Zen 4 lineup of six-core CPUs while falling just short of the more expensive Ryzen 5 9600X.\u003C/p>\u003C/div>\u003Cfigure class=\"block-wrapper block-wrapper--media\">\u003Cdiv class=\"block-wrapper__content\">\u003Cblockquote class=\"block-quote block-quote--text-small\">\u003Cdiv class=\"block-quote__line\">\u003C/div>\u003Cdiv class=\"block-quote__text\">On average, the Ryzen 5 9500F is 9% faster than the Ryzen 5 7500F in gaming—exactly as AMD promised. In some titles like \u003Ci>Hogwarts Legacy\u003C/i> and \u003Ci>Homeworld 3\u003C/i>, the lead can stretch to over 15%.\u003C/div>\u003C/blockquote>\u003C/div>\u003C/figure>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">The 9500F establishes a confident lead over both the Core i5-14400 and the Core Ultra 5 225. Only the pricier Core i5-14600K and Ryzen 5 9600X consistently deliver higher frame rates. This places the 9500F in a sweet spot for budget-to-mid-range gaming builds.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch2 class=\"block-content\">Gaming Performance (4K/2160p)\u003C/h2>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">At 4K, the GPU becomes the primary bottleneck, causing the performance gaps between CPUs to shrink. However, the hierarchy remains the same. The Ryzen 5 9500F maintains its advantage over the Zen 4 six-cores, the Core i5-14400, and the Core Ultra 5 225. While the absolute differences are smaller, the 9500F proves it's a capable CPU that won't hold back a high-end graphics card, even at higher resolutions.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch2 class=\"block-content\">Power Consumption and Thermals\u003C/h2>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">The Ryzen 5 9500F is an efficient and cool-running processor. Its combination of lower clocks, an 88W power limit, and a soldered heatspreader keeps temperatures in check. Under a full multi-core load, its power consumption is capped, making it more efficient than the Ryzen 5 7600X and Core i5-14600K.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">In gaming, the 9500F typically consumes around 80W, which is slightly more than the 7500F and its direct Intel competitors. The standout for efficiency is the Core Ultra 5 225, which proves to be significantly more frugal, especially in single-threaded tasks and gaming. Nonetheless, the 9500F doesn't require an expensive cooling solution and is very easy to manage thermally.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch2 class=\"block-content\">Conclusion\u003C/h2>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">The Ryzen 5 9500F is a logical and well-balanced addition to AMD's Zen 5 lineup. It’s not a revolutionary product but a smart refinement, bringing the latest architecture to a price point where it can truly shine. It successfully continues the legacy of previous Ryzen 5 processors as a potential \"people's champion.\"\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">By essentially offering a slightly down-clocked Ryzen 5 9600X without integrated graphics, AMD has created a CPU that delivers nearly identical real-world performance for less money. It provides a noticeable 9-12% uplift over the 7500F and even surpasses the 7600X, all while being more efficient and easier to cool.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cul class=\"block-content content-block-list-unordered\">\u003Cli class=\"content-block-list-item\">\u003Cb>Excellent Performance:\u003C/b> Delivers a meaningful performance boost over the previous generation and beats its direct Intel rivals in most tasks.\u003C/li>\u003Cli class=\"content-block-list-item\">\u003Cb>Efficient and Cool:\u003C/b> Easy to manage thermally, making it suitable for budget-friendly builds with standard air coolers.\u003C/li>\u003Cli class=\"content-block-list-item\">\u003Cb>Great Value:\u003C/b> With an expected price of around $170-$180, it hits a sweet spot, offering better performance than similarly priced competitors.\u003C/li>\u003Cli class=\"content-block-list-item\">\u003Cb>AM5 Platform:\u003C/b> Provides a solid entry point into a modern and future-proof ecosystem with an excellent upgrade path, including future 3D V-Cache CPUs.\u003C/li>\u003C/ul>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">If the retail price holds steady, the Ryzen 5 9500F is poised to be an extremely popular choice. While the Ryzen 5 7500F may remain the absolute cheapest entry into the AM5 platform, the 9500F offers a compelling reason to spend a little more for a significant jump in performance. It is, without a doubt, a top contender for the best budget CPU on the market today.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">\u003Ca href=\"/tag/amd\" target=\"_blank\" rel=\"noopener noreferrer ugc\">#amd\u003C/a>\u003Ca href=\"/tag/#ryzen\" target=\"_blank\" rel=\"noopener noreferrer ugc\"> #ryzen\u003C/a>\u003C/p>\u003C/div>\u003C/div>",{"time":635,"blocks":636,"version":123},1760939472063,[637,641,645,649,653,657,661,665,682,686,690,694,698,702,735,739,768,772,776,780,784,788,792,796,800,804,808,812,816,820,824,828,849,853],{"id":638,"type":14,"data":639},"intro_p_1",{"text":640},"The AM5 platform has officially gone mainstream. AMD has successfully populated the market not only with high-end CPUs for enthusiasts but also with affordable motherboards and processors starting around $120. However, budget-conscious builders have been limited to previous-generation Zen 4 chips like the Ryzen 5 7500F. While effective, these 2022-era CPUs are starting to show their age.",{"id":642,"type":14,"data":643},"intro_p_2",{"text":644},"Now, AMD is shaking things up by bringing its latest architecture to the budget segment. The new \u003Cb>Ryzen 5 9500F\u003C/b> is the successor to the popular 7500F, built on the cutting-edge Zen 5 architecture. This upgrade promises a significant performance boost where it matters most for gamers.",{"id":646,"type":60,"data":647},"quote_1",{"quoteText":648,"authorText":63},"AMD claims the Zen 5 architecture delivers a 16% improvement in instructions per clock (IPC) over Zen 4, translating to a 9-11% frame rate increase in esports and CPU-intensive AAA titles.",{"id":650,"type":14,"data":651},"intro_p_3",{"text":652},"Interestingly, the Ryzen 5 9500F isn't about making AM5 systems cheaper. It's priced slightly higher than its predecessor. Instead, AMD's goal is to elevate the performance of budget AM5 builds, making them a more compelling alternative to Intel's popular Core i5 series, which remains a strong competitor even after the launch of the Arrow Lake family.",{"id":654,"type":14,"data":655},"intro_p_4",{"text":656},"In this review, we'll determine if AMD has crafted a worthy rival to the Core i5-14400F and i5-14600KF, and whether the new Ryzen 5 9500F has what it takes to become the next go-to processor for budget-savvy builders.",{"id":658,"type":27,"data":659},"features_header_1",{"text":660,"level":137},"Ryzen 5 9500F: Key Features",{"id":662,"type":14,"data":663},"features_p_1",{"text":664},"The Ryzen 5 9500F is a straightforward six-core, twelve-thread processor based on the Zen 5 architecture, featuring a 32 MB L3 cache. Its specifications are very similar to its predecessor, the Ryzen 5 7500F, with the primary difference being the upgraded core architecture. However, there are a few key distinctions:",{"id":666,"type":37,"data":667},"features_list_1",{"style":39,"meta":668,"items":669},{},[670,674,678],{"content":671,"meta":672,"items":673},"\u003Cb>No Integrated Graphics:\u003C/b> The \"F\" suffix indicates the lack of an integrated GPU, which is a non-issue for gamers who will be using a discrete graphics card.",{},[],{"content":675,"meta":676,"items":677},"\u003Cb>Clock Speeds:\u003C/b> The 9500F has a maximum boost clock of 5.0 GHz, which is 400 MHz lower than the Ryzen 5 9600X. While Precision Boost Overdrive (PBO) can add up to 200 MHz, it won't reach the levels of the top-tier 9600X.",{},[],{"content":679,"meta":680,"items":681},"\u003Cb>Power Profile:\u003C/b> Built on TSMC's 4nm process, the Zen 5 cores have a different power-to-frequency curve. Within the same 65W TDP (up to 88W peak), the 9500F runs at slightly lower all-core frequencies (4.8 GHz) compared to the 7500F (4.9 GHz) under heavy load.",{},[],{"id":683,"type":14,"data":684},"features_p_2",{"text":685},"Despite the slightly lower all-core clock, the architectural improvements of Zen 5 are expected to more than compensate for this difference. Thermals are also well-managed. Thanks to a conservative power limit and a high-quality soldered thermal interface, the 9500F runs cool, peaking at just 67°C in our tests with a liquid cooler, compared to 72°C for the 7500F. Even the included Wraith Stealth stock cooler should be sufficient for stock operation.",{"id":687,"type":14,"data":688},"features_p_3",{"text":689},"For enthusiasts, the 9500F remains fully unlocked. By enabling PBO, lifting power limits, and using Curve Optimizer, it's possible to push the CPU to a stable 5.25 GHz on all cores. However, this transforms it into a much hotter chip, drawing up to 125W and requiring a high-performance cooling solution.",{"id":691,"type":27,"data":692},"test_method_header",{"text":693,"level":137},"Test Methodology",{"id":695,"type":14,"data":696},"test_method_p_1",{"text":697},"We tested the Ryzen 5 9500F against a full lineup of its direct competitors in the sub-$200 price category, as well as its slightly more expensive siblings, to provide a complete performance picture.",{"id":699,"type":27,"data":700},"test_cpu_header",{"text":701,"level":30},"Processors Tested",{"id":703,"type":37,"data":704},"cpu_list_1",{"style":39,"meta":705,"items":706},{},[707,711,715,719,723,727,731],{"content":708,"meta":709,"items":710},"AMD Ryzen 5 9600X (6-core, Zen 5)",{},[],{"content":712,"meta":713,"items":714},"\u003Cb>AMD Ryzen 5 9500F (6-core, Zen 5)\u003C/b>",{},[],{"content":716,"meta":717,"items":718},"AMD Ryzen 5 7600X (6-core, Zen 4)",{},[],{"content":720,"meta":721,"items":722},"AMD Ryzen 5 7500F (6-core, Zen 4)",{},[],{"content":724,"meta":725,"items":726},"Intel Core Ultra 5 225 (6P+4E, Arrow Lake)",{},[],{"content":728,"meta":729,"items":730},"Intel Core i5-14600K (6P+8E, Raptor Lake Refresh)",{},[],{"content":732,"meta":733,"items":734},"Intel Core i5-14400 (6P+4E, Alder Lake)",{},[],{"id":736,"type":27,"data":737},"test_hardware_header",{"text":738,"level":30},"Test Bench Hardware",{"id":740,"type":37,"data":741},"hardware_list_1",{"style":39,"meta":742,"items":743},{},[744,748,752,756,760,764],{"content":745,"meta":746,"items":747},"\u003Cb>CPU Cooler:\u003C/b> Custom Liquid Cooling Loop (EKWB components)",{},[],{"content":749,"meta":750,"items":751},"\u003Cb>Motherboards:\u003C/b> MSI MEG Z890 Unity-X (LGA1851), ASUS ROG Maximus Z790 Apex (LGA1700), MSI MPG X670E Carbon WiFi (AM5)",{},[],{"content":753,"meta":754,"items":755},"\u003Cb>Memory:\u003C/b> 32GB (2x16GB) G.Skill Ripjaws S5 DDR5-6400 (run at DDR5-6000 for AM5)",{},[],{"content":757,"meta":758,"items":759},"\u003Cb>Graphics Card:\u003C/b> Palit GeForce RTX 5090 GameRock",{},[],{"content":761,"meta":762,"items":763},"\u003Cb>Storage:\u003C/b> 2TB Intel SSD 760p",{},[],{"content":765,"meta":766,"items":767},"\u003Cb>Power Supply:\u003C/b> 1200W Deepcool PX1200G (80+ Gold)",{},[],{"id":769,"type":27,"data":770},"performance_header_1",{"text":771,"level":137},"Application Performance",{"id":773,"type":14,"data":774},"performance_p_1",{"text":775},"In productivity and content creation tasks, the Ryzen 5 9500F fits perfectly into AMD's lineup. On average, it delivers a \u003Cb>12% performance increase\u003C/b> over its predecessor, the 7500F. This is enough to also surpass the last-gen Ryzen 5 7600X by a comfortable 5-6% margin.",{"id":777,"type":14,"data":778},"performance_p_2",{"text":779},"When pitted against Intel, the 9500F shows its strength. It's about 10% faster than the Core i5-14400 and also edges out the newer Core Ultra 5 225 in most workloads. The only competitor that remains out of reach is the Core i5-14600K, whose 14-core design provides a 20-25% advantage in heavily multi-threaded applications. However, the i5-14600K also sits in a higher price bracket.",{"id":781,"type":27,"data":782},"gaming_header_1080p",{"text":783,"level":137},"Gaming Performance (1080p)",{"id":785,"type":14,"data":786},"gaming_p_1080p_1",{"text":787},"In gaming, the results are consistent. The Ryzen 5 9500F is a clear upgrade over the entire Zen 4 lineup of six-core CPUs while falling just short of the more expensive Ryzen 5 9600X.",{"id":789,"type":60,"data":790},"gaming_quote_1080p",{"quoteText":791,"authorText":63},"On average, the Ryzen 5 9500F is 9% faster than the Ryzen 5 7500F in gaming—exactly as AMD promised. In some titles like \u003Ci>Hogwarts Legacy\u003C/i> and \u003Ci>Homeworld 3\u003C/i>, the lead can stretch to over 15%.",{"id":793,"type":14,"data":794},"gaming_p_1080p_2",{"text":795},"The 9500F establishes a confident lead over both the Core i5-14400 and the Core Ultra 5 225. Only the pricier Core i5-14600K and Ryzen 5 9600X consistently deliver higher frame rates. This places the 9500F in a sweet spot for budget-to-mid-range gaming builds.",{"id":797,"type":27,"data":798},"gaming_header_4k",{"text":799,"level":137},"Gaming Performance (4K/2160p)",{"id":801,"type":14,"data":802},"gaming_p_4k_1",{"text":803},"At 4K, the GPU becomes the primary bottleneck, causing the performance gaps between CPUs to shrink. However, the hierarchy remains the same. The Ryzen 5 9500F maintains its advantage over the Zen 4 six-cores, the Core i5-14400, and the Core Ultra 5 225. While the absolute differences are smaller, the 9500F proves it's a capable CPU that won't hold back a high-end graphics card, even at higher resolutions.",{"id":805,"type":27,"data":806},"power_header",{"text":807,"level":137},"Power Consumption and Thermals",{"id":809,"type":14,"data":810},"power_p_1",{"text":811},"The Ryzen 5 9500F is an efficient and cool-running processor. Its combination of lower clocks, an 88W power limit, and a soldered heatspreader keeps temperatures in check. Under a full multi-core load, its power consumption is capped, making it more efficient than the Ryzen 5 7600X and Core i5-14600K.",{"id":813,"type":14,"data":814},"power_p_2",{"text":815},"In gaming, the 9500F typically consumes around 80W, which is slightly more than the 7500F and its direct Intel competitors. The standout for efficiency is the Core Ultra 5 225, which proves to be significantly more frugal, especially in single-threaded tasks and gaming. Nonetheless, the 9500F doesn't require an expensive cooling solution and is very easy to manage thermally.",{"id":817,"type":27,"data":818},"conclusion_header",{"text":819,"level":137},"Conclusion",{"id":821,"type":14,"data":822},"conclusion_p_1",{"text":823},"The Ryzen 5 9500F is a logical and well-balanced addition to AMD's Zen 5 lineup. It’s not a revolutionary product but a smart refinement, bringing the latest architecture to a price point where it can truly shine. It successfully continues the legacy of previous Ryzen 5 processors as a potential \"people's champion.