Blog — Page 13
Engineering notes, case studies, and lessons from the field.
DeepSeek Is Building Its Own Inference Chip to Escape Both Nvidia and Huawei
Reuters reports China's DeepSeek has spent a year exploring custom AI accelerators and is now hiring chip designers and talking to foundries. The goal: silicon independence from every supplier it currently depends on.
Gemini 3.5 Pro Lands July 17 With a 2-Million-Token Context Window
A leaked launch plan puts Google DeepMind's Gemini 3.5 Pro at general availability on July 17, with double the context of any frontier rival. Deep Think reasoning will sit behind the $250/month Ultra tier.
Meta Charges for AI for the First Time: Muse Spark 1.1 API Undercuts OpenAI and Anthropic by 4x
Muse Spark 1.1 is Meta's first paid API model — $1.25 input and $4.25 output per million tokens, a 1M-token context window, and benchmark wins over Claude Opus 4.8 and GPT-5.5 on MCP Atlas and Humanity's Last Exam. The AI price war just got a third front.
Apple Sues OpenAI, Calling Its Hardware Business 'Rotten to Its Core' With Stolen Trade Secrets
Apple filed suit in the Northern District of California, accusing OpenAI of systematic trade secret theft — from interview 'show and tell' sessions with Apple parts to an engineer who allegedly kept a work laptop and siphoned confidential files. OpenAI denies it.
US Drops License Requirement for AI Chip Sales to the UAE
The Commerce Department reclassified the UAE to Country Group A:5, letting approved entities like G42 buy advanced AI chips from Nvidia, AMD, and Cerebras license-free. The move cements the US-UAE AI framework and rewards Abu Dhabi's trillion-dollar investment pledge.
Mistral Ships Robostral Navigate: An 8B Model That Steers Robots With One Cheap Camera
Mistral's first robotics model navigates unfamiliar buildings using a single RGB camera and a plain-language instruction — no lidar, no depth sensor, no pre-built map. It beats multi-sensor systems on the R2R-CE benchmark despite using a fraction of the hardware.
TypeScript 7.0 Ships: The Native Go Compiler Makes Full Builds 8-12x Faster
Microsoft released TypeScript 7.0, the long-awaited native port of the compiler written in Go. VSCode's codebase now type-checks in 10.6 seconds instead of 125.7, and strict mode is finally the default.
Apple Commits Over $30 Billion to Broadcom for US-Made Chips Through 2031
Apple signed its largest American Manufacturing Program deal yet: a multiyear agreement with Broadcom covering more than 15 billion US-made chips and a $1.5 billion expansion of Broadcom's Fort Collins, Colorado fab.
Meta Puts Its Iris AI Chip Into Production in September to Double Compute to 14 Gigawatts
An internal memo reveals Meta's custom MTIA accelerator, designed with Broadcom and fabbed by TSMC, enters production in September. The chip cleared bug testing in six weeks and anchors a plan to double datacenter compute from 7 to 14 gigawatts.
Page 13 of 61 · 549 articles