Last Updated on June 1, 2026 by Taya Ziv
In 2002, eBay bought PayPal for $1.5 billion. Over the next decade, the employees who left built YouTube, LinkedIn, Tesla, Palantir, Yelp, and a dozen other companies. The press called them the “PayPal Mafia.” Collectively, across 12+ ventures and 15 years, they created more value than PayPal itself.
That story felt extraordinary at the time. It took an entire cohort, scattering in every direction, building across multiple industries, to eventually outgrow the mothership.
What Anthropic just did makes the PayPal Mafia look slow.
On Wednesday, the company announced a $65 billion Series H round at a $965 billion post-money valuation. That number makes Anthropic the most valuable AI company on Earth, surpassing OpenAI’s $852 billion valuation from March. The round was led by Altimeter Capital, Dragoneer, Greenoaks, and Sequoia Capital. Micron, Samsung, and SK Hynix participated. So did Amazon, with another $5 billion.
Anthropic was founded in January 2021. It is four years and five months old. It was started by seven people who walked out of OpenAI because they disagreed with where the company was going. Dario Amodei, the CEO, was OpenAI’s Vice President of Research. He brought his sister Daniela and five colleagues with him.
Their seed round was $124 million. Today, the company is worth 7,782 times that. And the company they left? It’s worth less.
The Fastest Flip in Startup History
Let me put some numbers next to each other so you can feel how ridiculous this timeline is.
February 2024: Anthropic valued at $18.4 billion. OpenAI valued at $86 billion. Anthropic is worth 21% of OpenAI.
February 2026: Anthropic Series G at $380 billion. OpenAI at $300 billion (pre its March round). Anthropic has already pulled ahead.
May 28, 2026: Anthropic at $965 billion. OpenAI at $852 billion. The gap is $113 billion and widening.
In eight months, Anthropic went from $183 billion to $965 billion. That’s a 5.3x increase in valuation while the company was already a late-stage behemoth. The last time a company this large moved this fast was… actually, there is no precedent. Apple needed 16 years to go from $100 billion to $1 trillion. Anthropic is doing it in months.
Why Dario Left (and Why It Matters That He Did)
Here’s the quote that explains everything: “It is incredibly unproductive to try and argue with someone else’s vision.”
That’s Amodei explaining why he didn’t stay at OpenAI and fight for his approach to AI safety. He didn’t get fired. He wasn’t pushed out. He left because he realized that convincing Sam Altman to change direction was a worse bet than starting from scratch.
Most founders I know would have stayed. Most founders treat leaving a powerful position as failure. Amodei treated it as strategy. He had clarity about what he wanted to build (AI that’s simultaneously the most capable AND the most carefully constructed) and clarity about the fact that OpenAI’s leadership didn’t share that priority ordering.
So he took his people and left. No product. No customers. No code. Just a thesis and a team.
Five years later, Anthropic’s revenue is growing faster than any enterprise software company in history. The run rate crossed $47 billion in May. They have over 500 customers paying more than $1 million annually. The 300,000 businesses using Claude include Netflix, Goldman Sachs, KPMG, and Salesforce.
The Second-Mover Thesis (It’s Not What You Think)
The standard pitch deck slide about competition says “first-mover advantage.” Every founder has written those words at some point. And every VC has nodded along, because the intuition feels right: get there first, build the moat, defend the hill.
But the Anthropic story says something different. And it’s not the simplistic “second mover can learn from first mover’s mistakes” argument.
Here’s what actually happened: OpenAI built a product (ChatGPT) and then tried to find a business model. Anthropic built a business model (enterprise AI safety and reliability) and then built the product to serve it. OpenAI went consumer-first and is now scrambling to monetize. Anthropic went enterprise-first and is printing money.
The $47 billion run rate didn’t come from consumers paying $20/month. It came from companies embedding Claude into their production workflows and paying per token. OpenAI told the IRS it doesn’t need to make money and then filed for an IPO. Anthropic never pretended it wasn’t a business.
That’s the real second-mover advantage: you get to see the first mover’s constraint and build around it. OpenAI’s constraint was its nonprofit origin. Its hybrid structure. Its internal governance chaos (remember the Sam Altman firing?). Its consumer obsession. Its inability to say no to governments. Its sprawl into robotics, search, hardware, movies.
Anthropic looked at all of that and said: “We’re going to be a for-profit enterprise AI company that does one thing really well.” And then Claude Code became the fastest-growing developer tool ever, hitting $2.5 billion in annualized revenue within nine months of general availability.
The Pattern for Founders
Let me make this concrete for the people reading this who are building companies right now.
The pattern is not “leave big company, start competitor.” That’s obvious and usually stupid. The pattern is more specific:
1. You leave because you have a CONSTRAINT THESIS, not a feature idea. Amodei didn’t leave OpenAI because he wanted to build a better chatbot. He left because he believed AI safety and AI capability were not trade-offs, and that building for safety-conscious enterprises was a bigger market than consumer chatbots.
2. You take the TEAM, not the technology. Anthropic started with zero code. They rebuilt everything from scratch. What they brought was seven people who had deep conviction about a specific approach. The team IS the moat.
3. You pick the adjacent market, not the same one. Anthropic didn’t compete with ChatGPT for consumers. They competed for the enterprise customers who needed reliability guarantees that OpenAI couldn’t credibly offer because of its governance drama.
4. You let the first mover define the category and then you REDEFINE what winning means. OpenAI defined “AI” as “chatbot that can do anything.” Anthropic redefined winning as “AI that enterprises trust with production workloads.” One definition leads to demos. The other leads to $47 billion in revenue.
The Chip Makers Know Something You Don’t
One detail in this round that everyone is overlooking: Micron, Samsung, and SK Hynix invested. These are the companies that make the physical memory chips that AI runs on. They are not typical venture investors. They don’t write checks to be nice.
When the companies that manufacture the actual silicon invest in your AI company, they’re making a bet about who will consume the most compute in 5 years. They looked at OpenAI and Anthropic and concluded that Anthropic’s enterprise trajectory will burn through more chips. That’s a supply-chain-validated bet on who wins the volume game.
The Uncomfortable Truth
If you’re a founder reading this, the uncomfortable takeaway is not “go start a competitor to your employer.” It’s this:
Being early doesn’t protect you. Having more money doesn’t protect you. Having more users doesn’t protect you. Having a clear thesis about what you’re building AND why AND for whom, and then executing with discipline for five years, is the only thing that compounds.
OpenAI raised more capital. Launched earlier. Had more users. Had more brand recognition. Had more partnerships. Had more media coverage. Had more everything.
Anthropic had seven people with a shared disagreement and the discipline to not get distracted.
Now one of those companies is worth $965 billion, and it’s not the one that started first.