\"",{"id":825,"type":14,"data":826},"conclusion_p_2",{"text":827},"By essentially offering a slightly down-clocked Ryzen 5 9600X without integrated graphics, AMD has created a CPU that delivers nearly identical real-world performance for less money. It provides a noticeable 9-12% uplift over the 7500F and even surpasses the 7600X, all while being more efficient and easier to cool.",{"id":829,"type":37,"data":830},"conclusion_list_1",{"style":39,"meta":831,"items":832},{},[833,837,841,845],{"content":834,"meta":835,"items":836},"\u003Cb>Excellent Performance:\u003C/b> Delivers a meaningful performance boost over the previous generation and beats its direct Intel rivals in most tasks.",{},[],{"content":838,"meta":839,"items":840},"\u003Cb>Efficient and Cool:\u003C/b> Easy to manage thermally, making it suitable for budget-friendly builds with standard air coolers.",{},[],{"content":842,"meta":843,"items":844},"\u003Cb>Great Value:\u003C/b> With an expected price of around $170-$180, it hits a sweet spot, offering better performance than similarly priced competitors.",{},[],{"content":846,"meta":847,"items":848},"\u003Cb>AM5 Platform:\u003C/b> Provides a solid entry point into a modern and future-proof ecosystem with an excellent upgrade path, including future 3D V-Cache CPUs.",{},[],{"id":850,"type":14,"data":851},"conclusion_p_3",{"text":852},"If the retail price holds steady, the Ryzen 5 9500F is poised to be an extremely popular choice. While the Ryzen 5 7500F may remain the absolute cheapest entry into the AM5 platform, the 9500F offers a compelling reason to spend a little more for a significant jump in performance. It is, without a doubt, a top contender for the best budget CPU on the market today.",{"id":854,"type":14,"data":855},"KYvqAik4Cv",{"text":856},"#amd #ryzen",{"authorId":128,"name":129,"avatar":130,"username":131},[859,860],{"emojiId":141,"count":141,"reactedByUser":22},{"emojiId":30,"count":141,"reactedByUser":22},{"comments":136,"favorites":141,"views":407,"hits":862},48,"2025-10-20T05:48:09.512","2025-10-20T14:44:25.076",{"postId":866,"title":867,"slug":868,"preview":869,"html":870,"content":871,"author":1408,"authorUsername":1412,"topic":132,"status":138,"reactions":1413,"counters":1415,"createdAt":1418,"updatedAt":1419,"publishedPostId":145,"favorite":22},"29cdf764-f6de-49ae-ae68-88f85d897003","New AMD EPYC 4005 and Ryzen Pro 9000: Hardware Digest for September","573621-new-amd-epyc-4005-ryzen-pro-9000-hardware-digest-september","\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">September was a packed month for the hardware industry. Market leaders unveiled new processors and GPUs, while innovative startups launched their own accelerators. In this hardware digest, we’ll explore the key releases from September, dive into new solutions from emerging vendors, and share our insights on the latest trends in AI hardware.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--media\">\u003Cfigure class=\"block-content content-block-image\">\u003Cimg src=\"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/264ff1c5-f42a-46ac-8a96-7a26a5b538a3/1184.webp\" alt=\"Image\" class=\"rounded-xl medium-zoom-image\" loading=\"lazy\">\u003C/figure>\u003C/div>","\u003Cdiv class=\"content-wrapper break-words\">\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">September was a packed month for the hardware industry. Market leaders unveiled new processors and GPUs, while innovative startups launched their own accelerators. In this hardware digest, we’ll explore the key releases from September, dive into new solutions from emerging vendors, and share our insights on the latest trends in AI hardware.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--media\">\u003Cfigure class=\"block-content content-block-image\">\u003Cimg src=\"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/264ff1c5-f42a-46ac-8a96-7a26a5b538a3/1184.webp\" alt=\"Image\" class=\"rounded-xl medium-zoom-image\" loading=\"lazy\">\u003C/figure>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch2 class=\"block-content\">Server Platforms\u003C/h2>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch3 class=\"block-content\">MSI Servers with NVIDIA RTX Pro 6000 Blackwell Edition GPUs\u003C/h3>\u003C/div>\u003Cdiv class=\"block-wrapper--media\">\u003Cfigure class=\"block-content content-block-image\">\u003Cimg src=\"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/fdfb27d8-3454-4ebc-876b-4be1159da59e/1184.webp\" alt=\"Image\" class=\"rounded-xl medium-zoom-image\" loading=\"lazy\">\u003C/figure>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">MSI has \u003Ca href=\"https://www.msi.com/news/detail/MSI-Expands-NVIDIA-RTX-PRO-Server-Lineup---Accelerated-by-RTX-PRO-6000-Blackwell-Server-Edition-GPU-146930\">unveiled\u003C/a> two new servers based on the NVIDIA MGX modular architecture: the 4U CG480-S5063 and the 2U CG290-S3063. Both are designed for AI workloads and feature NVIDIA RTX Pro 6000 Blackwell Server Edition accelerators with 96 GB of memory.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch2 class=\"block-content\">Features of the CG480-S5063 (4U)\u003C/h2>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cul class=\"block-content content-block-list-unordered\">\u003Cli class=\"content-block-list-item\">Dual Intel Xeon 6500P / 6700P / 6700E series processors (up to 350W TDP).\u003C/li>\u003Cli class=\"content-block-list-item\">Up to 32 DDR5 slots (RDIMM / MRDIMM) per CPU.\u003C/li>\u003Cli class=\"content-block-list-item\">Up to 20 front-facing E1.S NVMe bays (PCIe 5.0 x4).\u003C/li>\u003Cli class=\"content-block-list-item\">Two internal M.2 slots (2280 / 22110) via PCIe 5.0 x2.\u003C/li>\u003Cli class=\"content-block-list-item\">Expansion options include eight double-width PCIe 5.0 x16 slots, supporting up to eight accelerators.\u003C/li>\u003Cli class=\"content-block-list-item\">Air cooling with hot-swappable fans.\u003C/li>\u003Cli class=\"content-block-list-item\">Four 3200W power supply units (80 PLUS Titanium certified).\u003C/li>\u003C/ul>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">The 2U CG290-S3063 model features a single Intel processor, half the memory slots, and support for up to four GPUs. Both systems are designed to offer an excellent price-to-performance ratio by balancing total VRAM with overall cost.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">With their rich I/O, NVMe storage, and modern memory, these servers are well-suited for both AI training and inference. They align with the current trend of integrating powerful accelerators into modular, high-density server platforms.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch3 class=\"block-content\">Giga Computing's AMD & Intel Blade Servers\u003C/h3>\u003C/div>\u003Cdiv class=\"block-wrapper--media\">\u003Cfigure class=\"block-content content-block-image\">\u003Cimg src=\"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/d6b2193c-cf62-43fe-8a48-108fec62df2c/1184.webp\" alt=\"Image\" class=\"rounded-xl medium-zoom-image\" loading=\"lazy\">\u003C/figure>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">Gigabyte's Giga Computing division has \u003Ca href=\"https://www.gigabyte.com/Press/News/2311\">announced\u003C/a> the B343 series of blade servers, targeting enterprise, cloud, and edge workloads. Available in two 3U versions—the AMD-based B343-C40 and the Intel-based B343-X40—each chassis houses ten nodes with flexible memory and networking configurations.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cul class=\"block-content content-block-list-unordered\">\u003Cli class=\"content-block-list-item\">\u003Cb>Form Factor:\u003C/b> 3U chassis with ten air-cooled nodes.\u003C/li>\u003Cli class=\"content-block-list-item\">\u003Cb>Node Design:\u003C/b> Each node is a self-contained server with its own CPU, memory, I/O, and network interfaces.\u003C/li>\u003Cli class=\"content-block-list-item\">\u003Cb>Power:\u003C/b> Four 2000W or two 3200W PSUs (80 PLUS Titanium).\u003C/li>\u003C/ul>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">The B343-C40 supports AMD EPYC Embedded 4005 (Grado) or Ryzen 9000 series processors, while the B343-X40 is designed for Intel Xeon 6300/E-2400 CPUs.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">Each node features up to four DDR5 ECC slots, one M.2 2280/22110 slot (PCIe 3.0 x1), two SFF NVMe/SATA slots, one FHHL expansion slot (PCIe 5.0 x16), and three OCP NIC 3.0 slots (PCIe 4.0 x4). Various network port configurations are supported, with a maximum of two 25 GbE ports.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">Packing ten servers into a 3U chassis delivers high density for cloud, edge, and scalable deployments. The choice between AMD and Intel provides the flexibility to optimize for specific workloads and ecosystems.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch2 class=\"block-content\">Processors\u003C/h2>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch3 class=\"block-content\">AMD EPYC Embedded 4005 for the Edge\u003C/h3>\u003C/div>\u003Cdiv class=\"block-wrapper--media\">\u003Cfigure class=\"block-content content-block-image\">\u003Cimg src=\"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/4d08f83d-3f8d-4da2-a3fa-92e4a2d98115/1184.webp\" alt=\"Image\" class=\"rounded-xl medium-zoom-image\" loading=\"lazy\">\u003C/figure>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">AMD has \u003Ca href=\"https://www.amd.com/en/products/embedded/epyc/4005-series.html\">introduced\u003C/a> the EPYC Embedded 4005 processors for industrial and edge systems like factory servers and next-generation firewalls. These new CPUs extend the EPYC 4005 Grado lineup and are based on the Ryzen 9000 architecture.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch2 class=\"block-content\">Key Features\u003C/h2>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cul class=\"block-content content-block-list-unordered\">\u003Cli class=\"content-block-list-item\">Zen 5 architecture, 4nm process technology, and chiplet design.\u003C/li>\u003Cli class=\"content-block-list-item\">Compatibility with the AM5 socket.\u003C/li>\u003Cli class=\"content-block-list-item\">Support for DDR5-5600 ECC, 28 PCIe 5.0 lanes, and AVX-512 instructions.\u003C/li>\u003Cli class=\"content-block-list-item\">Up to 128 MB of L3 cache in the top model with 3D V-Cache.\u003C/li>\u003Cli class=\"content-block-list-item\">Integrated AMD Infinity Guard security features.\u003C/li>\u003Cli class=\"content-block-list-item\">Flexible TDP ranging from 65W to 170W.\u003C/li>\u003Cli class=\"content-block-list-item\">A 7-year supply cycle, crucial for industrial solutions.\u003C/li>\u003C/ul>\u003C/div>\u003Cdiv class=\"block-wrapper--media\">\u003Cfigure class=\"block-content content-block-image\">\u003Cimg src=\"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/b82370bf-1734-4ce9-9bc4-7cbcb095eff2/1184.webp\" alt=\"Image\" class=\"rounded-xl medium-zoom-image\" loading=\"lazy\">\u003C/figure>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">The AMD EPYC Embedded 4005 family brings server-grade features like AVX-512, PCIe 5.0, and DDR5 ECC to a form factor suitable for edge, embedded, and NGFW applications. Its flexible TDP and long-term support make it a viable solution for system integrators and edge infrastructure operators.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch3 class=\"block-content\">AMD Ryzen Pro 9000 for Workstations\u003C/h3>\u003C/div>\u003Cdiv class=\"block-wrapper--media\">\u003Cfigure class=\"block-content content-block-image\">\u003Cimg src=\"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/b70f4b61-cc34-48fb-ba1f-036f63d17b4a/1184.webp\" alt=\"Image\" class=\"rounded-xl medium-zoom-image\" loading=\"lazy\">\u003C/figure>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">AMD has quietly \u003Ca href=\"https://www.amd.com/en/products/processors/ryzen-for-professionals.html#overview\">updated\u003C/a> its business processor lineup by adding the Ryzen Pro 9000 series, based on the Zen 5 architecture. These are professional versions of the consumer Ryzen 9000 CPUs, enhanced with security and enterprise management features for workstations and business PCs.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch2 class=\"block-content\">Key Features\u003C/h2>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cul class=\"block-content content-block-list-unordered\">\u003Cli class=\"content-block-list-item\">Based on the Zen 5 architecture.\u003C/li>\u003Cli class=\"content-block-list-item\">Utilizes the AM5 socket.\u003C/li>\u003Cli class=\"content-block-list-item\">Maintains a 65W TDP across all models.\u003C/li>\u003Cli class=\"content-block-list-item\">Includes AMD PRO technologies for enhanced security, memory protection (Memory Guard), and manageability.