Weekly AI Roundup: Anthropic Partners with SpaceX Because Nothing Makes Sense Anymore, Commits $200B to Google Just to Keep the Lights On, and Pennsylvania Sues a Chatbot for Practicing Medicine
A week where Anthropic made deals with both Google and Elon Musk, a chatbot claimed to be a licensed psychiatrist in Pennsylvania, and Claude Code users got double the rate limits because SpaceX apparently has GPUs to spare. Normal stuff.
Anthropic Rents Elon’s GPUs, Doubles Your Rate Limits
Anthropic signed a deal with SpaceX to use all of the compute capacity at Colossus 1, xAI’s Memphis data center — over 300 megawatts of capacity, equivalent to more than 220,000 Nvidia GPUs coming online within the month. The immediate payoff: Claude Code’s five-hour rate limits are doubled across Pro, Max, Team, and Enterprise tiers, and peak-hour restrictions are gone for Pro and Max.
The optics are delicious. Elon Musk has called Anthropic “misanthropic and evil” and spent years antagonizing Dario Amodei. Now he says he’s “impressed” with the Anthropic team and Claude will “probably” be good. Money talks, apparently, even between people who publicly despise each other. Anthropic also expressed interest in partnering with SpaceX on “orbital AI compute capacity,” which is either the most ambitious infrastructure play in history or the most expensive way to avoid paying a cloud provider’s markup.
$200 Billion for Google, $900 Billion Valuation, and an IPO on Deck
In case SpaceX wasn’t enough, Anthropic also committed to spending $200 billion with Google Cloud over five years for multiple gigawatts of TPU compute. That’s on top of the up to $40 billion Google is investing in Anthropic directly. When your biggest cloud provider is also your biggest investor, the word “committed” does a lot of heavy lifting.
Meanwhile, Anthropic is weighing a funding round that would value it at $900 billion — eclipsing OpenAI’s $852 billion February valuation. Annualized revenue has crossed $30 billion. A June IPO is reportedly in play. The company also announced a new enterprise AI services company with Blackstone, Hellman & Friedman, and Goldman Sachs, because when you’re raising $50 billion you might as well build a consulting arm too.
Pennsylvania Sues a Chatbot for Practicing Medicine Without a License
Pennsylvania filed suit against Character.AI after a state investigator discovered a chatbot named “Emilie” that claimed to be a licensed psychiatrist, said she went to medical school at Imperial College London, and produced a fabricated Pennsylvania medical license number on request. The bot told the investigator they were her “patient” and offered to book a depression assessment.
The state is seeking a preliminary injunction to stop AI companion bots from posing as licensed professionals. This is the first time a state has sued an AI company specifically for unauthorized practice of medicine, and it probably won’t be the last. Character.AI has already been the subject of multiple lawsuits over teen safety — this one adds a new legal theory that could apply to any chatbot that claims professional credentials it doesn’t have.
Also Noted
The White House wants to vet AI models like drugs. National Economic Council Director Kevin Hassett said the administration is drafting an executive order to evaluate new AI models before release, comparing the process to FDA drug approval. Google DeepMind, Microsoft, and xAI have already signed agreements to let a government agency test their models before launch.
OpenAI’s AI phone is accelerating. Mass production is now targeted for first half 2027, using a custom MediaTek Dimensity 9600 chip. Analyst Ming-Chi Kuo projects up to 30 million units shipped by 2028. An always-on AI that remembers everything you do — what could possibly go wrong.
EPAM trains 20,000 employees on Claude. Anthropic and EPAM Systems announced a multi-year partnership to build a dedicated Claude practice with 10,000+ certified architects. Enterprise AI consulting is now officially its own industry.
Filed from the wire. Back next Friday.