devtake.dev — NvidiaArticles on devtake.dev covering Nvidia.https://devtake.dev/en-usSpaceX's S-1 revealed who's paying for Colossus. Anthropic just locked in $45B through 2029.https://devtake.dev/article/anthropic-spacex-colossus-15-billion/https://devtake.dev/article/anthropic-spacex-colossus-15-billion/Anthropic is paying SpaceX $1.25 billion a month for Colossus 1 and 2 capacity. The contract runs through May 2029 and books about 83% of SpaceX's revenue.Thu, 21 May 2026 11:45:00 GMTaianthropicspacexcolossusai-infrastructurenvidiagb200data-centersclaudedieter-morelliAlibaba's new Zhenwu M890 chip is 3x faster and aimed straight at agent workloadshttps://devtake.dev/article/alibaba-zhenwu-m890-qwen-3-7-max/https://devtake.dev/article/alibaba-zhenwu-m890-qwen-3-7-max/Alibaba showed the Zhenwu M890 at its Cloud Summit on May 19. 144 GB of memory, 800 GB/s interchip bandwidth, and Qwen3.7-Max riding on top.Wed, 20 May 2026 10:00:00 GMThardwarealibabaai-chipscustom-siliconqwenai-infrastructurenvidiaai-agentssemiconductorhiro-tanakaCisco cut 4,000 jobs on its record revenue day. AI infrastructure orders jumped to $9 billion.https://devtake.dev/article/cisco-4000-layoffs-record-q3-revenue/https://devtake.dev/article/cisco-4000-layoffs-record-q3-revenue/Cisco posted $15.8B in Q3 FY26 revenue on May 14, up 12% YoY, then sent layoff notices to 4,000 staff the same day. AI orders are now tracking to $9B for the fiscal year.Tue, 19 May 2026 09:30:00 GMThardwareciscolayoffsai-infrastructurenetworkingrestructuringearningsai-chipshiro-tanakaCerebras priced its IPO at $185 and closed at $311. Andrew Feldman and Sean Lie became billionaires.https://devtake.dev/article/cerebras-ipo-debut-nasdaq/https://devtake.dev/article/cerebras-ipo-debut-nasdaq/Cerebras raised $5.55 billion on May 14 and closed its first day at a $95 billion market cap. The wafer-scale AI chip maker shipped the year's biggest tech IPO.Fri, 15 May 2026 09:30:00 GMThardwarecerebrasai-chipsipocustom-siliconai-infrastructurenasdaqopenaiamazonhiro-tanakaAnthropic doubled Claude Code's limits by renting 220,000 GPUs from xAIhttps://devtake.dev/article/claude-code-rate-limits-doubled-may-2026/https://devtake.dev/article/claude-code-rate-limits-doubled-may-2026/Anthropic doubled Claude Code's 5-hour limits, killed peak-hours throttling, and raised Opus API tiers. The capacity comes from xAI's Colossus 1, via a SpaceX deal.Thu, 07 May 2026 12:30:00 GMTaianthropicclaude-codeclaude-opusspacexxaiai-infrastructureai-modelscolossusdieter-morelliAMD's data center revenue jumped 57% to $5.8 billion in Q1. Its stock closed up 18%.https://devtake.dev/article/amd-q1-2026-ai-chip-record-rally/https://devtake.dev/article/amd-q1-2026-ai-chip-record-rally/AMD reported Q1 revenue of $10.25B, up 38% YoY. Data center hit $5.8B (+57%) and Q2 guidance midpoint of $11.2B beat consensus by roughly $700M.Thu, 07 May 2026 11:45:00 GMThardwareamdai-chipsai-infrastructuresemiconductordata-centerslisa-suearningsnvidiahiro-tanakaSamsung Q1 profit hit 57.2 trillion won. Memory chips for Nvidia drove 93% of it.https://devtake.dev/article/samsung-q1-2026-record-ai-chip-profit/https://devtake.dev/article/samsung-q1-2026-record-ai-chip-profit/Samsung Electronics posted record Q1 2026 results on April 30: 133.9 trillion won revenue and 57.2 trillion won operating profit. Semiconductors did 93% of the work.Thu, 30 Apr 2026 09:00:00 GMThardwaresamsunghbmai-chipsdramsemiconductornvidiamemoryai-infrastructurehiro-tanakaAMD Zen 7 'Florence' leak: 288 cores, a separate cache die, and a 25% IPC swinghttps://devtake.dev/article/amd-zen-7-florence-288-core-leak/https://devtake.dev/article/amd-zen-7-florence-288-core-leak/Leaked slides describe Zen 7 EPYC scaling to 288 cores per socket, with all L3 moved off the CCD onto a stacked cache die. Tape-out is October 2026, launch late 2028.Mon, 27 Apr 2026 20:00:00 GMThardwareamdzen-7epycsemiconductorai-chipsipcserver-cpushardwarehiro-tanakaArcee's Trinity-Large-Thinking is a 399B open MoE that costs 96% less than Opushttps://devtake.dev/article/arcee-trinity-large-thinking-reasoning/https://devtake.dev/article/arcee-trinity-large-thinking-reasoning/Arcee