\u003C/li>\u003C/ul>\u003C/div>\u003Cdiv class=\"block-wrapper--media\">\u003Cfigure class=\"block-content content-block-image\">\u003Cimg src=\"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/e04bb2fd-b30f-4617-a011-83ec27461dee/1184.webp\" alt=\"Image\" class=\"rounded-xl medium-zoom-image\" loading=\"lazy\">\u003C/figure>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">This \"soft launch\" is likely aimed at OEM partners. The Ryzen Pro 9000 series offers an important upgrade for the workstation and corporate PC market, delivering the higher IPC and clock speeds of Zen 5 with a strong focus on security and manageability, all without increasing TDP.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch2 class=\"block-content\">GPUs and Accelerators\u003C/h2>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch3 class=\"block-content\">Innosilicon Fenghua 3: China's First Full-Featured RISC-V GPU\u003C/h3>\u003C/div>\u003Cdiv class=\"block-wrapper--media\">\u003Cfigure class=\"block-content content-block-image\">\u003Cimg src=\"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/018a67b4-4188-4132-87de-686f69c17774/1184.webp\" alt=\"Image\" class=\"rounded-xl medium-zoom-image\" loading=\"lazy\">\u003C/figure>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">Innosilicon has \u003Ca href=\"https://videocardz.com/newz/innosilicon-unveils-fenghua-3-gpu-with-directx12-support-and-hardware-ray-tracing\">unveiled\u003C/a> Fenghua 3, China's first full-featured, domestically designed GPU based on the open RISC-V architecture and compatible with CUDA. The new accelerator targets AI computing, model inference, scientific visualization, and cloud gaming.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch2 class=\"block-content\">Key Features\u003C/h2>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cul class=\"block-content content-block-list-unordered\">\u003Cli class=\"content-block-list-item\">Up to 112 GB of HBM memory, a record for a Chinese GPU.\u003C/li>\u003Cli class=\"content-block-list-item\">Based on the OpenCore RISC-V Nanhu V3 architecture, moving away from the PowerVR used in previous models.\u003C/li>\u003Cli class=\"content-block-list-item\">Supports DirectX 12, Vulkan 1.2, OpenGL 4.6, and hardware ray tracing.\u003C/li>\u003Cli class=\"content-block-list-item\">A single accelerator can handle models up to 72 billion parameters, while an eight-GPU cluster can run models like DeepSeek 671B.\u003C/li>\u003Cli class=\"content-block-list-item\">Claimed compatibility with CUDA, simplifying the adaptation of existing AI frameworks.\u003C/li>\u003C/ul>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">Fenghua 3 is also the first Chinese graphics card to support the YUV444 format for maximum color accuracy, a significant advantage for video editing and CAD. Furthermore, it's the world's first GPU with hardware support for DICOM, allowing for the correct display of X-rays, MRIs, and CT scans on standard monitors.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">The release of Fenghua 3 marks a strategic shift for China, moving from reliance on licensed GPU technology to fully proprietary architectures. Its use of RISC-V and claimed CUDA compatibility represent a major step toward a self-sufficient graphics ecosystem.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch3 class=\"block-content\">Metis M.2 Max: An Accelerator for AI Inference\u003C/h3>\u003C/div>\u003Cdiv class=\"block-wrapper--media\">\u003Cfigure class=\"block-content content-block-image\">\u003Cimg src=\"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/8f3140d4-2c9c-4248-9916-d606c2384779/1184.webp\" alt=\"Image\" class=\"rounded-xl medium-zoom-image\" loading=\"lazy\">\u003C/figure>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">Dutch startup Axelera AI has \u003Ca href=\"https://axelera.ai/news/axelera-ai-boosts-llms-at-the-edge-by-2x-with-metis-m.2-max-introduction\">announced\u003C/a> Metis M.2 Max, an enhanced version of its Metis M.2 module designed for AI inference on edge devices. The company claims it boosts performance for LLM and VLM tasks through increased memory bandwidth, improved cooling, and enhanced security.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch2 class=\"block-content\">Key Features\u003C/h2>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cul class=\"block-content content-block-list-unordered\">\u003Cli class=\"content-block-list-item\">Form Factor: M.2 2280 (M-Key) with a PCIe 3.0 x4 interface.\u003C/li>\u003Cli class=\"content-block-list-item\">Architecture: Based on the Metis AIPU with four RISC-V cores.\u003C/li>\u003Cli class=\"content-block-list-item\">Memory: Up to 16 GB with double the bandwidth of the original Metis M.2.\u003C/li>\u003Cli class=\"content-block-list-item\">Security: Integrated Root of Trust, secure boot, and protection against unauthorized software.\u003C/li>\u003Cli class=\"content-block-list-item\">Power Consumption: Averages 6.5W.\u003C/li>\u003C/ul>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">The device reportedly delivers a 33% performance increase in convolutional neural networks (CNNs) and doubles the token generation speed for LLM and VLM models. It's compatible with the Voyager SDK, which simplifies model optimization and application integration.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">The Metis M.2 Max maintains a compact and energy-efficient profile while offering significant improvements in memory bandwidth and security. It is particularly compelling for systems that require a balance of performance, size, and resilience in demanding operating conditions.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch3 class=\"block-content\">NVIDIA Blackwell Ultra: Key Details and Advantages\u003C/h3>\u003C/div>\u003Cdiv class=\"block-wrapper--media\">\u003Cfigure class=\"block-content content-block-image\">\u003Cimg src=\"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/9182306b-86e6-4a33-a2be-3889a9eb6045/1184.webp\" alt=\"Image\" class=\"rounded-xl medium-zoom-image\" loading=\"lazy\">\u003C/figure>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">NVIDIA has \u003Ca href=\"https://developer.nvidia.com/blog/inside-nvidia-blackwell-ultra-the-chip-powering-the-ai-factory-era/\">revealed\u003C/a> architectural details of Blackwell Ultra, an evolution of its Blackwell accelerator. The update includes a higher transistor count, a more powerful tensor core engine, and expanded memory, targeting AI inference, reasoning models, and high-density data center scaling.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch2 class=\"block-content\">Architecture and Specifications\u003C/h2>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cul class=\"block-content content-block-list-unordered\">\u003Cli class=\"content-block-list-item\">\u003Cb>Dual-Die Design:\u003C/b> Blackwell Ultra maintains a dual-die configuration connected by an NV-HBI interface with up to 10 TB/s of bandwidth.\u003C/li>\u003Cli class=\"content-block-list-item\">\u003Cb>Transistor Count:\u003C/b> A total of 208 billion transistors.\u003C/li>\u003Cli class=\"content-block-list-item\">\u003Cb>Tensor Cores:\u003C/b> The architecture includes 160 Streaming Multiprocessors (SMs) and approximately 640 fifth-generation Tensor Cores optimized for low-precision formats (NVFP4, FP8).\u003C/li>\u003Cli class=\"content-block-list-item\">\u003Cb>New FP Format:\u003C/b> NVFP4 is a new 4-bit floating-point format that NVIDIA claims offers accuracy close to FP8 while using up to 3.5 times less memory than FP16.\u003C/li>\u003C/ul>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">Compared to the original Blackwell, NVFP4 performance has increased from 10 to 15 PFLOPS. For attention layers, the throughput of special function units (SFUs) has been doubled, accelerating computations on long contexts and reducing time-to-first-token.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch2 class=\"block-content\">Memory Subsystem and Interconnect\u003C/h2>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cul class=\"block-content content-block-list-unordered\">\u003Cli class=\"content-block-list-item\">\u003Cb>Memory:\u003C/b> 288 GB of HBM3e with an 8192-bit bus, providing 8 TB/s of bandwidth.\u003C/li>\u003Cli class=\"content-block-list-item\">\u003Cb>NVLink 5:\u003C/b> Enables the connection of up to 576 GPUs into a single compute fabric.\u003C/li>\u003Cli class=\"content-block-list-item\">\u003Cb>Host Interface:\u003C/b> PCIe 6.0 x16.\u003C/li>\u003Cli class=\"content-block-list-item\">\u003Cb>Networking:\u003C/b> Integrates ConnectX-8 SuperNIC and Quantum-X800 InfiniBand / Spectrum-X Ethernet, with up to 800 Gb/s per GPU.\u003C/li>\u003C/ul>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">Blackwell Ultra's architectural changes signal a clear shift toward inference and reasoning tasks. The increased memory allows large models to reside entirely on the GPU, reducing latency. This positions the new accelerator as the foundation for modular and scalable AI factories.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch3 class=\"block-content\">D-Matrix JetStream Card for Distributed Inference\u003C/h3>\u003C/div>\u003Cdiv class=\"block-wrapper--media\">\u003Cfigure class=\"block-content content-block-image\">\u003Cimg src=\"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/596e4aad-c3c5-43bd-9033-2db3b57f0ec3/1184.webp\" alt=\"Image\" class=\"rounded-xl medium-zoom-image\" loading=\"lazy\">\u003C/figure>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">Startup d-Matrix has \u003Ca href=\"https://www.d-matrix.ai/announcements/jetstream/\">introduced\u003C/a> JetStream, a specialized I/O card designed to distribute AI inference workloads across servers. It works with the company's Corsair accelerators to reduce latency without requiring changes to existing network infrastructure.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch2 class=\"block-content\">Key Features\u003C/h2>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cul class=\"block-content content-block-list-unordered\">\u003Cli class=\"content-block-list-item\">\u003Cb>Form Factor:\u003C/b> PCIe 5.0 x16 expansion card.\u003C/li>\u003Cli class=\"content-block-list-item\">\u003Cb>Networking:\u003C/b> Two 200 GbE ports or one 400 GbE port with standard Ethernet (QSFP-DD).\u003C/li>\u003Cli class=\"content-block-list-item\">\u003Cb>Power Consumption:\u003C/b> Approximately 150W, cooled by a heatsink and heat pipes.\u003C/li>\u003C/ul>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">A single JetStream card serves up to four Corsair accelerators. The card eliminates the I/O bottleneck when scaling inference workloads, especially when a model is distributed across multiple servers, achieving a claimed network latency of just 2 μs. The company asserts that their solution is up to 10 times faster and three times more power-efficient than GPU-based alternatives for models over 100 billion parameters. Samples are available now, with mass production planned for the end of the year.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch2 class=\"block-content\">Memory\u003C/h2>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch3 class=\"block-content\">SK hynix Finalizes HBM4 Development\u003C/h3>\u003C/div>\u003Cdiv class=\"block-wrapper--media\">\u003Cfigure class=\"block-content content-block-image\">\u003Cimg src=\"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/bfd5484a-6e1f-4801-bf1a-527ff7b6c1ac/1184.webp\" alt=\"Image\" class=\"rounded-xl medium-zoom-image\" loading=\"lazy\">\u003C/figure>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">SK hynix has \u003Ca href=\"https://news.skhynix.com/sk-hynix-completes-worlds-first-hbm4-development-and-readies-mass-production/\">announced\u003C/a> the completion of its HBM4 memory development. The company has already sent 12-layer samples to clients and is preparing for mass production in the second half of 2025. HBM4 is set to become a key component in future AI accelerators like NVIDIA's Rubin, offering double the bandwidth and significantly improved power efficiency over HBM3/3E.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch2 class=\"block-content\">Key Features\u003C/h2>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cul class=\"block-content content-block-list-unordered\">\u003Cli class=\"content-block-list-item\">\u003Cb>2048-bit Interface:\u003C/b> Doubles the width compared to HBM3/E, a first since 2015.\u003C/li>\u003Cli class=\"content-block-list-item\">\u003Cb>Data Transfer Rate:\u003C/b> Over 10 Gbps per pin, 25% higher than the official JEDEC standard for HBM3E.\u003C/li>\u003Cli class=\"content-block-list-item\">\u003Cb>Bandwidth:\u003C/b> Twice that of HBM3E.\u003C/li>\u003Cli class=\"content-block-list-item\">\u003Cb>Power Efficiency:\u003C/b> Improved by 40%.\u003C/li>\u003Cli class=\"content-block-list-item\">\u003Cb>AI Performance:\u003C/b> Expected to boost AI accelerator performance by up to 69% by eliminating memory bottlenecks.\u003C/li>\u003C/ul>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">The memory is manufactured using a fifth-generation 10nm-class (1β nm) process and an innovative Advanced MR-MUF (Molded Underfill) assembly technique, which ensures mechanical stability and better heat dissipation. This new memory will be used in upcoming products from NVIDIA, AMD, Broadcom, and other AI industry leaders.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch2 class=\"block-content\">Storage\u003C/h2>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch3 class=\"block-content\">Solidigm D7-PS1010: A Liquid-Cooled E1.S SSD\u003C/h3>\u003C/div>\u003Cdiv class=\"block-wrapper--media\">\u003Cfigure class=\"block-content content-block-image\">\u003Cimg src=\"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/dfbfcb23-04b7-4dd8-8f67-38fc28f9aee8/1184.webp\" alt=\"Image\" class=\"rounded-xl medium-zoom-image\" loading=\"lazy\">\u003C/figure>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">Solidigm (an SK hynix brand) has \u003Ca href=\"https://www.ixbt.com/news/2025/03/22/ssd-solidigm-d7-ps1010-e1-s-essd.html\">introduced\u003C/a> one of the world's first liquid-cooled enterprise SSDs, the D7-PS1010 E1.S. It is designed for high-density servers that lack traditional air cooling. The key feature is a cold plate that contacts both sides of the hot-swappable drive.