metir Blog

Nevada regulators unanimously approved commercial robotaxi permits for Tesla, Uber and Waymo covering up to 7,000 vehicles in Clark County. Here is what the permits actually authorize, and why three very different autonomy bets now share one regulatory framework.

Israeli firm Dream documented a four-day campaign against Taiwan that ran on open-source AI agents with minimal human steering. What was found, what near-autonomous actually means, and why it matters.

CBRE's 2026 Scoring Tech Talent report finds New York has more tech workers than the San Francisco Bay Area for the first time in 13 years. Here is what the headcount flip does and does not say about AI.

Varonis disclosed CoSnitch, a critical Microsoft Copilot vulnerability letting one click exfiltrate account data, with Copilot itself helping map the flaw.

AI now writes half of Linear's issues and coding agents tripled weekly pull requests, yet product development time still went up. Here is what the 2026 data says is actually happening.

Nvidia is paying $6 billion to license Poolside's AI model-building software and hiring 109 staff, plus a $1 billion equity stake, without a traditional acquisition.

Broadcom is reportedly in talks for up to $100 billion in debt to fund custom AI chips for Anthropic and others. How the tranches work, and what it means.

Google DeepMind says Gemma has crossed 1 billion cumulative downloads and 100,000-plus community variants. What the milestone does, and doesn't, tell us about open weights.

Nvidia is in early discussions with Korean AI inference chip designer Rebellions about a technical partnership, an investment, or an acquisition. What is confirmed, and what is not.

Anthropic now embeds invisible watermarks in Claude's text worldwide under EU AI Act rules. How the mechanism works, what it proves, and what it does not.

Nvidia is in talks to back a funding round valuing AI data supplier Mercor at $20 billion, roughly double its valuation from under a year earlier.

Cerebras unveiled the CS-4, claiming up to 30x faster AI inference than GPU-based solutions. How wafer-scale chips work, and what the numbers do and don't prove.

OpenAI began rolling out ChatGPT for Teens on August 18, 2026. How account age, self-reported age, and age prediction route 13-to-17 accounts into a more restrictive experience, and the privacy tradeoffs involved.

Samsung Foundry is raising prices up to 15% on its 4nm, 5nm and 8nm nodes as AI chip demand overflows from TSMC. What it means for downstream AI costs.

SpaceX closed its $60 billion acquisition of AI coding tool Cursor on August 15, 2026. What the deal means for Elon Musk's AI-coding stack, and the Cognition rumor that followed it.

AI chip startup Etched raised $700 million led by Jane Street at a $21 billion valuation, doubling in a month. A neutral look at the transformer-specialized Sohu chip, the customer-as-investor signal, and the specialization trade-off.

UK chip startup Fractile is reportedly raising at a $6.5 billion valuation after an Anthropic supply deal, six times its May mark. A neutral look at in-memory compute, the DRAM shortage driving interest, and why the chips are a 2027 story.

Marvell granted Google a warrant worth up to $12.2 billion tied to a custom AI chip deal. A neutral breakdown of how the warrant vests against chip purchases, why Marvell stock jumped, and what it signals about hyperscaler silicon.

Anthropic says Claude designed protein binders that bound 14 of 15 lab targets, above typical success rates. A neutral analysis of the experiment, why a general model doing specialized science matters, and the limits of a lab-bench result.

Google is offering eligible students a free year of Google AI Pro in the US and AI Plus across 140-plus markets. A neutral look at what is included, how to qualify, and why free student AI tiers are really about long-term habits.

Anthropic is reported to be in talks to buy Israeli startup Decart for about $6 billion, with later reports near $7 billion. Why AI efficiency is the draw.

Higgsfield raised $400 million at a $5.4 billion valuation, roughly quadrupling its prior mark. A neutral look at what a $700M annualized run-rate says about AI video moving from experiment to production.

Groq raised $350 million at a $3.5 billion valuation, below its 2025 mark, as it pivots from making its own inference chips to running a neocloud on Nvidia GPUs. A neutral analysis.

A Wall Street Journal analysis found roughly $3 trillion of off-balance-sheet AI commitments across nine tech companies. What that means in plain accounting terms, and why it matters.

Google DeepMind researchers argue that scaling text-based LLMs alone will not be enough. A neutral survey of world models, VLA, and state-space alternatives in 2026.