A lot happened in AI this week, but three things stood out: Anthropic quietly filed for an IPO, Congress dropped a massive bill that could bulldoze every state AI law on the books, and OpenAI rolled out a memory upgrade that's actually worth paying attention to.
Let's get into it.
Anthropic Just Filed an S-1 and the Numbers Are Insane
Anthropic confidentially filed a draft S-1 with the SEC on June 1st. The headline: their revenue run rate hit roughly $47 billion in May 2026. A year ago that number was around $10 billion.
That's not a typo. Nearly 5x growth in twelve months.
If you've been sleeping on Anthropic because OpenAI gets more press, wake up. Claude is clearly winning real enterprise business at a scale most people aren't tracking. The confidential filing means we don't have full details yet, but this is the kind of growth that makes an IPO not just possible but inevitable.
My take: this is the most important AI business story of 2026 so far. A company built on "responsible AI" is now one of the fastest-growing software businesses in history. Whether that's ironic or inspiring probably depends on where you sit.
Congress Dropped a 269-Page AI Bill That Could Override State Laws
Congress released a 269-page federal AI bill this week. The big deal here isn't the length. It's the preemption clause.
If this passes as written, it could override every state-level AI law in the country. That means California's AI regulations, Texas's rules, all of it would take a back seat to whatever the federal government decides.
On one hand, a patchwork of 50 different state AI laws is a nightmare for any company trying to ship a product nationally. On the other hand, federal preemption means one lobbying target instead of fifty, which is great news for the big tech companies that can afford to shape federal policy.
We don't know yet how this bill will evolve, and bills this size almost always change dramatically before anything gets signed. But the intent is clear: Washington wants to own AI regulation, and they're moving fast.
Worth watching closely over the next few months.
OpenAI's Memory Upgrade Is the Quiet Win Nobody's Talking About
OpenAI started rolling out what they're calling the Dreaming V3 memory architecture to ChatGPT Plus and Pro users this week. It's being described as the biggest memory upgrade since the original ChatGPT launch.
Memory in AI assistants sounds boring until you actually use a system with good memory and realize how much time you waste re-explaining context every single session. The difference between a tool that remembers you and one that doesn't is the difference between a capable assistant and a very smart stranger you have to brief from scratch every morning.
I haven't tested Dreaming V3 yet, but if the memory improvements are as significant as OpenAI claims, this could actually shift how people use ChatGPT day-to-day. That matters more than most benchmark announcements.
Keep an eye on whether Anthropic's Claude follows with something similar. Memory is where the real stickiness lives in AI products right now.