Image created with OpenAI GPT-Image-1. Image prompt: mid‑1990s web‑browser screenshot, CRT glow, 256‑color dithering — Animated e‑mail mailbox GIF bouncing in corner — microchip sprite rotating caption “AI Chips” — crisp pixel edges, screen‑door scan‑lines, phosphor glow
Intel layoffs begin. Cuts 15k-20k jobs company-wide to become leaner after a $19B loss and shrinking AI chip share. oregonlive. com/silicon-forest/2025/07/intel-layoffs-begin-chipmaker-is-cutting-many-thousands-of-jobs.html https://x.com/rohanpaul_ai/status/1942479845318897753
Congrats to @NVIDIA, the first public $4T company! Today, compute is 100000x cheaper, and $NVDA 4000x more valuable than in the 1990s when we worked on unleashing the true potential of neural networks. Thanks to Jensen Huang (see image) for generously funding our research 🚀 https://x.com/SchmidhuberAI/status/1943671639620645140
Nvidia CEO Jensen Huang to visit China again as firm plans China-only AI chip launch in September < World < 기사본문 - The Korea Post https://www.koreapost.com/news/articleView.html?idxno=45220 GPU by hand ✍️ I drew this to show how a GPU speeds up an array operation of 8 elements in parallel over 4 threads in 2 clock cycles. Read more 👇 CPU • It has one core. • Its global memory has 120 locations (0-119). • To use the GPU, it needs to copy data from the global https://x.com/ProfTomYeh/status/1942718838904418509 🎉 Big news – Cloudflare’s new pay per crawl model is here! This permission-based approach for AI content scraping empowers creators to control and monetize their content, ensuring fair compensation and building a sustainable future for original work online. Learn more: https://x.com/Cloudflare/status/1940063055368298617 IBM rolls out new chips and servers, aims for simplified AI | Reuters https://www.reuters.com/business/ibm-rolls-out-new-chips-servers-aims-simplified-ai-2025-07-08/ crazy that in 2025 i can converse in 1000 tokens/sec on my single GPU machine with AI that’s world-class at math and programming but i still have to type. i can’t speak to it, at least not in low-enough latency to carry a conversation we don’t have this tech yet. why not?”” / X https://x.com/jxmnop/status/1941995444730540050 Nvidia challenger Groq expands with first European data center https://www.cnbc.com/2025/07/07/ai-chip-startup-groq-expands-with-first-european-data-center.html 📊 Deployment Metrics in LangGraph Platform! View your deployment’s CPU & memory usage, API request latency, pending/active run count, and much more all in one place. ➡ Docs: https://x.com/LangChainAI/status/1943013330005954644 It was wonderful to get to spend time with the legendary @LisaSu today. She’s such an inspirational leader, it is no wonder that @AMD is on fire! 🔥 https://x.com/clattner_llvm/status/1943364501417660466 The story of hybrid architectures is honestly fascinating! I’ve been diving deep into why Transformers became the default choice, and looking at new model architectures. It’s not because “”attention is all you need”” (though catchy!) It’s because they exploited GPU parallelism so https://x.com/realSharonZhou/status/1943326596804345863 we’re also building out our physical infrastructure team — welcome to many amazing new team members!”” / X https://x.com/gdb/status/1943043253009551608 Impact of PCIe 5.0 Bandwidth on GPU Content Creation Performance | Puget Systems https://www.pugetsystems.com/labs/articles/impact-of-pcie-5-0-bandwidth-on-gpu-content-creation-performance/ RT @rajan__vivek: This could be the next dominant paradigm once we figure out how to make EBTs tractable at scale. Starting with a nebulous…”” / X https://x.com/_akhaliq/status/1941920969590792701 The Big Beautiful Bill lets U.S. hyperscalers and AI labs fully expense GPUs and training upfront, likely providing tens of billions in subsidies for compute through Trump’s term. Surprised this isn’t getting more attention; journalists frame the bill as bad for AI somehow?”” (see comments for counterpoint) / X https://x.com/tamaybes/status/1941633298893242444 Custom silicon vendors were pitching NVLink fusion in January and February UALink 1.0 spec had no surprises for anyone because consortiums discuss everything for 6 months before anything released Nvidia is far more worried about Broadcom SUE not UALink. Expert calls are just”” / X https://x.com/dylan522p/status/1942453912885186788 OK, I run the fp8 mamf-finder on B200 and indeed as others suggested fp8 efficiency is improving over NVIDIA generations: H100: 70.9% H200: 73.4% B200: 76.3% h100 and h200 should be the same compute, higher cuda version for h200 run is probably it https://x.com/StasBekman/status/1942972268851888606 The three biggest hps for stable training in everything are lr, bs, and beta2. We’ve built up good intuitions on how to tune them over time, but this lays it all out analytically and convincingly. this is definitely my new handbook for training big models on small gpus.”” / X https://x.com/sainingxie/status/1943453528099258529
