Welcome to Jumble, your go-to source for AI news updates. This week, Anthropic signed a deal with SpaceX to take over an entire data center. Meanwhile, the White House is drafting an executive order to vet AI models like FDA drugs. Let’s dive in ⬇️
In today’s newsletter:
🔌 Anthropic and SpaceX team up
🏛️ FDA-style AI approval process
🍎 Apple opens iOS 27 to major models
📱 OpenAI fast-tracks first AI phone
🧪 Weekly Challenge: Let AI find money you're wasting
🛰️ Anthropic Just Rented Elon's Entire Supercomputer
Anthropic signed a deal with SpaceX to take all of the compute at the Colossus 1 data center in Memphis, picking up 220,000 NVIDIA GPUs and 300 megawatts of capacity within a month. The agreement lands weeks before SpaceX's planned IPO and joins a stack of compute deals worth hundreds of billions.
⚡ Claude Code Limits Just Doubled
Five-hour rate limits for Claude Code doubled across Pro, Max, Team, and Enterprise plans, effective immediately. Peak-hour restrictions also got dropped for Pro and Max users, easing the throttling that frustrated developers all spring.
🤝 The Musk Heel Turn
Musk previously called Anthropic "misanthropic and evil," but posted that he's "impressed" after meeting the team. He also reserved the right to "reclaim the compute" if Claude misbehaves, which is either a joke or the most ominous SLA clause in tech.
🚀 Data Centers In Space
Both companies expressed interest in orbital AI compute at gigawatt scale, meaning literal data centers in space. Musk separately announced xAI is being dissolved into SpaceX under the new name SpaceXAI, consolidating his entire AI footprint under one roof.
💊 The White House Wants To Regulate AI Like A Drug
NEC Director Kevin Hassett said Wednesday the White House is drafting an executive order requiring AI models to pass FDA-style safety vetting before release. The trigger is Anthropic's Mythos, a model that finds decades-old vulnerabilities and is locked to just 12 partners.
Should AI models go through an FDA-style approval process?
🧪 The Approval Pipeline
Models would have to be "tested left and right" before release, Hassett said. The Center for AI Standards and Innovation would likely run the reviews, and Hassett said the rules would "really quite likely" apply to all frontier labs, not just Anthropic.
🥊 Internal Disagreement
Hours after Hassett's interview, Chief of Staff Susie Wiles posted that Trump is "not in the business of picking winners and losers." Translation: the order is real, but the strength of it is still being fought over inside the West Wing. Sources expect a signed order within two weeks.
Weekly Scoop 🍦
🎯 Weekly Challenge: Let AI Find Money You're Wasting
Challenge: Most people leak money through forgotten subscriptions and overlapping services. This week, let AI do the audit you've been putting off.
Here’s what to do:
📄 Step 1: Pull a statement Grab last month's credit card bill, phone bill, or list of subscriptions. Strip any sensitive info before uploading.
🔍 Step 2: Ask AI to audit it Upload to ChatGPT, Claude, or Gemini and ask: "Find redundant services, forgotten subscriptions, and anything I could cut without missing it."
✂️ Step 3: Cancel one thing Pick the most embarrassing find and kill it today. Reply and tell us how much you saved.
This will take you 5-10 minutes, but could save you a lot of money. Good luck on cutting down on those pesky subscriptions.
Was this the week Anthropic became too big to fail, or the week Washington finally caught up to it? See you next time! 🚀
Stay informed, stay curious, and stay ahead with Jumble!
Zoe from Jumble
