Browsing Tag
AI Chips
20 posts
AI accelerators, advanced processors, semiconductor supply chains, and chip policy tied to artificial intelligence.
Google Adds Secure TPU 8tx and Dual-Chip 8i Max, but Leaves Pricing Open
Google’s TPU 8tx protects data in use, while 8i Max doubles compute and memory bandwidth. Pricing, regions and availability remain undisclosed.
Meta’s $97.5B AI Financing Pulls Nvidia and AMD Into the Cap Table
Meta priced $37.5 billion of immediate financing and plans a possible $60 billion Class C program as Nvidia and AMD invest in their own major AI customer.
TorchNeuron Runs PyTorch on Trainium, but Beta Gaps Matter
Amazon's TorchNeuron makes Trainium a native PyTorch device. Here are the public beta's compatibility gaps, performance claims and migration checks.
Microsoft Maia 300: Specs, 2027 Rollout, and What the Claims Mean
Microsoft’s Maia 300 packs eight HBM4 stacks into a 950W accelerator and 96 chips into a 167kW rack. Here’s what its 2027 rollout means.
AMD’s $8.2B World Labs Deal Is a Bet on AI Beyond Language
AMD is buying Fei-Fei Li’s World Labs for $8.2 billion. The strategic prize is a model lab that can shape future chips, robotics simulation, and the physical AI stack.
Alibaba’s Zhenwu V900 Turns China’s AI Push Into a Full-Stack Scale Test
Alibaba’s Zhenwu V900 pairs 216 GB of memory with a Qwen 4 roadmap and a 20 GW cloud target. Here is what is shipping, what remains a claim, and what buyers should verify.
Dell’s $95B AI Server Backlog Turns Compute Demand Into a Delivery Test
Dell reported $60.9 billion in AI server orders, $16.4 billion in AI-optimized server revenue, and a $95 billion backlog in its fiscal second quarter. The numbers show how AI infrastructure demand is moving from GPU hype into a harder test of supply chains, racks, storage, networking, and delivery schedules.
OpenAI’s Jalapeño Benchmarks Turn Inference Power Into the AI Chip Fight
OpenAI published first benchmark results for Jalapeño, its Broadcom-built inference chip, claiming higher performance per watt and lower latency than leading commercial AI systems. The useful question is not whether it replaces Nvidia immediately, but whether custom inference silicon can turn power limits into a product advantage for ChatGPT, Codex, and agentic AI workloads.
NVIDIA’s Ohio Backstop Turns OpenAI’s Data Center Demand Into Infrastructure Finance
NVIDIA is investing $1.5 billion in SB Energy and backing land, power, and shell capacity for OpenAI's planned PORTS-Pike AI data center in Ohio. The deal shows how frontier AI demand is pushing chipmakers deeper into power, real estate, financing, and local infrastructure.
Qualcomm’s Chip Price Hike Could Make Android Upgrades More Expensive
Qualcomm has reportedly told customers it will raise chip prices by a double-digit percentage for shipments after September 1. The move could push Android phone makers, smart-glasses vendors, and Windows-on-Arm PC builders toward higher prices, tighter specs, or delayed launches.