The smart money in San Francisco just switched horses mid-race.
The Summary
- At HumanX 2026, VCs made it clear: Anthropic is now Silicon Valley's favorite AI lab, a sharp reversal from last year's OpenAI consensus
- Anthropic's run-rate revenue hit $30 billion (up from $9 billion in late 2025) while trading at a $380 billion valuation versus OpenAI's $852 billion
- The shift isn't just vibes: Claude Code became the product that made enterprise buyers believers
The Signal
Twelve months ago at HumanX in Vegas, you couldn't find a VC who wasn't bullish on OpenAI. This week at Moscone Center, you couldn't find one who still is. That's not typical Silicon Valley fickleness. That's a market repricing what "winning" looks like in the foundation model race.
The numbers tell part of the story. Anthropic's revenue tripled in four months to hit a $30 billion annual run rate. OpenAI still leads in absolute revenue, but Anthropic is closing the gap while trading at less than half the valuation multiple. For VCs doing the math on which IPO they want exposure to, that's a no-brainer arbitrage play.
But the real story is product. Claude Code didn't just ship. It became the thing developers actually use. Not the thing they demo. Not the thing they tweet about. The thing that writes production code while they're in standup meetings.
"The focus on enterprise, frontier capabilities, and coding while deliberately avoiding consumer use cases were great decisions."
Anthropic made a bet most AI labs wouldn't: they said no to the dopamine hits of viral consumer products. No ChatGPT moment. No race to put a chatbot in every pocket. They built for the people who sign $10 million checks, not the people who sign up for free trials. Turns out enterprises don't want cool demos. They want models that actually work inside their security perimeter and don't hallucinate when you ask them to touch prod.
Key differences in approach:
- OpenAI chased consumer scale and brand recognition
- Anthropic optimized for enterprise trust and deployment velocity
- OpenAI valued growth at all costs; Anthropic valued margin and focus
The conference timing matters. Both companies are reportedly preparing to go public. That means real scrutiny of real business models, not just "AI is the future" handwaving. Anthropic's enterprise focus gives them a story public market investors understand: recurring revenue from companies that don't churn, fat margins on API calls, a customer list that reads like the Fortune 500.
OpenAI's story is messier. Massive consumer usage that hasn't translated to sustainable unit economics. A valuation that assumes permanent market dominance. And whatever acquisition drama the article mentions but cuts off mid-sentence.
The Implication
If you're building AI tooling or agent infrastructure, watch where the enterprise dollars flow next. Anthropic's model proves you can win the foundation model race without winning Twitter. The companies building on top should take note: the real money isn't in viral moments. It's in replacing $500/hour consultants with $2/hour API calls that CFOs can actually model.
For OpenAI, this isn't a death sentence. Models advance too fast for any lead to hold. But they're learning what Microsoft learned in the mobile era: being first doesn't matter if you pick the wrong customer. The next six months will show whether they can pivot to enterprise credibility or whether they've already ceded that ground for good.