Ai Hardware

🚨 OpenAI just launched its first hardware product, but it's probably not what you were expecting. Instead of an AI gadget for everyone, it released the 𝘊𝘰𝘥𝘦𝘹 𝘔𝘪𝘤𝘳𝘰, a $230 desktop controller built for develop…

"Show me where you last saw my XR glasses." Milliseconds later, your robot responds. No cloud. No server. Wi-Fi off. I like this one. Our team built it to prove a point about Qdrant.to/Edge: the full retrieval stack can…

🤖 Quand le RAISE Summit transforme Paris en capitale de l'IA... 👀 Cette semaine, se déroulait le Raise Summit, dernière grande conférence tech avant la pause estivale... et quelle conférence ! Le line-up était une fo…

AMD GPU users: the longer your context, the more this attention kernel beats PyTorch SDPA - one line to enable in transformers. I ported the Composable Kernel (CK) FlashAttention from ROCm/aiter to the Hugging Face Kerne…

I’m speaking on the Compute track at the AI Infra Summit, taking place Sep 15-17 at the Santa Clara Convention Center! Register for your pass here: https://lnkd.in/ejrbY9Md The AI Infra Summit is the ultimate stage fo…

LET's go 🏆 (*) Yesterday, Techleap organised the Leading Entrepreneurs in Tech (LET) award which celebrates the entrepreneurs building big and ambitious things. A great event, with change makers, but also a great demon…

‼️THIS IS A COMPLETE MASSACRE IN KOREAN CHIP STOCKS: SK Hynix dropped -15.4% on Monday, posting its largest daily decline EVER. This even exceeded the declines seen during the Great Financial Crisis. On Tuesday, the s…

Une puce. Des tokens moins chers.  Et un impact direct sur la façon dont on fait tourner l'IA en production. OpenAI et Broadcom viennent de dévoiler Jalapeno - une puce d'inférence conçue spécifiquement pour les LLM. P…

Most people will read SpaceX’s IPO and Cursor deal wrong. They still see SpaceX as Elon’s Mars rocket company, not an AI infrastructure company. That made sense for years, given SpaceX started with rockets, Mars and imp…

We've introduced a suite of new chipmaking systems for building the advanced 3D chip architectures that power next-generation AI. 🔹Enhanced Centura™ Prime™ Epi: Bringing logic-class epitaxy to DRAM, boosting speed and …