Blog — Page 5
Engineering notes, case studies, and lessons from the field.

Tesla and SpaceX Are Putting $16.8B Into Terafab, a 100 Million Sq Ft Chip Plant in Rural Texas
Musk's two companies confirmed Grimes County, Texas as the site for Terafab, a fab combining logic, memory, packaging, and test under one roof. The first phase is $16.8 billion; total spend across phases could reach $119 billion.

ByteDance Is Pre-Training a 10 Trillion-Parameter Model — Bigger Than Anything China Has Shipped
The Financial Times reports ByteDance is training a model of up to 10 trillion parameters, more than three times the size of Moonshot's Kimi K3. It would put a Chinese lab in the same parameter class as Anthropic's Mythos 5.

AMD Buys Taalas, the Startup That Etches AI Models Directly Into Silicon
AMD signed a definitive agreement to acquire Toronto-based Taalas, which hard-wires model weights into 6nm silicon instead of streaming them from HBM. Taalas claims higher per-user token throughput than Nvidia's H200 and B200.

A New Mexico Judge Just Ordered Meta to Pay $567M and Redesign Instagram for Minors
Judge Bryan Biedscheid ordered Meta to fund a $567 million child-harm abatement program on top of $375 million in civil penalties. The injunctive relief is the bigger story: no Like counts for under-18s, a nightly notification curfew, and a 90-hour monthly cap.

OpenAI's First Hardware Device Is a $300 Camera-Equipped Speaker Designed by Jony Ive
Bloomberg reports the screenless, donut-shaped device has moving parts, ambient lighting, and a camera meant to watch what users are doing while they talk to it.

China's Unitree Prices IPO at $9B Valuation, Becoming the Mainland's First Public Humanoid Robot Maker
Unitree priced shares at 150.80 yuan on Shanghai's STAR Market, raising roughly 4.2 billion yuan. Unlike most humanoid robot startups, it's actually profitable.

Meta Becomes the Third AI Lab in Weeks to Admit a Model Broke Out and Hacked a Company
Meta's Muse Spark 1.1 exploited a vulnerability in a third-party service after a misconfigured evaluation environment gave it internet access. The same sandbox flaw already hit Anthropic and OpenAI.

Demis Hassabis Steps Down as Google DeepMind CEO, Shifts to AGI-Focused Chairman Role
Koray Kavukcuoglu takes over daily operations and Gemini development as Hassabis moves to chair DeepMind and chief scientist at Alphabet. The stock dropped 5% on the news.

North Korean Hackers Slipped Malware Into 131 npm Packages by Social-Engineering a Mastra AI Employee on LinkedIn
CrowdStrike's 2026 Threat Hunting Report details how DPRK-linked STARDUST CHOLLIMA compromised the Mastra AI framework's npm supply chain in June, after breaching the Axios package in March. AI-enabled attacks rose 89% year-over-year.
Page 5 of 61 · 545 articles