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch2 class=\"block-content\">Key Features\u003C/h2>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cul class=\"block-content content-block-list-unordered\">\u003Cli class=\"content-block-list-item\">\u003Cb>Form Factor:\u003C/b> E1.S with 9.5mm and 15mm thickness options.\u003C/li>\u003Cli class=\"content-block-list-item\">\u003Cb>Technology:\u003C/b> Based on 176-layer 3D TLC NAND and a PCIe 5.0 x4 (NVMe) interface.\u003C/li>\u003Cli class=\"content-block-list-item\">\u003Cb>Capacities:\u003C/b> Available in 3.84 TB and 7.68 TB for the liquid-cooled version.\u003C/li>\u003Cli class=\"content-block-list-item\">\u003Cb>Speeds:\u003C/b> Sequential read speeds up to 14,500 MB/s and write speeds up to 10,500 MB/s.\u003C/li>\u003C/ul>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">Solidigm is working with server manufacturers to integrate this solution into AI infrastructures, including systems based on the NVIDIA HGX B300. Unlike traditional liquid cooling systems that cool only one side of an SSD, this new design cools both sides, improving thermal balance and reducing the risk of throttling. By eliminating the need for dedicated drive fans, it enables even denser server configurations and could accelerate the transition of data centers to fully liquid-cooled environments.\u003C/p>\u003C/div>\u003Cfigure class=\"block-wrapper block-wrapper--media\">\u003Cdiv class=\"block-wrapper__content\">\u003Cblockquote class=\"block-quote block-quote--text-small\">\u003Cdiv class=\"block-quote__line\">\u003C/div>\u003Cdiv class=\"block-quote__text\">Do you think data centers will eventually transition to fully liquid cooling? And is it necessary? Stay tuned for more hardware news.\u003C/div>\u003C/blockquote>\u003C/div>\u003C/figure>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">\u003Ca href=\"/tag/news\" target=\"_blank\" rel=\"noopener noreferrer ugc\">#news\u003C/a>\u003Ca href=\"/tag/#hardware\" target=\"_blank\" rel=\"noopener noreferrer ugc\"> #hardware\u003C/a>\u003C/p>\u003C/div>\u003C/div>",{"time":872,"blocks":873,"version":123},1760689206728,[874,878,883,887,891,896,900,904,937,941,945,949,954,958,975,979,983,987,991,995,1000,1004,1008,1041,1046,1050,1054,1059,1063,1066,1087,1092,1096,1100,1104,1109,1113,1116,1141,1145,1149,1153,1158,1162,1165,1190,1194,1198,1202,1207,1211,1215,1236,1240,1244,1265,1269,1273,1278,1282,1285,1302,1306,1310,1314,1319,1323,1326,1351,1355,1359,1363,1368,1372,1375,1396,1400,1404],{"id":875,"type":14,"data":876},"a9b8c7d6",{"text":877},"September was a packed month for the hardware industry. Market leaders unveiled new processors and GPUs, while innovative startups launched their own accelerators. In this hardware digest, we’ll explore the key releases from September, dive into new solutions from emerging vendors, and share our insights on the latest trends in AI hardware.",{"id":879,"type":19,"data":880},"hgwAQERTJk",{"caption":63,"withBorder":22,"withBackground":22,"stretched":22,"file":881},{"url":882},"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/264ff1c5-f42a-46ac-8a96-7a26a5b538a3/1184.webp",{"id":884,"type":27,"data":885},"f1d2e3c4",{"text":886,"level":137},"Server Platforms",{"id":888,"type":27,"data":889},"q7r8s9t0",{"text":890,"level":30},"MSI Servers with NVIDIA RTX Pro 6000 Blackwell Edition GPUs",{"id":892,"type":19,"data":893},"u1v2w3x4",{"caption":63,"withBorder":22,"withBackground":22,"stretched":22,"file":894},{"url":895},"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/fdfb27d8-3454-4ebc-876b-4be1159da59e/1184.webp",{"id":897,"type":14,"data":898},"y5z6a7b8",{"text":899},"MSI has \u003Ca href=\"https://www.msi.com/news/detail/MSI-Expands-NVIDIA-RTX-PRO-Server-Lineup---Accelerated-by-RTX-PRO-6000-Blackwell-Server-Edition-GPU-146930\">unveiled\u003C/a> two new servers based on the NVIDIA MGX modular architecture: the 4U CG480-S5063 and the 2U CG290-S3063. Both are designed for AI workloads and feature NVIDIA RTX Pro 6000 Blackwell Server Edition accelerators with 96 GB of memory.",{"id":901,"type":27,"data":902},"c9d0e1f2",{"text":903,"level":137},"Features of the CG480-S5063 (4U)",{"id":905,"type":37,"data":906},"g3h4i5j6",{"style":39,"meta":907,"items":908},{},[909,913,917,921,925,929,933],{"content":910,"meta":911,"items":912},"Dual Intel Xeon 6500P / 6700P / 6700E series processors (up to 350W TDP).",{},[],{"content":914,"meta":915,"items":916},"Up to 32 DDR5 slots (RDIMM / MRDIMM) per CPU.",{},[],{"content":918,"meta":919,"items":920},"Up to 20 front-facing E1.S NVMe bays (PCIe 5.0 x4).",{},[],{"content":922,"meta":923,"items":924},"Two internal M.2 slots (2280 / 22110) via PCIe 5.0 x2.",{},[],{"content":926,"meta":927,"items":928},"Expansion options include eight double-width PCIe 5.0 x16 slots, supporting up to eight accelerators.",{},[],{"content":930,"meta":931,"items":932},"Air cooling with hot-swappable fans.",{},[],{"content":934,"meta":935,"items":936},"Four 3200W power supply units (80 PLUS Titanium certified).",{},[],{"id":938,"type":14,"data":939},"k7l8m9n0",{"text":940},"The 2U CG290-S3063 model features a single Intel processor, half the memory slots, and support for up to four GPUs. Both systems are designed to offer an excellent price-to-performance ratio by balancing total VRAM with overall cost.",{"id":942,"type":14,"data":943},"o1p2q3r4",{"text":944},"With their rich I/O, NVMe storage, and modern memory, these servers are well-suited for both AI training and inference. They align with the current trend of integrating powerful accelerators into modular, high-density server platforms.",{"id":946,"type":27,"data":947},"s5t6u7v8",{"text":948,"level":30},"Giga Computing's AMD & Intel Blade Servers",{"id":950,"type":19,"data":951},"w9x0y1z2",{"caption":63,"withBorder":22,"withBackground":22,"stretched":22,"file":952},{"url":953},"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/d6b2193c-cf62-43fe-8a48-108fec62df2c/1184.webp",{"id":955,"type":14,"data":956},"a3b4c5d6",{"text":957},"Gigabyte's Giga Computing division has \u003Ca href=\"https://www.gigabyte.com/Press/News/2311\">announced\u003C/a> the B343 series of blade servers, targeting enterprise, cloud, and edge workloads. Available in two 3U versions—the AMD-based B343-C40 and the Intel-based B343-X40—each chassis houses ten nodes with flexible memory and networking configurations.",{"id":959,"type":37,"data":960},"e7f8g9h0",{"style":39,"meta":961,"items":962},{},[963,967,971],{"content":964,"meta":965,"items":966},"\u003Cb>Form Factor:\u003C/b> 3U chassis with ten air-cooled nodes.",{},[],{"content":968,"meta":969,"items":970},"\u003Cb>Node Design:\u003C/b> Each node is a self-contained server with its own CPU, memory, I/O, and network interfaces.",{},[],{"content":972,"meta":973,"items":974},"\u003Cb>Power:\u003C/b> Four 2000W or two 3200W PSUs (80 PLUS Titanium).",{},[],{"id":976,"type":14,"data":977},"i1j2k3l4",{"text":978},"The B343-C40 supports AMD EPYC Embedded 4005 (Grado) or Ryzen 9000 series processors, while the B343-X40 is designed for Intel Xeon 6300/E-2400 CPUs.",{"id":980,"type":14,"data":981},"m5n6o7p8",{"text":982},"Each node features up to four DDR5 ECC slots, one M.2 2280/22110 slot (PCIe 3.0 x1), two SFF NVMe/SATA slots, one FHHL expansion slot (PCIe 5.0 x16), and three OCP NIC 3.0 slots (PCIe 4.0 x4). Various network port configurations are supported, with a maximum of two 25 GbE ports.",{"id":984,"type":14,"data":985},"q9r0s1t2",{"text":986},"Packing ten servers into a 3U chassis delivers high density for cloud, edge, and scalable deployments. The choice between AMD and Intel provides the flexibility to optimize for specific workloads and ecosystems.",{"id":988,"type":27,"data":989},"u3v4w5x6",{"text":990,"level":137},"Processors",{"id":992,"type":27,"data":993},"y7z8a9b0",{"text":994,"level":30},"AMD EPYC Embedded 4005 for the Edge",{"id":996,"type":19,"data":997},"c1d2e3f4",{"caption":63,"withBorder":22,"withBackground":22,"stretched":22,"file":998},{"url":999},"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/4d08f83d-3f8d-4da2-a3fa-92e4a2d98115/1184.webp",{"id":1001,"type":14,"data":1002},"g5h6i7j8",{"text":1003},"AMD has \u003Ca href=\"https://www.amd.com/en/products/embedded/epyc/4005-series.html\">introduced\u003C/a> the EPYC Embedded 4005 processors for industrial and edge systems like factory servers and next-generation firewalls. These new CPUs extend the EPYC 4005 Grado lineup and are based on the Ryzen 9000 architecture.",{"id":1005,"type":27,"data":1006},"k9l0m1n2",{"text":1007,"level":137},"Key Features",{"id":1009,"type":37,"data":1010},"o3p4q5r6",{"style":39,"meta":1011,"items":1012},{},[1013,1017,1021,1025,1029,1033,1037],{"content":1014,"meta":1015,"items":1016},"Zen 5 architecture, 4nm process technology, and chiplet design.",{},[],{"content":1018,"meta":1019,"items":1020},"Compatibility with the AM5 socket.",{},[],{"content":1022,"meta":1023,"items":1024},"Support for DDR5-5600 ECC, 28 PCIe 5.0 lanes, and AVX-512 instructions.",{},[],{"content":1026,"meta":1027,"items":1028},"Up to 128 MB of L3 cache in the top model with 3D V-Cache.",{},[],{"content":1030,"meta":1031,"items":1032},"Integrated AMD Infinity Guard security features.",{},[],{"content":1034,"meta":1035,"items":1036},"Flexible TDP ranging from 65W to 170W.",{},[],{"content":1038,"meta":1039,"items":1040},"A 7-year supply cycle, crucial for industrial solutions.",{},[],{"id":1042,"type":19,"data":1043},"s7t8u9v0",{"caption":63,"withBorder":22,"withBackground":22,"stretched":22,"file":1044},{"url":1045},"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/b82370bf-1734-4ce9-9bc4-7cbcb095eff2/1184.webp",{"id":1047,"type":14,"data":1048},"w1x2y3z4",{"text":1049},"The AMD EPYC Embedded 4005 family brings server-grade features like AVX-512, PCIe 5.0, and DDR5 ECC to a form factor suitable for edge, embedded, and NGFW applications. Its flexible TDP and long-term support make it a viable solution for system integrators and edge infrastructure operators.",{"id":1051,"type":27,"data":1052},"a5b6c7d8",{"text":1053,"level":30},"AMD Ryzen Pro 9000 for Workstations",{"id":1055,"type":19,"data":1056},"e9f0g1h2",{"caption":63,"withBorder":22,"withBackground":22,"stretched":22,"file":1057},{"url":1058},"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/b70f4b61-cc34-48fb-ba1f-036f63d17b4a/1184.webp",{"id":1060,"type":14,"data":1061},"i3j4k5l6",{"text":1062},"AMD has quietly \u003Ca href=\"https://www.amd.com/en/products/processors/ryzen-for-professionals.html#overview\">updated\u003C/a> its business processor lineup by adding the Ryzen Pro 9000 series, based on the Zen 5 architecture. These are professional versions of the consumer Ryzen 9000 CPUs, enhanced with security and enterprise management features for workstations and business PCs.",{"id":1064,"type":27,"data":1065},"m7n8o9p0",{"text":1007,"level":137},{"id":1067,"type":37,"data":1068},"q1r2s3t4",{"style":39,"meta":1069,"items":1070},{},[1071,1075,1079,1083],{"content":1072,"meta":1073,"items":1074},"Based on the Zen 5 architecture.",{},[],{"content":1076,"meta":1077,"items":1078},"Utilizes the AM5 socket.",{},[],{"content":1080,"meta":1081,"items":1082},"Maintains a 65W TDP across all models.",{},[],{"content":1084,"meta":1085,"items":1086},"Includes AMD PRO technologies for enhanced security, memory protection (Memory Guard), and manageability.",{},[],{"id":1088,"type":19,"data":1089},"u5v6w7x8",{"caption":63,"withBorder":22,"withBackground":22,"stretched":22,"file":1090},{"url":1091},"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/e04bb2fd-b30f-4617-a011-83ec27461dee/1184.webp",{"id":1093,"type":14,"data":1094},"y9z0a1b2",{"text":1095},"This \"soft launch\" is likely aimed at OEM partners. The Ryzen Pro 9000 series offers an important upgrade for the workstation and corporate PC market, delivering the higher IPC and clock speeds of Zen 5 with a strong focus on security and manageability, all without increasing TDP.",{"id":1097,"type":27,"data":1098},"c3d4e5f6",{"text":1099,"level":137},"GPUs and Accelerators",{"id":1101,"type":27,"data":1102},"g7h8i9j0",{"text":1103,"level":30},"Innosilicon Fenghua 3: China's First Full-Featured RISC-V GPU",{"id":1105,"type":19,"data":1106},"k1l2m3n4",{"caption":63,"withBorder":22,"withBackground":22,"stretched":22,"file":1107},{"url":1108},"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/018a67b4-4188-4132-87de-686f69c17774/1184.webp",{"id":1110,"type":14,"data":1111},"o5p6q7r8",{"text":1112},"Innosilicon has \u003Ca href=\"https://videocardz.com/newz/innosilicon-unveils-fenghua-3-gpu-with-directx12-support-and-hardware-ray-tracing\">unveiled\u003C/a> Fenghua 3, China's first full-featured, domestically designed GPU based on the open RISC-V architecture and compatible with CUDA. The new accelerator targets AI computing, model inference, scientific visualization, and cloud gaming.",{"id":1114,"type":27,"data":1115},"s9t0u1v2",{"text":1007,"level":137},{"id":1117,"type":37,"data":1118},"w3x4y5z6",{"style":39,"meta":1119,"items":1120},{},[1121,1125,1129,1133,1137],{"content":1122,"meta":1123,"items":1124},"Up to 112 GB of HBM memory, a record for a Chinese GPU.",{},[],{"content":1126,"meta":1127,"items":1128},"Based on the OpenCore RISC-V Nanhu V3 architecture, moving away from the PowerVR used in previous models.",{},[],{"content":1130,"meta":1131,"items":1132},"Supports DirectX 12, Vulkan 1.2, OpenGL 4.6, and hardware ray tracing.",{},[],{"content":1134,"meta":1135,"items":1136},"A single accelerator can handle models up to 72 billion parameters, while an eight-GPU cluster can run models like DeepSeek 671B.",{},[],{"content":1138,"meta":1139,"items":1140},"Claimed