released Trinity-Large-Thinking on April 1: a 399B-param sparse MoE with 13B active, Apache 2.0 weights, $0.88 per million output tokens, and PinchBench just behind Opus 4.6.Mon, 27 Apr 2026 13:00:00 GMTopen-sourcearceetrinityllmai-modelsopen-weightsmoereasoningapache-2-0soren-vanekCerebras files for an IPO again, this time with $510M in revenue and a $10B OpenAI deal in its pockethttps://devtake.dev/article/cerebras-ipo-filing/https://devtake.dev/article/cerebras-ipo-filing/Cerebras filed an S-1 on April 17 listing as 'CBRS,' targeting roughly $23B at the prior private mark. The OpenAI inference deal is the line item that changed the story.Sat, 25 Apr 2026 08:30:00 GMThardwarecerebrasai-chipsiposemiconductornvidiaopenaicustom-siliconwafer-scalehiro-tanakaIntel's foundry finally has a whale: Musk says Tesla's Terafab will run on 14Ahttps://devtake.dev/article/intel-tesla-14a-foundry/https://devtake.dev/article/intel-tesla-14a-foundry/Elon Musk said Tesla plans to manufacture at Terafab in Austin on Intel's 14A process. It's the first major external customer Intel CEO Lip-Bu Tan needs to keep the foundry alive.Thu, 23 Apr 2026 10:30:00 GMThardwareintelteslaterafab14a-processfoundryelon-musklip-bu-tansemiconductorhiro-tanakaGoogle splits its AI chip: the TPU 8t trains, the TPU 8i serves, and Axion handles the resthttps://devtake.dev/article/google-tpu-ironwood-axion-launch/https://devtake.dev/article/google-tpu-ironwood-axion-launch/At Cloud Next 2026, Google unveiled TPU 8t and TPU 8i, an 8th-generation split that matches Nvidia on agentic inference. Ironwood hit GA the same day.Thu, 23 Apr 2026 09:00:00 GMThardwaregoogletputpu-8ironwoodaxionai-infrastructurenvidiacloud-nexthiro-tanakaGPUHammer grew up: three new Rowhammer attacks take full control of Nvidia machineshttps://devtake.dev/article/gpuhammer-rowhammer-nvidia-attack/https://devtake.dev/article/gpuhammer-rowhammer-nvidia-attack/IEEE S&P 2026 papers extend GPUHammer with GeForge, GDDRHammer, and GPUBreach. They flip GDDR6 bits to break out of the GPU and own the host.Tue, 21 Apr 2026 11:00:00 GMTsecurityrowhammernvidiagpu-securitygddr6gpuhammerhardware-securityieee-sp-2026luca-reinhardtNSA is running Anthropic's Mythos. The Pentagon says Anthropic is a supply-chain risk.https://devtake.dev/article/nsa-anthropic-mythos-blacklist/https://devtake.dev/article/nsa-anthropic-mythos-blacklist/Axios reports the NSA is using Anthropic's unreleased Mythos model even though the Defense Department has blacklisted Anthropic. One government, two positions.Mon, 20 Apr 2026 15:30:00 GMTaianthropicclaude-mythosnsapentagoncybersecurityproject-glasswingnational-securitydario-amodeidieter-morelliCloudflare open-sourced a lossless LLM compressor that shaves 22% off model weightshttps://devtake.dev/article/cloudflare-unweight-lossless-llm-compression/https://devtake.dev/article/cloudflare-unweight-lossless-llm-compression/Unweight is Cloudflare Research's new BF16 weight compressor. 22% smaller bundles, 13% smaller inference footprint, 30-40% throughput overhead, BSD license.Sun, 19 Apr 2026 12:00:00 GMTopen-sourcecloudflareunweightllmcompressionbf16huffmanh100open-sourcesoren-vanekCursor wants $50B for an AI editor that's burning cash on individualshttps://devtake.dev/article/cursor-50-billion-valuation-raise/https://devtake.dev/article/cursor-50-billion-valuation-raise/Cursor is in talks to raise $2B at a $50B valuation, nearly double its September mark. Revenue is up, but it's still losing money per indie seat.Sat, 18 Apr 2026 08:00:00 GMTaicursoranysphereai-codingfundingandreessen-horowitzclaude-codeopenai-codexdieter-morelliMeta just committed 1 gigawatt of custom AI chips with Broadcom through 2029https://devtake.dev/article/meta-broadcom-custom-ai-chip-deal/https://devtake.dev/article/meta-broadcom-custom-ai-chip-deal/Meta and Broadcom expanded their partnership to co-develop four generations of MTIA chips on a 2nm process. Here's what the deal includes and why Nvidia should care.Thu, 16 Apr 2026 08:30:00 GMThardwaremetabroadcomai-chipsmtiacustom-siliconnvidiainfrastructurehiro-tanaka