Sunday, April 19, 2026 — Three big stories define this weekend's AI landscape: Anthropic quietly surpassed OpenAI in revenue for the first time, Meta's overhauled AI strategy produced its first real model, and Stanford's annual AI Index confirmed what many suspected — China has nearly closed the gap with the United States.
Anthropic Overtakes OpenAI in Revenue
In a landmark reversal, Anthropic has eclipsed OpenAI in annualized revenue, hitting a $30 billion run rate in March 2026 — up roughly 1,400% year-over-year from $9 billion at the close of 2025. OpenAI, which closed a $120 billion funding round at an $850 billion valuation last month, sits at approximately $25 billion in annualized revenue. The gap isn't enormous, but the symbolism is hard to ignore: the younger, safety-focused spinoff now out-earns the company that started the generative-AI gold rush.
Enterprise adoption is the engine. Roughly 80% of Anthropic's revenue comes from more than 300,000 business customers, a sign that Claude has become the default AI layer for serious back-office and developer workflows. Both companies are now eyeing 2026 IPOs — Anthropic at a possible $380 billion valuation in October, OpenAI targeting a Q4 NASDAQ listing near $850 billion — making the revenue race very much a public-market story too. Sources: Medium, OpenClawAI, CMC Markets.
Meta's Muse Spark: A Ground-Up Overhaul Under Alexandr Wang
After spending an estimated $14 billion to bring in Scale AI founder Alexandr Wang and stand up Meta Superintelligence Labs, Meta has shipped Muse Spark — the first model out of that new organization. Announced April 8, Muse Spark is a natively multimodal reasoning model that supports visual chain-of-thought, tool use, and multi-agent orchestration. It scored 52 on the Intelligence Index and is intentionally kept small and fast for consumer deployment.
The practical reach is staggering: Muse Spark is rolling out across WhatsApp, Instagram, Facebook, Messenger, and Meta's AI glasses. Novel features include a "Contemplating" mode that spins up multiple reasoning agents in parallel, strong health-question handling developed with a team of physicians, and visual coding that lets users build dashboards or mini-games straight from a text prompt. It's free to use, though rate limits may apply. The model accepts voice, text, and image inputs but currently outputs text only. Sources: TechCrunch, Meta, Axios.
Stanford AI Index: China Has Nearly Erased the U.S. Lead
Stanford's 2026 AI Index, released this week, contains a number that should focus minds in Washington: the performance gap between the top U.S. and top Chinese models has shrunk to just 39 Arena points, with Anthropic's Claude Opus 4.6 leading China's Dola-Seed 2.0 by a mere 2.7%. A year ago that margin was far wider. The U.S. still produces more top-tier models — 50 versus China's 30 — but China now generates more AI publication citations, and the talent pipeline flowing into America has slowed. Meanwhile, AI data centers globally now draw 29.6 gigawatts of power (enough for all of New York State at peak demand), and the environmental costs are mounting: running GPT-4o alone may consume more water annually than 1.2 million people drink. Sources: IEEE Spectrum, Fortune.
What to Watch
The next 30 days are loaded. Keep an eye on whether Anthropic files IPO paperwork — any S-1 filing would be the most-scrutinized document in tech this year. Watch how Meta's billion-user distribution gives Muse Spark adoption data that no benchmark can replicate. And track the U.S.–China model race closely: with the gap now measured in single-digit percentage points, the next major model drop from either side could flip the leaderboard entirely. The era of a comfortable American lead in frontier AI may already be over.