Five stories worth knowing about from the AI industry this week — a revenue milestone at the top of the frontier-lab race, a quiet shift in who actually processes developer traffic, and a regulatory summit trying to keep pace with all of it.
 |
| Photo by Mizuno K from Pexels |
1. Anthropic Overtakes OpenAI on Revenue
Anthropic reported a $47 billion annualized revenue run rate as of May 2026, ahead of OpenAI's most recently disclosed $25–33 billion — a lead of roughly $14–22 billion on a strict revenue basis. OpenAI still leads on total users (ChatGPT's roughly 1.1 billion monthly actives dwarf Claude's smaller, more commercially focused base) and on total capital raised. The split says something real about strategy: Anthropic is optimizing for revenue per user and enterprise commercial focus, while OpenAI is optimizing for reach.
2. Chinese Open-Weight Models Now Handle Nearly Half of OpenRouter Traffic
Chinese AI providers now account for an estimated 45% of all traffic on OpenRouter, up from under 2% a year earlier. Xiaomi's MiMo-V2-Pro alone processes 4.21 trillion tokens weekly, with a 21.1% platform share — more than double OpenAI's 7.5% — driven by strong coding benchmarks, a 1-million-token context window, and pricing several times lower than US frontier models. Notably, the model built developer trust anonymously before Xiaomi publicly attributed it, echoing a similar pattern with Meituan's LongCat-2.0.
3. Grok 4.5 Enters Private Beta at SpaceX and Tesla
Elon Musk announced that Grok 4.5 has entered internal private beta, built on a 1.5 trillion-parameter foundation model — roughly triple the size of the model currently handling production Grok traffic on X — and represents a 50% scale jump from Grok 4.4 in about a month. Musk said early evaluations show performance approaching, and in some cases exceeding, Claude Opus. The model reportedly includes supplemental training on Cursor data.
4. EU Mandates AI Driver-Distraction Detection in All New Cars
As of July 7, every newly registered car in the European Union must include a driver-distraction detection system that analyzes gaze and head movement to flag inattention, without recording or transmitting footage. It's a small mandate with a big signal: AI safety features are moving from optional add-on to regulatory baseline for an entire vehicle category.
5. UN Convenes Global AI Governance Dialogue in Geneva
The UN's Independent International Scientific Panel on Artificial Intelligence — 40 experts spanning every world region — published its first report on July 1, feeding into a Global Dialogue on AI Governance held in Geneva. Panel members, including Yoshua Bengio, warned that current science can't guarantee that advancing AI capabilities won't cause serious harm, either autonomously or through misuse, and flagged a widening "AI divide" between the US/China frontier and countries still building basic digital infrastructure.
What to Watch This Week
Voluntary AI model release standards from the White House are expected around August 1, following advanced talks between the administration and major AI labs — worth watching for how (or whether) the framework applies evenly across frontier developers.
The CODEW | AI Watch runs every Monday.
Erwin Castro
Founder & Editor • The CODEW
Erwin Castro is the founder and editor of The CODEW, covering technology mergers and acquisitions, startup exits, artificial intelligence, enterprise software, and Build vs Buy strategy. With more than a decade of journalism experience, he has contributed to Sportskeeda, IBTimes, University Herald, US Blasting News, and Seeking Alpha. His work focuses on explaining the business strategy behind technology deals and their impact on the global technology industry.
Reviewed by Erwin Castro
on
Monday, July 13, 2026
Rating:
No comments: