Anthropic Just Lapped OpenAI in the Valuation Game — And Sam Altman Better Stop Pretending Otherwise

Anthropic raised $65 billion at a $965 billion post-money valuation this week. That's a bigger number than OpenAI's last round. I'll wait while you read that again.

Let me be crystal clear: the "scrappy safety startup" narrative is dead. Anthropic is no longer playing catch-up. Anthropic crossed $47 billion in annualized run-rate revenue this month — the month they dropped Claude Opus 4.8 — and they've got Amazon, Google Cloud, and Microsoft Azure all fighting to be their best friend simultaneously. 1

That's not a challenger. That's a champion.

Here's the hot take nobody in Silicon Valley wants to say out loud: OpenAI won the first quarter of this AI season, and now they're trying to live off that highlight reel while Anthropic is actually winning the current game.

Think about it. What's the last concrete, undeniable W that OpenAI scored? A valuation round? Anthropic just beat them there too. GPT performance? Claude Opus 4.8 just shipped with stronger coding and agentic task scores. Enterprise adoption? Anthropic just locked in "global enterprises across industries" while expanding into Milan, Seoul, and beyond.

OpenAI still has the name recognition. They still have ChatGPT as a consumer product. Those matter. But in the enterprise league — where the real contracts and the real money live — Anthropic is currently the better team on the court.

The comparison is brutal: OpenAI's last fundraise valued them at ~$730B. Anthropic is now at $965B. That's not a rounding error. That's a $235 billion gap, and it flipped in one funding cycle. 1

For three years, the conventional wisdom was: Anthropic is the responsible team, which means they'll always be a step behind. Safety frameworks cost you speed. Caution costs you market share.

Tell that to Sequoia Capital, who just co-led this round. Tell that to Altimeter, Dragoneer, Greenoaks, and the 20+ other institutional investors who all decided Anthropic is where the money goes. 1

Alfred Lin at Sequoia said it directly: Claude is learning "how businesses actually operate — the context, the processes, the judgment." That's the pitch for a dominant franchise, not a niche player. You don't raise $65 billion being cautious for the sake of optics. You raise $65 billion because you've actually built something enterprise buyers trust more than the alternative.

Here's what Anthropic figured out that OpenAI is still struggling with: enterprise buyers don't just want capability. They want reliability. They want predictability. They want a model that won't embarrass them in front of their board. Claude's positioning — "more honest when it makes mistakes," "consistent enough to handle long-running work" — that's not a feature list. That's a value proposition that prints contracts.

While we were watching the Anthropic vs. OpenAI main event, a new player crashed onto the scene.

Tencent's Hy3 preview — a model barely anyone outside of China had heard of — jumped to the #2 position on OpenRouter's token usage rankings for May 25, with volume 50% higher than Claude. It's coming out of nowhere, served exclusively through Singapore-based SiliconFlow, with 98% of its usage being input tokens (data processing workloads) and top apps accounting for less than 1% of activity. 2

Nobody knows who the whale customer is. The benchmarks don't justify the usage at all — Hy3's coding scores don't come close to Claude Opus 4.7 or GPT 5.5. Something is running massive data pipelines through this model quietly, at scale, and doesn't care about benchmarks.

This is the DeepSeek playbook redux. Remember when everyone laughed at the "low budget Chinese model"? And then R1 dropped and the whole American AI industry had a collective meltdown? Tencent — with the resources of a $450B company — just quietly slipped something into the supply chain while everyone's eyes were on the Anthropic funding news.

I'm not saying Hy3 beats Claude or GPT in quality. I'm saying that someone who processes a lot of tokens decided it was good enough at a price that worked, and that's exactly how DeepSeek got traction before anyone was paying attention. 2

Google has more money than every other team in this league combined. They have the best infrastructure. They have five gigawatts of TPU capacity committed to Anthropic — they're literally funding the team that's beating them on valuation. That's the most Google thing I've ever heard.

Sundar Pichai just gave a full interview to The Verge about "the future of search" while DuckDuckGo's installs went up 33% after Google I/O — specifically because users are recoiling from Google's AI-infested search experience. 3

Google's AI Mode response: show users "preferred sources." That's the product answer to a trust crisis. That's a team that knows they're losing the crowd and is desperately trying to remember how to play fundamental basketball.

I'm going on record right now. Anthropic's IPO — when it comes — prices above $1 trillion. Not because $965B is close to $1T and rounding is easy. Because by the time they actually file, run-rate revenue won't be $47B. It'll be higher. The customer base will be bigger. The moat — three years of enterprise trust and safety credibility — will be deeper.

And here's the twist that'll break the internet: if Anthropic IPOs above OpenAI's eventual IPO valuation, the whole "Sam Altman won the AI race" narrative collapses in public, in real time, with stock prices as the scoreboard. That's not a technicality — that's a dynasty change.

The safety team didn't just keep up. The safety team is winning.

#AILeague