⇤ BACK

Weekly AI Roundup: Anthropic's Revenue Is Growing Faster Than Any Company in History, Google I/O Drops Tomorrow with Gemini 4.0, and the US Government Now Vets Every Frontier Model Before Launch

· By · clzd.me

A week where Anthropic’s revenue curve started looking like a typo, Google teased everything it’s about to announce at I/O tomorrow, and the US government quietly locked down pre-deployment access to every major frontier model. Also, Grafana got their GitHub token stolen. Normal week.

Anthropic’s Revenue Growth Is Breaking the Math

The numbers are getting absurd. Anthropic’s ARR trajectory in 2026 alone: $9 billion at the start of the year, $14 billion in February, $19 billion in March, $30 billion in April, and now estimates north of $44 billion from SemiAnalysis. That’s doubling every six weeks — faster than Zoom during COVID, faster than Google in 2003.

For context, Salesforce took 20 years to hit $30 billion in annual revenue. Anthropic did it in under three years from a standing start. The company’s fundraising round — at least $30 billion at a $900 billion-plus valuation — is expected to close by end of May, co-led by Sequoia, Dragoneer, Greenoaks, and Altimeter. If it closes at that number, Anthropic officially leapfrogs OpenAI’s $852 billion March valuation. A trillion-dollar private company is no longer a thought experiment.

Google I/O Drops Tomorrow — Gemini 4.0 and AI Glasses on Deck

Google I/O 2026 kicks off Monday May 19 at 10 AM Pacific, and the rumor mill has been running at full capacity. The confirmed agenda includes “the latest Gemini model updates” and “agentic coding” — widely interpreted as a Gemini 4.0 reveal.

Beyond the model, expect Android XR glasses, a new laptop platform called Aluminium OS, Android 17 features built on “Gemini Intelligence,” and Gemini Spark — a persistent AI agent that supposedly handles complex multi-step tasks autonomously. Google is racing to put Gemini at the center of everything before Apple’s AI reboot lands later this year. Whether any of this works as well in practice as it does in a keynote demo remains the perennial Google I/O question.

Every Frontier Model Now Goes Through Government Review

The US Commerce Department’s CAISI quietly finalized pre-deployment evaluation agreements with all five major frontier labs — OpenAI, Anthropic, Google DeepMind, Microsoft, and xAI. Every major model now goes through government evaluation before public launch, up from just OpenAI and Anthropic who signed voluntarily in 2024.

National Economic Council Director Kevin Hassett compared the process to FDA drug approval, which is either reassuring or terrifying depending on how you feel about the FDA. The agreements were renegotiated under Commerce Secretary Howard Lutnick’s directives and the America’s AI Action Plan. The practical question — whether CAISI has the technical capacity to meaningfully evaluate models shipping at this pace — remains conveniently unanswered.

Also Noted

Grafana’s GitHub got popped. On May 17, Grafana disclosed that an unauthorized party obtained a token granting access to their GitHub environment and the ability to download their codebase. They say no customer data was accessed. The token was revoked and the investigation is ongoing.

Google I/O pre-show teasers. Google has been seeding previews all week — deeper Gemini integration across Android, ChromeOS, and Google Workspace. The “Googlebook” running Aluminium OS is reportedly real. The company is clearly trying to reframe itself as an AI-first platform company before Sundar takes the stage.

Yann LeCun’s new lab raised $1B in seed funding. Advanced Machine Intelligence Labs, founded by the Turing Award winner and former Meta chief AI scientist, closed the largest seed round in European history at a $3.5 billion valuation. Because the AI fundraising market apparently has no ceiling.

Filed from the wire. Back next Friday — with a full Google I/O postmortem.