compatibility with CUDA, simplifying the adaptation of existing AI frameworks.",{},[],{"id":1142,"type":14,"data":1143},"a7b8c9d0",{"text":1144},"Fenghua 3 is also the first Chinese graphics card to support the YUV444 format for maximum color accuracy, a significant advantage for video editing and CAD. Furthermore, it's the world's first GPU with hardware support for DICOM, allowing for the correct display of X-rays, MRIs, and CT scans on standard monitors.",{"id":1146,"type":14,"data":1147},"e1f2g3h4",{"text":1148},"The release of Fenghua 3 marks a strategic shift for China, moving from reliance on licensed GPU technology to fully proprietary architectures. Its use of RISC-V and claimed CUDA compatibility represent a major step toward a self-sufficient graphics ecosystem.",{"id":1150,"type":27,"data":1151},"i5j6k7l8",{"text":1152,"level":30},"Metis M.2 Max: An Accelerator for AI Inference",{"id":1154,"type":19,"data":1155},"m9n0o1p2",{"caption":63,"withBorder":22,"withBackground":22,"stretched":22,"file":1156},{"url":1157},"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/8f3140d4-2c9c-4248-9916-d606c2384779/1184.webp",{"id":1159,"type":14,"data":1160},"q3r4s5t6",{"text":1161},"Dutch startup Axelera AI has \u003Ca href=\"https://axelera.ai/news/axelera-ai-boosts-llms-at-the-edge-by-2x-with-metis-m.2-max-introduction\">announced\u003C/a> Metis M.2 Max, an enhanced version of its Metis M.2 module designed for AI inference on edge devices. The company claims it boosts performance for LLM and VLM tasks through increased memory bandwidth, improved cooling, and enhanced security.",{"id":1163,"type":27,"data":1164},"u7v8w9x0",{"text":1007,"level":137},{"id":1166,"type":37,"data":1167},"y1z2a3b4",{"style":39,"meta":1168,"items":1169},{},[1170,1174,1178,1182,1186],{"content":1171,"meta":1172,"items":1173},"Form Factor: M.2 2280 (M-Key) with a PCIe 3.0 x4 interface.",{},[],{"content":1175,"meta":1176,"items":1177},"Architecture: Based on the Metis AIPU with four RISC-V cores.",{},[],{"content":1179,"meta":1180,"items":1181},"Memory: Up to 16 GB with double the bandwidth of the original Metis M.2.",{},[],{"content":1183,"meta":1184,"items":1185},"Security: Integrated Root of Trust, secure boot, and protection against unauthorized software.",{},[],{"content":1187,"meta":1188,"items":1189},"Power Consumption: Averages 6.5W.",{},[],{"id":1191,"type":14,"data":1192},"c5d6e7f8",{"text":1193},"The device reportedly delivers a 33% performance increase in convolutional neural networks (CNNs) and doubles the token generation speed for LLM and VLM models. It's compatible with the Voyager SDK, which simplifies model optimization and application integration.",{"id":1195,"type":14,"data":1196},"g9h0i1j2",{"text":1197},"The Metis M.2 Max maintains a compact and energy-efficient profile while offering significant improvements in memory bandwidth and security. It is particularly compelling for systems that require a balance of performance, size, and resilience in demanding operating conditions.",{"id":1199,"type":27,"data":1200},"k3l4m5n6",{"text":1201,"level":30},"NVIDIA Blackwell Ultra: Key Details and Advantages",{"id":1203,"type":19,"data":1204},"o7p8q9r0",{"caption":63,"withBorder":22,"withBackground":22,"stretched":22,"file":1205},{"url":1206},"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/9182306b-86e6-4a33-a2be-3889a9eb6045/1184.webp",{"id":1208,"type":14,"data":1209},"s1t2u3v4",{"text":1210},"NVIDIA has \u003Ca href=\"https://developer.nvidia.com/blog/inside-nvidia-blackwell-ultra-the-chip-powering-the-ai-factory-era/\">revealed\u003C/a> architectural details of Blackwell Ultra, an evolution of its Blackwell accelerator. The update includes a higher transistor count, a more powerful tensor core engine, and expanded memory, targeting AI inference, reasoning models, and high-density data center scaling.",{"id":1212,"type":27,"data":1213},"w5x6y7z8",{"text":1214,"level":137},"Architecture and Specifications",{"id":1216,"type":37,"data":1217},"a9b0c1d2",{"style":39,"meta":1218,"items":1219},{},[1220,1224,1228,1232],{"content":1221,"meta":1222,"items":1223},"\u003Cb>Dual-Die Design:\u003C/b> Blackwell Ultra maintains a dual-die configuration connected by an NV-HBI interface with up to 10 TB/s of bandwidth.",{},[],{"content":1225,"meta":1226,"items":1227},"\u003Cb>Transistor Count:\u003C/b> A total of 208 billion transistors.",{},[],{"content":1229,"meta":1230,"items":1231},"\u003Cb>Tensor Cores:\u003C/b> The architecture includes 160 Streaming Multiprocessors (SMs) and approximately 640 fifth-generation Tensor Cores optimized for low-precision formats (NVFP4, FP8).",{},[],{"content":1233,"meta":1234,"items":1235},"\u003Cb>New FP Format:\u003C/b> NVFP4 is a new 4-bit floating-point format that NVIDIA claims offers accuracy close to FP8 while using up to 3.5 times less memory than FP16.",{},[],{"id":1237,"type":14,"data":1238},"e3f4g5h6",{"text":1239},"Compared to the original Blackwell, NVFP4 performance has increased from 10 to 15 PFLOPS. For attention layers, the throughput of special function units (SFUs) has been doubled, accelerating computations on long contexts and reducing time-to-first-token.",{"id":1241,"type":27,"data":1242},"i7j8k9l0",{"text":1243,"level":137},"Memory Subsystem and Interconnect",{"id":1245,"type":37,"data":1246},"m1n2o3p4",{"style":39,"meta":1247,"items":1248},{},[1249,1253,1257,1261],{"content":1250,"meta":1251,"items":1252},"\u003Cb>Memory:\u003C/b> 288 GB of HBM3e with an 8192-bit bus, providing 8 TB/s of bandwidth.",{},[],{"content":1254,"meta":1255,"items":1256},"\u003Cb>NVLink 5:\u003C/b> Enables the connection of up to 576 GPUs into a single compute fabric.",{},[],{"content":1258,"meta":1259,"items":1260},"\u003Cb>Host Interface:\u003C/b> PCIe 6.0 x16.",{},[],{"content":1262,"meta":1263,"items":1264},"\u003Cb>Networking:\u003C/b> Integrates ConnectX-8 SuperNIC and Quantum-X800 InfiniBand / Spectrum-X Ethernet, with up to 800 Gb/s per GPU.",{},[],{"id":1266,"type":14,"data":1267},"q5r6s7t8",{"text":1268},"Blackwell Ultra's architectural changes signal a clear shift toward inference and reasoning tasks. The increased memory allows large models to reside entirely on the GPU, reducing latency. This positions the new accelerator as the foundation for modular and scalable AI factories.",{"id":1270,"type":27,"data":1271},"u9v0w1x2",{"text":1272,"level":30},"D-Matrix JetStream Card for Distributed Inference",{"id":1274,"type":19,"data":1275},"y3z4a5b6",{"caption":63,"withBorder":22,"withBackground":22,"stretched":22,"file":1276},{"url":1277},"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/596e4aad-c3c5-43bd-9033-2db3b57f0ec3/1184.webp",{"id":1279,"type":14,"data":1280},"c7d8e9f0",{"text":1281},"Startup d-Matrix has \u003Ca href=\"https://www.d-matrix.ai/announcements/jetstream/\">introduced\u003C/a> JetStream, a specialized I/O card designed to distribute AI inference workloads across servers. It works with the company's Corsair accelerators to reduce latency without requiring changes to existing network infrastructure.",{"id":1283,"type":27,"data":1284},"g1h2i3j4",{"text":1007,"level":137},{"id":1286,"type":37,"data":1287},"k5l6m7n8",{"style":39,"meta":1288,"items":1289},{},[1290,1294,1298],{"content":1291,"meta":1292,"items":1293},"\u003Cb>Form Factor:\u003C/b> PCIe 5.0 x16 expansion card.",{},[],{"content":1295,"meta":1296,"items":1297},"\u003Cb>Networking:\u003C/b> Two 200 GbE ports or one 400 GbE port with standard Ethernet (QSFP-DD).",{},[],{"content":1299,"meta":1300,"items":1301},"\u003Cb>Power Consumption:\u003C/b> Approximately 150W, cooled by a heatsink and heat pipes.",{},[],{"id":1303,"type":14,"data":1304},"o9p0q1r2",{"text":1305},"A single JetStream card serves up to four Corsair accelerators. The card eliminates the I/O bottleneck when scaling inference workloads, especially when a model is distributed across multiple servers, achieving a claimed network latency of just 2 μs. The company asserts that their solution is up to 10 times faster and three times more power-efficient than GPU-based alternatives for models over 100 billion parameters. Samples are available now, with mass production planned for the end of the year.",{"id":1307,"type":27,"data":1308},"s3t4u5v6",{"text":1309,"level":137},"Memory",{"id":1311,"type":27,"data":1312},"w7x8y9z0",{"text":1313,"level":30},"SK hynix Finalizes HBM4 Development",{"id":1315,"type":19,"data":1316},"a1b2c3d4e",{"caption":63,"withBorder":22,"withBackground":22,"stretched":22,"file":1317},{"url":1318},"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/bfd5484a-6e1f-4801-bf1a-527ff7b6c1ac/1184.webp",{"id":1320,"type":14,"data":1321},"f5g6h7i8j",{"text":1322},"SK hynix has \u003Ca href=\"https://news.skhynix.com/sk-hynix-completes-worlds-first-hbm4-development-and-readies-mass-production/\">announced\u003C/a> the completion of its HBM4 memory development. The company has already sent 12-layer samples to clients and is preparing for mass production in the second half of 2025. HBM4 is set to become a key component in future AI accelerators like NVIDIA's Rubin, offering double the bandwidth and significantly improved power efficiency over HBM3/3E.",{"id":1324,"type":27,"data":1325},"k9l0m1n2o",{"text":1007,"level":137},{"id":1327,"type":37,"data":1328},"p3q4r5s6t",{"style":39,"meta":1329,"items":1330},{},[1331,1335,1339,1343,1347],{"content":1332,"meta":1333,"items":1334},"\u003Cb>2048-bit Interface:\u003C/b> Doubles the width compared to HBM3/E, a first since 2015.",{},[],{"content":1336,"meta":1337,"items":1338},"\u003Cb>Data Transfer Rate:\u003C/b> Over 10 Gbps per pin, 25% higher than the official JEDEC standard for HBM3E.",{},[],{"content":1340,"meta":1341,"items":1342},"\u003Cb>Bandwidth:\u003C/b> Twice that of HBM3E.",{},[],{"content":1344,"meta":1345,"items":1346},"\u003Cb>Power Efficiency:\u003C/b> Improved by 40%.",{},[],{"content":1348,"meta":1349,"items":1350},"\u003Cb>AI Performance:\u003C/b> Expected to boost AI accelerator performance by up to 69% by eliminating memory bottlenecks.",{},[],{"id":1352,"type":14,"data":1353},"u7v8w9x0y",{"text":1354},"The memory is manufactured using a fifth-generation 10nm-class (1β nm) process and an innovative Advanced MR-MUF (Molded Underfill) assembly technique, which ensures mechanical stability and better heat dissipation. This new memory will be used in upcoming products from NVIDIA, AMD, Broadcom, and other AI industry leaders.",{"id":1356,"type":27,"data":1357},"z1a2b3c4d",{"text":1358,"level":137},"Storage",{"id":1360,"type":27,"data":1361},"e5f6g7h8i",{"text":1362,"level":30},"Solidigm D7-PS1010: A Liquid-Cooled E1.S SSD",{"id":1364,"type":19,"data":1365},"j9k0l1m2n",{"caption":63,"withBorder":22,"withBackground":22,"stretched":22,"file":1366},{"url":1367},"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/dfbfcb23-04b7-4dd8-8f67-38fc28f9aee8/1184.webp",{"id":1369,"type":14,"data":1370},"o3p4q5r6s",{"text":1371},"Solidigm (an SK hynix brand) has \u003Ca href=\"https://www.ixbt.com/news/2025/03/22/ssd-solidigm-d7-ps1010-e1-s-essd.html\">introduced\u003C/a> one of the world's first liquid-cooled enterprise SSDs, the D7-PS1010 E1.S. It is designed for high-density servers that lack traditional air cooling. The key feature is a cold plate that contacts both sides of the hot-swappable drive.",{"id":1373,"type":27,"data":1374},"t7u8v9w0x",{"text":1007,"level":137},{"id":1376,"type":37,"data":1377},"y1z2a3b4c",{"style":39,"meta":1378,"items":1379},{},[1380,1384,1388,1392],{"content":1381,"meta":1382,"items":1383},"\u003Cb>Form Factor:\u003C/b> E1.S with 9.5mm and 15mm thickness options.",{},[],{"content":1385,"meta":1386,"items":1387},"\u003Cb>Technology:\u003C/b> Based on 176-layer 3D TLC NAND and a PCIe 5.0 x4 (NVMe) interface.",{},[],{"content":1389,"meta":1390,"items":1391},"\u003Cb>Capacities:\u003C/b> Available in 3.84 TB and 7.68 TB for the liquid-cooled version.",{},[],{"content":1393,"meta":1394,"items":1395},"\u003Cb>Speeds:\u003C/b> Sequential read speeds up to 14,500 MB/s and write speeds up to 10,500 MB/s.",{},[],{"id":1397,"type":14,"data":1398},"d5e6f7g8h",{"text":1399},"Solidigm is working with server manufacturers to integrate this solution into AI infrastructures, including systems based on the NVIDIA HGX B300. Unlike traditional liquid cooling systems that cool only one side of an SSD, this new design cools both sides, improving thermal balance and reducing the risk of throttling. By eliminating the need for dedicated drive fans, it enables even denser server configurations and could accelerate the transition of data centers to fully liquid-cooled environments.",{"id":1401,"type":60,"data":1402},"i9j0k1l2m",{"quoteText":1403,"authorText":63},"Do you think data centers will eventually transition to fully liquid cooling? And is it necessary? Stay tuned for more hardware news.",{"id":1405,"type":14,"data":1406},"w6ZljsUvxc",{"text":1407},"#news #hardware",{"authorId":1409,"name":1410,"avatar":1411,"username":1412},"W3jnb37WdYhjXAZ3MUO7kkHpoAf2","Corey Maldonado","https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/9a3e623c-f419-4e4e-b9ec-406851cb78e1/1184.webp","id492585336165",[1414],{"emojiId":30,"count":30,"reactedByUser":22},{"comments":136,"favorites":137,"views":1416,"hits":1417},50,132,"2025-10-17T08:10:54.79","2025-10-17T20:09:14.155",{"postId":1421,"title":1422,"slug":1423,"preview":1424,"html":1425,"content":1426,"author":1643,"authorUsername":1647,"topic":1648,"status":138,"reactions":1649,"counters":1654,"createdAt":1657,"updatedAt":1658,"publishedPostId":145,"favorite":22},"94165de0-604f-4257-8311-1a0a99f73fce","AI agents for working with clients, mail, and databases: how the no-code constructor from OpenAI Agent Builder is used","220995-ai-agents-for-working-with-clients-mail-databases-openai-agent-builder","\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">Explore how to create powerful custom AI assistants without writing a single line of code. We'll cover core features, prompt examples, integration capabilities, and how to deploy your finished agent.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--media\">\u003Cfigure class=\"block-content content-block-image\">\u003Cimg src=\"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/30f3382a-89dd-437d-82bd-fb0e2c760e8a/1184.webp\" alt=\"The Agent Builder interface. Source: Medium\" class=\"rounded-xl medium-zoom-image\" loading=\"lazy\">\u003C/figure>\u003Cspan class=\"caption-text text-sm mt-3\">The Agent Builder interface. Source: Medium\u003C/span>\u003C/div>","\u003Cdiv class=\"content-wrapper break-words\">\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">Explore how to create powerful custom AI assistants without writing a single line of code. We'll cover core features, prompt examples, integration capabilities, and how to deploy your finished agent.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--media\">\u003Cfigure class=\"block-content content-block-image\">\u003Cimg src=\"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/30f3382a-89dd-437d-82bd-fb0e2c760e8a/1184.webp\" alt=\"The Agent Builder interface. Source: Medium\" class=\"rounded-xl medium-zoom-image\" loading=\"lazy\">\u003C/figure>\u003Cspan class=\"caption-text text-sm mt-3\">The Agent Builder interface. Source: Medium\u003C/span>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch2 class=\"block-content\">What is Agent Builder?\u003C/h2>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">Unveiled in October 2025, OpenAI's Agent Builder is a no-code platform for creating custom AI agents. These agents can range from simple chatbots to sophisticated business automation tools that analyze data or summarize meetings. A key feature is their ability to integrate with third-party applications like Google Drive, Slack, and Zoom.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">You can access the service on the \u003Ca href=\"https://platform.openai.com/agent-builder\">OpenAI Platform website\u003C/a>. To begin, log in with your ChatGPT account and link a payment method to cover token usage.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Chr class=\"block-content content-block-delimiter\">\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch2 class=\"block-content\">How It Works: The Node-Based Canvas\u003C/h2>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">The builder uses a visual, node-based system where each component is configured individually. The core component is the 'Agent' node, where you define the AI's role, instructions, and behavioral rules. You can also select the underlying OpenAI model for your agent, with options starting from GPT-3.5.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--media\">\u003Cfigure class=\"block-content content-block-image\">\u003Cvideo src=\"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/dda39b4f-8b18-41a2-97da-97bf139c3f68/bd1c6f56-025a-4ba3-b70e-f8261d4f3ffa5204707452659221237.mp4\" class=\"rounded-xl w-full\" preload=\"auto\" autoplay muted playsinline loop>Your browser does not support the video tag.\u003C/video>\u003C/figure>\u003Cspan class=\"caption-text text-sm mt-3\">The Agent Builder canvas and its component settings. Source: Testing Catalog News\u003C/span>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch3 class=\"block-content\">Core Capabilities\u003C/h3>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cul class=\"block-content content-block-list-unordered\">\u003Cli class=\"content-block-list-item\">\u003Cb>Web Search:\u003C/b> Natively search the internet for information.\u003C/li>\u003Cli class=\"content-block-list-item\">\u003Cb>Knowledge Base Analysis:\u003C/b> Analyze information from uploaded files.\u003C/li>\u003Cli class=\"content-block-list-item\">\u003Cb>Code Generation:\u003C/b> Write and execute code.\u003C/li>\u003Cli class=\"content-block-list-item\">\u003Cb>Autonomous Actions:\u003C/b> Fill out forms, send emails, and operate within other applications.\u003C/li>\u003Cli class=\"content-block-list-item\">\u003Cb>Image Generation:\u003C/b> Create images based on prompts.\u003C/li>\u003C/ul>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch3 class=\"block-content\">Advanced Nodes\u003C/h3>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">You can drag additional nodes from the side panel to add custom logic and functionality:\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Col class=\"block-content content-block-list-ordered\">\u003Cli class=\"content-block-list-item\">\u003Cb>File Search:\u003C/b> Allows the agent to use your uploaded documents as a dedicated knowledge base.\u003C/li>\u003Cli class=\"content-block-list-item\">\u003Cb>User Approval:\u003C/b> Pauses the agent's workflow to request confirmation from a human before proceeding with a critical action, like finalizing a payment or refund.\u003C/li>\u003C/ol>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Chr class=\"block-content content-block-delimiter\">\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch2 class=\"block-content\">Use Cases & Prompt Examples\u003C/h2>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">Developers are actively sharing creative use cases and ready-to-use instructions for various business needs.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch3 class=\"block-content\">Email Assistant\u003C/h3>\u003C/div>\u003Cfigure class=\"block-wrapper block-wrapper--media\">\u003Cdiv class=\"block-wrapper__content\">\u003Cblockquote class=\"block-quote block-quote--text-small\">\u003Cdiv class=\"block-quote__line\">\u003C/div>\u003Cdiv class=\"block-quote__text\">You are an email assistant. Prepare summaries of unread emails, draft replies in my style, and flag all urgent and personal messages.\u003C/div>\u003C/blockquote>\u003C/div>\u003C/figure>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">\u003Cb>Connected Apps:\u003C/b> Gmail, Google Drive\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--media\">\u003Cfigure class=\"block-content content-block-image\">\u003Cimg src=\"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/6b675e26-55ff-4a46-a50c-dbdd522ad9be/1184.webp\" alt=\"An AI agent for managing email. Source: X\" class=\"rounded-xl medium-zoom-image\" loading=\"lazy\">\u003C/figure>\u003Cspan class=\"caption-text text-sm mt-3\">An AI agent for managing email. Source: X\u003C/span>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch3 class=\"block-content\">Lead Generation Assistant\u003C/h3>\u003C/div>\u003Cfigure class=\"block-wrapper block-wrapper--media\">\u003Cdiv class=\"block-wrapper__content\">\u003Cblockquote class=\"block-quote block-quote--text-small\">\u003Cdiv class=\"block-quote__line\">\u003C/div>\u003Cdiv class=\"block-quote__text\">You are a lead generation assistant. Find 20 new companies in the specified industry that match our criteria. Extract the names of their CEOs from LinkedIn and add them to a spreadsheet.\u003C/div>\u003C/blockquote>\u003C/div>\u003C/figure>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">\u003Cb>Connected Apps:\u003C/b> HubSpot, LinkedIn, Google Sheets\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch2 class=\"block-content\">Ready-Made Templates\u003C/h2>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">The platform also provides several pre-built templates. For example, the \"Customer Service\" agent comes with pre-configured sub-agents for handling returns and subscription cancellations. You can customize this template to fit your specific business needs.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cul class=\"block-content content-block-list-unordered\">\u003Cli class=\"content-block-list-item\">Information retrieval bots\u003C/li>\u003Cli class=\"content-block-list-item\">Document analysis and comparison tools\u003C/li>\u003Cli class=\"content-block-list-item\">Planning and scheduling assistants\u003C/li>\u003Cli class=\"content-block-list-item\">Internal Q&A bots based on a corporate knowledge base\u003C/li>\u003C/ul>\u003C/div>\u003Cdiv class=\"block-wrapper--media\">\u003Cfigure class=\"block-content content-block-image\">\u003Cimg src=\"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/f9d26b6c-3a1a-4d21-ba13-cf875b9b0169/1184.webp\" alt=\"The menu of ready-made templates on the OpenAI Platform.\" class=\"rounded-xl medium-zoom-image\" loading=\"lazy\">\u003C/figure>\u003Cspan class=\"caption-text text-sm mt-3\">The menu of ready-made templates on the OpenAI Platform.\u003C/span>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch2 class=\"block-content\">Integrating with External Apps\u003C/h2>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">Agent Builder supports the Model Context Protocol (MCP), allowing you to connect to file storage, CRMs, and third-party apps. The catalog includes popular services like Gmail, Slack, Outlook, Dropbox, Google Drive, Zoom, and Microsoft Teams. For example, connecting to Gmail enables your agent to read emails, search your inbox, and draft or send replies.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">However, a current limitation is that Agent Builder doesn't natively support third-party services as input triggers. For instance, you can't directly have a bot respond to messages in Telegram; this requires manual integration. This contrasts with open-source tools like \u003Ca href=\"https://n8n.io/\">n8n\u003C/a>, which offer hundreds of pre-built input integrations.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--media\">\u003Cfigure class=\"block-content content-block-image\">\u003Cimg src=\"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/5ce02fb4-560e-41d7-91c5-5737cb7b6295/1184.webp\" alt=\"The n8n interface, showing an AI agent connected to Telegram as an input source. Source: YouTube channel 'Productive Sovet'\" class=\"rounded-xl medium-zoom-image\" loading=\"lazy\">\u003C/figure>\u003Cspan class=\"caption-text text-sm mt-3\">The n8n interface, showing an AI agent connected to Telegram as an input source. Source: YouTube channel 'Productive Sovet'\u003C/span>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch2 class=\"block-content\">Publishing Your AI Agent\u003C/h2>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">To publish your agent, you first need to verify your organization by providing details such as a registration address or website. Once verified, you have two main export options:\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cul class=\"block-content content-block-list-unordered\">\u003Cli class=\"content-block-list-item\">\u003Cb>ChatKit:\u003C/b> Embed a chat interface directly into your products using an OpenAI API key.\u003C/li>\u003Cli class=\"content-block-list-item\">\u003Cb>Agent SDK:\u003C/b> Export the agent as code for self-hosting and custom deployment.\u003C/li>\u003C/ul>\u003C/div>\u003Cdiv class=\"block-wrapper--media\">\u003Cfigure class=\"block-content content-block-image\">\u003Cimg src=\"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/242eeb9e-d304-415d-8a41-111a5e22cb95/1184.webp\" alt=\"Two options for publishing a completed AI agent. Source: Medium\" class=\"rounded-xl medium-zoom-image\" loading=\"lazy\">\u003C/figure>\u003Cspan class=\"caption-text text-sm mt-3\">Two options for publishing a completed AI agent. Source: Medium\u003C/span>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">\u003Ca href=\"/tag/news\" target=\"_blank\" rel=\"noopener noreferrer ugc\">#news\u003C/a>\u003Ca href=\"/tag/#ai\" target=\"_blank\" rel=\"noopener noreferrer ugc\"> #ai\u003C/a>\u003Ca href=\"/tag/#openai\" target=\"_blank\" rel=\"noopener noreferrer ugc\"> #openai\u003C/a>\u003C/p>\u003C/div>\u003C/div>",{"time":1427,"blocks":1428,"version":123},1760608508217,[1429,1433,1439,1443,1447,1451,1454,1458,1462,1468,1472,1497,1501,1505,1518,1521,1525,1529,1533,1537,1541,1547,1551,1555,1559,1563,1567,1588,1594,1598,1602,1606,1612,1616,1620,1633,1639],{"id":1430,"type":14,"data":1431},"sV8fR3mN2w",{"text":1432},"Explore how to create powerful custom AI assistants without writing a single line of code. We'll cover core features, prompt examples, integration capabilities, and how to deploy your finished agent.",{"id":1434,"type":19,"data":1435},"eX7jY5bC8z",{"caption":1436,"withBorder":22,"withBackground":22,"stretched":22,"file":1437},"The Agent Builder interface. Source: Medium",{"url":1438},"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/30f3382a-89dd-437d-82bd-fb0e2c760e8a/1184.webp",{"id":1440,"type":27,"data":1441},"T6pI0uV7xR",{"text":1442,"level":137},"What is Agent Builder?",{"id":1444,"type":14,"data":1445},"L9wE2sF4gH",{"text":1446},"Unveiled in October 2025, OpenAI's Agent Builder is a no-code platform for creating custom AI agents. These agents can range from simple chatbots to sophisticated business automation tools that analyze data or summarize meetings. A key feature is their ability to integrate with third-party applications like Google Drive, Slack, and Zoom.",{"id":1448,"type":14,"data":1449},"K3mN8zB1vX",{"text":1450},"You can access the service on the \u003Ca href=\"https://platform.openai.com/agent-builder\">OpenAI Platform website\u003C/a>. To begin, log in with your ChatGPT account and link a payment method to cover token usage.",{"id":1452,"type":273,"data":1453},"Pq7sD9fG2k",{},{"id":1455,"type":27,"data":1456},"J1hY5cV6bR",{"text":1457,"level":137},"How It Works: The Node-Based Canvas",{"id":1459,"type":14,"data":1460},"G4fH9jL3pW",{"text":1461},"The builder uses a visual, node-based system where each component is configured individually. The core component is the 'Agent' node, where you define the AI's role, instructions, and behavioral rules. You can also select the underlying OpenAI model for your agent, with options starting from GPT-3.5.",{"id":1463,"type":19,"data":1464},"R8tY2uI5oZ",{"caption":1465,"withBorder":22,"withBackground":22,"stretched":22,"file":1466},"The Agent Builder canvas and its component settings. Source: Testing Catalog News",{"url":1467},"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/dda39b4f-8b18-41a2-97da-97bf139c3f68/bd1c6f56-025a-4ba3-b70e-f8261d4f3ffa5204707452659221237.mp4",{"id":1469,"type":27,"data":1470},"O3iP7sA9dF",{"text":1471,"level":30},"Core Capabilities",{"id":1473,"type":37,"data":1474},"M6nB1vX5cZ",{"style":39,"meta":1475,"items":1476},{},[1477,1481,1485,1489,1493],{"content":1478,"meta":1479,"items":1480},"\u003Cb>Web Search:\u003C/b> Natively search the internet for information.",{},[],{"content":1482,"meta":1483,"items":1484},"\u003Cb>Knowledge Base Analysis:\u003C/b> Analyze information from uploaded files.",{},[],{"content":1486,"meta":1487,"items":1488},"\u003Cb>Code Generation:\u003C/b> Write and execute code.",{},[],{"content":1490,"meta":1491,"items":1492},"\u003Cb>Autonomous Actions:\u003C/b> Fill out forms, send emails, and operate within other applications.",{},[],{"content":1494,"meta":1495,"items":1496},"\u003Cb>Image Generation:\u003C/b> Create images based on prompts.",{},[],{"id":1498,"type":27,"data":1499},"X5cZ9vB1mN",{"text":1500,"level":30},"Advanced Nodes",{"id":1502,"type":14,"data":1503},"V6bN2mK8jL",{"text":1504},"You can drag additional nodes from the side panel to add custom logic and functionality:",{"id":1506,"type":37,"data":1507},"T1yP5oI7uR",{"style":286,"meta":1508,"items":1509},{"counterType":288},[1510,1514],{"content":1511,"meta":1512,"items":1513},"\u003Cb>File Search:\u003C/b> Allows the agent to use your uploaded documents as a dedicated knowledge base.",{},[],{"content":1515,"meta":1516,"items":1517},"\u003Cb>User Approval:\u003C/b> Pauses the agent's workflow to request confirmation from a human before proceeding with a critical action, like finalizing a payment or refund.",{},[],{"id":1519,"type":273,"data":1520},"U4iP8sA9dF",{},{"id":1522,"type":27,"data":1523},"I7uR1tY3pW",{"text":1524,"level":137},"Use Cases & Prompt Examples",{"id":1526,"type":14,"data":1527},"S9dF3gH6jL",{"text":1528},"Developers are actively sharing creative use cases and ready-to-use instructions for various business needs.",{"id":1530,"type":27,"data":1531},"D2kF6gH9jL",{"text":1532,"level":30},"Email Assistant",{"id":1534,"type":60,"data":1535},"L3pW7sA9dF",{"quoteText":1536,"authorText":63},"You are an email assistant. Prepare summaries of unread emails, draft replies in my style, and flag all urgent and personal messages.",{"id":1538,"type":14,"data":1539},"A9dF3gH6jL",{"text":1540},"\u003Cb>Connected Apps:\u003C/b> Gmail, Google Drive",{"id":1542,"type":19,"data":1543},"F3gH6jL1pW",{"caption":1544,"withBorder":22,"withBackground":22,"stretched":22,"file":1545},"An AI agent for managing email. Source: X",{"url":1546},"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/6b675e26-55ff-4a46-a50c-dbdd522ad9be/1184.webp",{"id":1548,"type":27,"data":1549},"G6jL1pW7sA",{"text":1550,"level":30},"Lead Generation Assistant",{"id":1552,"type":60,"data":1553},"J1pW7sA9dF",{"quoteText":1554,"authorText":63},"You are a lead generation assistant. Find 20 new companies in the specified industry that match our criteria. Extract the names of their CEOs from LinkedIn and add them to a spreadsheet.",{"id":1556,"type":14,"data":1557},"P7sA9dF3gH",{"text":1558},"\u003Cb>Connected Apps:\u003C/b> HubSpot, LinkedIn, Google Sheets",{"id":1560,"type":27,"data":1561},"W7sA9dF3gH",{"text":1562,"level":137},"Ready-Made Templates",{"id":1564,"type":14,"data":1565},"E9dF3gH6jL",{"text":1566},"The platform also provides several pre-built templates. For example, the \"Customer Service\" agent comes with pre-configured sub-agents for handling returns and subscription cancellations. You can customize this template to fit your specific business needs.",{"id":1568,"type":37,"data":1569},"T3gH6jL1pW",{"style":39,"meta":1570,"items":1571},{},[1572,1576,1580,1584],{"content":1573,"meta":1574,"items":1575},"Information retrieval bots",{},[],{"content":1577,"meta":1578,"items":1579},"Document analysis and comparison tools",{},[],{"content":1581,"meta":1582,"items":1583},"Planning and scheduling assistants",{},[],{"content":1585,"meta":1586,"items":1587},"Internal Q&A bots based on a corporate knowledge base",{},[],{"id":1589,"type":19,"data":1590},"Q6jL1pW7sA",{"caption":1591,"withBorder":22,"withBackground":22,"stretched":22,"file":1592},"The menu of ready-made templates on the OpenAI Platform.",{"url":1593},"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/f9d26b6c-3a1a-4d21-ba13-cf875b9b0169/1184.webp",{"id":1595,"type":27,"data":1596},"C9vB1mN7kF",{"text":1597,"level":137},"Integrating with External Apps",{"id":1599,"type":14,"data":1600},"B1mN7kF4gH",{"text":1601},"Agent Builder supports the Model Context Protocol (MCP), allowing you to connect to file storage, CRMs, and third-party apps. The catalog includes popular services like Gmail, Slack, Outlook, Dropbox, Google Drive, Zoom, and Microsoft Teams. For example, connecting to Gmail enables your agent to read emails, search your inbox, and draft or send replies.",{"id":1603,"type":14,"data":1604},"N7kF4gH8jL",{"text":1605},"However, a current limitation is that Agent Builder doesn't natively support third-party services as input triggers. For instance, you can't directly have a bot respond to messages in Telegram; this requires manual integration. This contrasts with open-source tools like \u003Ca href=\"https://n8n.io/\">n8n\u003C/a>, which offer hundreds of pre-built input integrations.",{"id":1607,"type":19,"data":1608},"K4gH8jL2pW",{"caption":1609,"withBorder":22,"withBackground":22,"stretched":22,"file":1610},"The n8n interface, showing an AI agent connected to Telegram as an input source. Source: YouTube channel 'Productive Sovet'",{"url":1611},"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/5ce02fb4-560e-41d7-91c5-5737cb7b6295/1184.webp",{"id":1613,"type":27,"data":1614},"H8jL2pW5cZ",{"text":1615,"level":137},"Publishing Your AI Agent",{"id":1617,"type":14,"data":1618},"L2pW5cZ9vB",{"text":1619},"To publish your agent, you first need to verify your organization by providing details such as a registration address or website. Once verified, you have two main export options:",{"id":1621,"type":37,"data":1622},"Z9vB1mN7kF",{"style":39,"meta":1623,"items":1624},{},[1625,1629],{"content":1626,"meta":1627,"items":1628},"\u003Cb>ChatKit:\u003C/b> Embed a chat interface directly into your products using an OpenAI API key.",{},[],{"content":1630,"meta":1631,"items":1632},"\u003Cb>Agent SDK:\u003C/b> Export the agent as code for self-hosting and custom deployment.",{},[],{"id":1634,"type":19,"data":1635},"V1mN7kF4gH",{"caption":1636,"withBorder":22,"withBackground":22,"stretched":22,"file":1637},"Two options for publishing a completed AI agent. Source: Medium",{"url":1638},"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/242eeb9e-d304-415d-8a41-111a5e22cb95/1184.webp",{"id":1640,"type":14,"data":1641},"u15HQzg74-",{"text":1642},"#news #ai #openai",{"authorId":1644,"name":1645,"avatar":1646,"username":1647},"5Z1ou9MAx3T3lu666UcrJu4W2FD3","Hans-Günther Fliegner","https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/31eb57c5-3add-4374-85a6-bd7fe17d3167/1184.webp","id653426925003","AI",[1650,1651,1653],{"emojiId":30,"count":137,"reactedByUser":22},{"emojiId":1652,"count":141,"reactedByUser":22},11,{"emojiId":318,"count":141,"reactedByUser":22},{"comments":136,"favorites":137,"views":1655,"hits":1656},29,180,"2025-10-16T09:50:39.451","2025-10-16T14:27:54.778",{"postId":1660,"title":1661,"slug":1662,"preview":1663,"html":1664,"content":1665,"author":1721,"authorUsername":1725,"topic":210,"status":138,"reactions":1726,"counters":1730,"createdAt":1732,"updatedAt":1733,"publishedPostId":145,"favorite":22},"a26abe7d-e65d-4c34-8d83-1f0557637c04","Honor Unveils the Robot Phone Concept — A Smartphone with a Retractable Camera and AI Features","065838-honor-robot-phone-concept-retractable-camera-ai-features","\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">Honor has teased a futuristic smartphone concept called the 'Robot Phone,' distinguished by a unique motorized main camera. This innovative feature is powered by AI, allowing the camera to move autonomously.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--media\">\u003Cfigure class=\"block-content content-block-image\">\u003Cimg src=\"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/be874606-349d-4757-a25f-d5b8ec9b311f/1184.webp\" alt=\"Image Source: Honor\" class=\"rounded-xl medium-zoom-image\" loading=\"lazy\">\u003C/figure>\u003Cspan class=\"caption-text text-sm mt-3\">Image Source: Honor\u003C/span>\u003C/div>","\u003Cdiv class=\"content-wrapper break-words\">\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">Honor has teased a futuristic smartphone concept called the 'Robot Phone,' distinguished by a unique motorized main camera. This innovative feature is powered by AI, allowing the camera to move autonomously.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--media\">\u003Cfigure class=\"block-content content-block-image\">\u003Cimg src=\"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/be874606-349d-4757-a25f-d5b8ec9b311f/1184.webp\" alt=\"Image Source: Honor\" class=\"rounded-xl medium-zoom-image\" loading=\"lazy\">\u003C/figure>\u003Cspan class=\"caption-text text-sm mt-3\">Image Source: Honor\u003C/span>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch3 class=\"block-content\">Key Features of the AI Camera\u003C/h3>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cul class=\"block-content content-block-list-unordered\">\u003Cli class=\"content-block-list-item\">\u003Cb>Autonomous Movement:\u003C/b> The camera automatically extends from the phone's body and retracts when not in use.\u003C/li>\u003Cli class=\"content-block-list-item\">\u003Cb>Smart Tracking:\u003C/b> It can intelligently move and orient itself to follow subjects without manual input.\u003C/li>\u003Cli class=\"content-block-list-item\">\u003Cb>Integrated Design:\u003C/b> The entire mechanism is housed within a thickened rear camera module to accommodate the moving parts.\u003C/li>\u003C/ul>\u003C/div>\u003Cdiv class=\"block-wrapper--media\">\u003Cfigure class=\"block-content content-block-image\">\u003Cimg src=\"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/9da5b365-5133-4a6a-ab28-16c144907650/1184.webp\" alt=\"Image\" class=\"rounded-xl medium-zoom-image\" loading=\"lazy\">\u003C/figure>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch3 class=\"block-content\">What We Know So Far\u003C/h3>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">While the concept is intriguing, Honor has not released any technical specifications for the device. According to \u003Ca href=\"https://www.theverge.com/news/799944/honor-robot-phone-tease-announcement\">The Verge\u003C/a>, the company is planning a full reveal and live demonstration of the phone at the MWC (Mobile World Congress) exhibition in March 2026.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--media\">\u003Cfigure class=\"block-content content-block-image\">\u003Cimg src=\"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/1792d265-fcd9-44c0-a155-15afcfe811ff/1184.webp\" alt=\"Image\" class=\"rounded-xl medium-zoom-image\" loading=\"lazy\">\u003C/figure>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">\u003Ca href=\"/tag/news\" target=\"_blank\" rel=\"noopener noreferrer ugc\">#news\u003C/a>\u003Ca href=\"/tag/#honor\" target=\"_blank\" rel=\"noopener noreferrer ugc\"> #honor\u003C/a>\u003C/p>\u003C/div>\u003C/div>",{"time":1666,"blocks":1667,"version":123},1760600413804,[1668,1672,1678,1682,1699,1704,1708,1712,1717],{"id":1669,"type":14,"data":1670},"bV9dC1sHqN",{"text":1671},"Honor has teased a futuristic smartphone concept called the 'Robot Phone,' distinguished by a unique motorized main camera. This innovative feature is powered by AI, allowing the camera to move autonomously.",{"id":1673,"type":19,"data":1674},"aC7xR5fPqV",{"caption":1675,"withBorder":22,"withBackground":22,"stretched":22,"file":1676},"Image Source: Honor",{"url":1677},"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/be874606-349d-4757-a25f-d5b8ec9b311f/1184.webp",{"id":1679,"type":27,"data":1680},"sZ4yT8wEaB",{"text":1681,"level":30},"Key Features of the AI Camera",{"id":1683,"type":37,"data":1684},"wG6jK3mNfZ",{"style":39,"meta":1685,"items":1686},{},[1687,1691,1695],{"content":1688,"meta":1689,"items":1690},"\u003Cb>Autonomous Movement:\u003C/b> The camera automatically extends from the phone's body and retracts when not in use.",{},[],{"content":1692,"meta":1693,"items":1694},"\u003Cb>Smart Tracking:\u003C/b> It can intelligently move and orient itself to follow subjects without manual input.",{},[],{"content":1696,"meta":1697,"items":1698},"\u003Cb>Integrated Design:\u003C/b> The entire mechanism is housed within a thickened rear camera module to accommodate the moving parts.",{},[],{"id":1700,"type":19,"data":1701},"xY1hV7uJdC",{"caption":63,"withBorder":22,"withBackground":22,"stretched":22,"file":1702},{"url":1703},"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/9da5b365-5133-4a6a-ab28-16c144907650/1184.webp",{"id":1705,"type":27,"data":1706},"mQ5pB4fLgH",{"text":1707,"level":30},"What We Know So Far",{"id":1709,"type":14,"data":1710},"rN2kE9jXzO",{"text":1711},"While the concept is intriguing, Honor has not released any technical specifications for the device. According to \u003Ca href=\"https://www.theverge.com/news/799944/honor-robot-phone-tease-announcement\">The Verge\u003C/a>, the company is planning a full reveal and live demonstration of the phone at the MWC (Mobile World Congress) exhibition in March 2026.",{"id":1713,"type":19,"data":1714},"uO3iF6gWzR",{"caption":63,"withBorder":22,"withBackground":22,"stretched":22,"file":1715},{"url":1716},"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/1792d265-fcd9-44c0-a155-15afcfe811ff/1184.webp",{"id":1718,"type":14,"data":1719},"LQTs4wr68C",{"text":1720},"#news #honor",{"authorId":1722,"name":1723,"avatar":1724,"username":1725},"1OEyVAS5ezRsF6LV34JwGeKqk6m1","lucía","https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/a78da6ce-961a-474a-8b80-646165b980a0/1184.webp","id177965120723",[1727,1728],{"emojiId":141,"count":30,"reactedByUser":22},{"emojiId":1729,"count":141,"reactedByUser":22},7,{"comments":136,"favorites":137,"views":320,"hits":1731},120,"2025-10-16T07:39:25.173","2025-10-16T14:28:00.83",{"postId":1735,"title":1736,"slug":1737,"preview":1738,"html":1739,"content":1740,"author":1807,"authorUsername":1811,"topic":1812,"status":138,"reactions":1813,"counters":1816,"createdAt":1819,"updatedAt":1820,"publishedPostId":145,"favorite":22},"28d9e66e-b4cb-47f8-a431-73e0163952a0","Kryptos Code Solution Found After 35 Years","797930-kryptos-code-solution-found-after-35-years","\u003Cdiv class=\"block-wrapper--media\">\u003Cfigure class=\"block-content content-block-image\">\u003Cimg src=\"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/5c923000-a413-44a2-8964-5f8e3e562403/1184.webp\" alt=\"The Kryptos sculpture at CIA headquarters in Virginia.\" class=\"rounded-xl medium-zoom-image\" loading=\"lazy\">\u003C/figure>\u003Cspan class=\"caption-text text-sm mt-3\">The Kryptos sculpture at CIA headquarters in Virginia.\u003C/span>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">The enigmatic copper sculpture, engraved with four coded messages, has captivated cryptographers since artist Jim Sanborn installed it at CIA headquarters in 1990. While the first three sections were cracked in the 1990s, the fourth and final passage, known as K4, has eluded even the most brilliant minds—until now.\u003C/p>\u003C/div>","\u003Cdiv class=\"content-wrapper break-words\">\u003Cdiv class=\"block-wrapper--media\">\u003Cfigure class=\"block-content content-block-image\">\u003Cimg src=\"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/5c923000-a413-44a2-8964-5f8e3e562403/1184.webp\" alt=\"The Kryptos sculpture at CIA headquarters in Virginia.\" class=\"rounded-xl medium-zoom-image\" loading=\"lazy\">\u003C/figure>\u003Cspan class=\"caption-text text-sm mt-3\">The Kryptos sculpture at CIA headquarters in Virginia.\u003C/span>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch2 class=\"block-content\">After 35 Years, the Final Code of the CIA's Kryptos Sculpture is Cracked\u003C/h2>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">The enigmatic copper sculpture, engraved with four coded messages, has captivated cryptographers since artist Jim Sanborn installed it at CIA headquarters in 1990. While the first three sections were cracked in the 1990s, the fourth and final passage, known as K4, has eluded even the most brilliant minds—until now.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">In a surprising turn of events, two writers have uncovered the solution, not by cracking the code, but by finding the plaintext answer hidden in plain sight within the Smithsonian Institution's archives.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch3 class=\"block-content\">An Accidental Discovery\u003C/h3>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">After years of failed attempts by codebreakers, Sanborn was preparing to auction the solution. However, on September 3, he received an unexpected email from journalists Jarett Kobek and Richard Byrne containing the complete decrypted text.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">Their breakthrough came after discovering that Sanborn's \"coding tables\" were part of the Smithsonian collection. After photographing the documents, Kobek identified glued-together scraps that revealed the original plaintext for K4, including the previously released clues \"BERLIN CLOCK\" and \"EAST NORTHEAST\".\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">Sanborn confirmed the solution's authenticity, admitting he had accidentally included the crucial scraps while archiving his documents years ago. He has since requested the Smithsonian seal the files for the next 50 years, a request they have granted.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Ch3 class=\"block-content\">Words vs. Method: The Controversy Continues\u003C/h3>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">Following the discovery, RR Auction, the firm managing the sale, warned Kobek and Byrne against publishing the text, threatening legal action. The pair has stated they have no intention of releasing the solution publicly.\u003C/p>\u003C/div>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">The auction house clarified that knowing the plaintext isn't the same as solving the puzzle. A true solution requires understanding the encryption method and artistic context, a secret only the auction winner will learn.\u003C/p>\u003C/div>\u003Cfigure class=\"block-wrapper block-wrapper--media\">\u003Cdiv class=\"block-wrapper__content\">\u003Cblockquote class=\"block-quote block-quote--text-small block-quote--has-author\">\u003Cdiv class=\"block-quote__line\">\u003C/div>\u003Cdiv class=\"block-quote__text\">It's one thing to have the words. It's another to reveal the method. If you can't show the method, your solution isn't valid.\u003C/div>\u003Cdiv class=\"block-quote__author\">Ilonka Dunin, Kryptos fan group moderator\u003C/div>\u003C/blockquote>\u003C/div>\u003C/figure>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">Regardless of the outcome, the fascination with Kryptos is sure to endure. The sculpture remains a testament to the allure of unsolved mysteries.\u003C/p>\u003C/div>\u003Cfigure class=\"block-wrapper block-wrapper--media\">\u003Cdiv class=\"block-wrapper__content\">\u003Cblockquote class=\"block-quote block-quote--text-small block-quote--has-author\">\u003Cdiv class=\"block-quote__line\">\u003C/div>\u003Cdiv class=\"block-quote__text\">For a piece of art to hold someone's attention for 10 minutes is an achievement. Sanborn created a work that has captivated people for 35 years.\u003C/div>\u003Cdiv class=\"block-quote__author\">Ilonka Dunin\u003C/div>\u003C/blockquote>\u003C/div>\u003C/figure>\u003Cdiv class=\"block-wrapper--default\">\u003Cp class=\"block-content\">\u003Ca href=\"/tag/news\" target=\"_blank\" rel=\"noopener noreferrer ugc\">#news\u003C/a>\u003C/p>\u003C/div>\u003C/div>",{"time":1741,"blocks":1742,"version":123},1760698165861,[1743,1749,1753,1757,1761,1765,1769,1773,1777,1781,1785,1789,1794,1798,1803],{"id":1744,"type":19,"data":1745},"e5702edd-5",{"caption":1746,"withBorder":22,"withBackground":22,"stretched":22,"file":1747},"The Kryptos sculpture at CIA headquarters in Virginia.",{"url":1748},"https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/5c923000-a413-44a2-8964-5f8e3e562403/1184.webp",{"id":1750,"type":27,"data":1751},"c9a2b1f8-0",{"text":1752,"level":137},"After 35 Years, the Final Code of the CIA's Kryptos Sculpture is Cracked",{"id":1754,"type":14,"data":1755},"a3d4e5f6-1",{"text":1756},"The enigmatic copper sculpture, engraved with four coded messages, has captivated cryptographers since artist Jim Sanborn installed it at CIA headquarters in 1990. While the first three sections were cracked in the 1990s, the fourth and final passage, known as K4, has eluded even the most brilliant minds—until now.",{"id":1758,"type":14,"data":1759},"b1c2d3e4-2",{"text":1760},"In a surprising turn of events, two writers have uncovered the solution, not by cracking the code, but by finding the plaintext answer hidden in plain sight within the Smithsonian Institution's archives.",{"id":1762,"type":27,"data":1763},"d5e6f7g8-3",{"text":1764,"level":30},"An Accidental Discovery",{"id":1766,"type":14,"data":1767},"f9g8h7i6-4",{"text":1768},"After years of failed attempts by codebreakers, Sanborn was preparing to auction the solution. However, on September 3, he received an unexpected email from journalists Jarett Kobek and Richard Byrne containing the complete decrypted text.",{"id":1770,"type":14,"data":1771},"h3i4j5k6-5",{"text":1772},"Their breakthrough came after discovering that Sanborn's \"coding tables\" were part of the Smithsonian collection. After photographing the documents, Kobek identified glued-together scraps that revealed the original plaintext for K4, including the previously released clues \"BERLIN CLOCK\" and \"EAST NORTHEAST\".",{"id":1774,"type":14,"data":1775},"i7j8k9l0-6",{"text":1776},"Sanborn confirmed the solution's authenticity, admitting he had accidentally included the crucial scraps while archiving his documents years ago. He has since requested the Smithsonian seal the files for the next 50 years, a request they have granted.",{"id":1778,"type":27,"data":1779},"k1l2m3n4-7",{"text":1780,"level":30},"Words vs. Method: The Controversy Continues",{"id":1782,"type":14,"data":1783},"l5m6n7o8-8",{"text":1784},"Following the discovery, RR Auction, the firm managing the sale, warned Kobek and Byrne against publishing the text, threatening legal action. The pair has stated they have no intention of releasing the solution publicly.",{"id":1786,"type":14,"data":1787},"m9n0o1p2-9",{"text":1788},"The auction house clarified that knowing the plaintext isn't the same as solving the puzzle. A true solution requires understanding the encryption method and artistic context, a secret only the auction winner will learn.",{"id":1790,"type":60,"data":1791},"n3o4p5q6-a",{"quoteText":1792,"authorText":1793},"It's one thing to have the words. It's another to reveal the method. If you can't show the method, your solution isn't valid.","Ilonka Dunin, Kryptos fan group moderator",{"id":1795,"type":14,"data":1796},"o7p8q9r0-b",{"text":1797},"Regardless of the outcome, the fascination with Kryptos is sure to endure. The sculpture remains a testament to the allure of unsolved mysteries.",{"id":1799,"type":60,"data":1800},"p1q2r3s4-c",{"quoteText":1801,"authorText":1802},"For a piece of art to hold someone's attention for 10 minutes is an achievement. Sanborn created a work that has captivated people for 35 years.","Ilonka Dunin",{"id":1804,"type":14,"data":1805},"MCcfqLVVFe",{"text":1806},"#news",{"authorId":1808,"name":1809,"avatar":1810,"username":1811},"ANx5bAorkDSfIPu4rPRs8XHqTaW2","Austin Dunn","https://media.suddo.io/file/2f3af38f-ffaf-465d-b6b8-b927250a64dc/e486133f-aff5-44bc-ba39-dc0000351764/1184.webp","id616693414223","SECURITY",[1814,1815],{"emojiId":141,"count":137,"reactedByUser":22},{"emojiId":318,"count":141,"reactedByUser":22},{"comments":136,"favorites":141,"views":1817,"hits":1818},156,87,"2025-10-17T10:24:32.813","2025-10-18T08:39:53.679",["Reactive",1822],{"$si18n:cached-locale-configs":1823,"$si18n:resolved-locale":63,"$sshowLoginModal":22},{},["Set"],["ShallowReactive",1826],{"post-hardware-445061-electronics-conquered-new-heights-scientists-stacked-six-cmos-transistors-on-top":-1,"post-slug-hardware-445061-electronics-conquered-new-heights-scientists-stacked-six-cmos-transistors-on-top":-1,"feed-global-none-none-hot-none-0":-1},true,"/hardware/445061-electronics-conquered-new-heights-scientists-stacked-six-cmos-transistors-on-top",{"auth":1830,"app":1831,"theme":1832,"main":1921,"i18n":1923,"ui":1927},{"logged":22,"user":145,"authReady":22},{"lastUpdatedPostId":145,"lastUpdateTime":145},{"layout":1833,"themeName":1834,"routerTransition":1835,"routerTransitionDuration":136,"rtl":22,"boxed":1836,"sidebar":1838,"footer":1844,"responsive":1845,"toolbarHeight":1852,"viewPadding":1849,"headerBarHeight":1854,"colors":1855,"borderRadius":1883,"lineHeight":1886,"fontSize":1888,"fontWeight":1891,"fontFamily":1894,"typography":1896,"isThemeSetByUser":22},"VerticalNav","light","fade-up",{"enabled":1827,"toolbar":1827,"width":1837},1600,{"autoClose":1827,"collapsed":22,"autoCloseBreakpoint":1839,"animEase":1840,"animDuration":1841,"openWidth":1842,"closeWidth":1843},1000,"ease-in-out",0.3,300,64,{"show":1827},{"breakpoint":1846,"override":1847},700,{"viewPadding":1848,"toolbarHeight":1851},{"desk":1849,"mobile":1850},40,20,{"desk":1852,"mobile":1853},80,70,60,{"light":1856,"dark":1871},{"sidebarBackground":1857,"bodyBackground":1858,"background":1857,"backgroundSecondary":1859,"text":1860,"textSecondary":1861,"textTertiary":1862,"border":1863,"hover":1864,"primary":1865,"info":1866,"success":1867,"warning":1868,"error":1869,"extra1":1866,"extra2":1870,"extra3":1868,"extra4":1869},"rgb(255, 255, 255)","rgb(245, 247, 249)","#F2F2F2","rgb(0, 0, 0)","#000","rgb(114, 115, 115)","rgb(226, 230, 233)","rgba(221, 224, 225, 0.5)","hsl(221.2, 83.2%, 53.3%)","rgb(97, 102, 255)","rgb(0, 179, 122)","rgb(255, 183, 0)","rgb(255, 0, 85)","rgb(255, 97, 200)",{"sidebarBackground":1872,"bodyBackground":1873,"background":1872,"backgroundSecondary":1874,"text":1875,"textSecondary":1876,"textTertiary":1877,"border":1878,"hover":1879,"primary":1880,"info":1866,"success":1881,"warning":1868,"error":1882,"extra1":1866,"extra2":1870,"extra3":1868,"extra4":1869},"rgb(35, 35, 36)","rgb(22, 22, 23)","rgb(44, 44, 45)","rgb(201, 204, 207)","#c9cccf","rgb(150, 157, 160)","rgb(53, 52, 54)","rgb(58, 58, 58)","rgb(65, 138, 244)","rgb(51, 160, 62)","rgb(214, 51, 62)",{"default":1884,"small":1885},"8px","4px",{"default":1887},"1.25",{"default":1889,"cardTitle":1890},"16px","18px",{"default":1892,"strong":1893,"cardTitle":1893},"400","600",{"default":1895,"display":1895,"mono":1895},"Roboto, system-ui, -apple-system, Segoe UI, Roboto, Ubuntu, Cantarell, Noto Sans, sans-serif",{"h1":1897,"h2":1902,"h3":1905,"h4":1908,"h5":1911,"h6":1914,"p":1918},{"fontFamily":1898,"fontSize":1899,"fontWeight":1900,"lineHeight":1901},"{fontFamily.display}","30px","700","38",{"fontFamily":1898,"fontSize":1903,"fontWeight":1900,"lineHeight":1904},"26px","33",{"fontFamily":1898,"fontSize":1906,"fontWeight":1900,"lineHeight":1907},"22px","28",{"fontFamily":1898,"fontSize":1890,"fontWeight":1909,"lineHeight":1910},"500","23",{"fontFamily":1898,"fontSize":1912,"fontWeight":1900,"lineHeight":1913},"14px","18",{"fontFamily":1915,"fontSize":1916,"fontWeight":1909,"lineHeight":1917},"{fontFamily.default}","12px","15",{"fontFamily":1915,"fontSize":1919,"lineHeight":1920},"{fontSize.default}","20",{"API_URL":-1,"forceRefresh":1922,"loadingBar":145},1760989773173,{"locale":1924,"availableLocales":1926},["Ref",1925],"en",[1925],{"feedRefreshKey":136}]