Claude Just Overtook ChatGPT's Parent Company - Here Is What Actually Happened
The Market Is Catching Up To What Builders Already Knew About Claude
A Thesis I Made Four Months Ago, And Why I Am Updating It
Before the end of 2025, I wrote that none of the AI labs had what you could honestly call a moat. Models were leapfrogging each other every few weeks. Switching costs were near zero. Whoever led in November would be second by January.
Four months later, I am updating that call.
There is a winner pulling ahead, and it is Claude.
Not "winner" in the sense of the race being over - it is not. OpenAI still has more users, more brand recognition, and a higher official valuation. But if you asked me in late 2025 whether any lab could build a real lead in the next six months, I would have said no. Today the answer is different, and the data backing it is harder to wave away than a benchmark chart.
This week Anthropic briefly passed OpenAI in implied valuation on Ventuals, a platform where traders speculate on private company valuations. Anthropic traded at around $846B against OpenAI's $839B. Every AI newsletter is going to lead with that number this week.
It is not the real story. The real story is what the market is reacting to, and most of it has been visible to builders for months.
The Valuation Flip Is A Lagging Indicator, Not A Leading One
Let me get the caveat out of the way first, because it matters.
The Ventuals price is not Anthropic's actual valuation. It is a speculative derivatives market where traders take long and short positions on private companies. OpenAI's last official round was $852B post-money. Anthropic's Series G in February closed at $380B. The real funding-round gap is still more than 2x in OpenAI's favor.
So why does the Ventuals flip matter at all?
Because funding rounds are frozen snapshots. A private company's "valuation" is whatever number was agreed on during the last raise, and it stays stuck there until the next one. That is fine when the business is moving at normal speed. It breaks down when the business is tripling in four months.
Ventuals is imperfect, but it is one of the few places where real money is being put behind a continuous view of what these companies are worth right now. And right now, that view says the gap is closing much faster than the official numbers suggest.
That is the lagging indicator finally moving. The leading indicators moved months ago.
What Builders Saw First
Here is what has actually happened while most of the public conversation stayed focused on ChatGPT.
Anthropic's revenue run-rate went from $9B at the end of 2025 to $14B in February to $30B in early April. That is roughly 3x in three months. There is no precedent for that kind of growth in enterprise software history. Not Salesforce, not AWS, not anything.
On the Ramp Index, which tracks spend across more than 50,000 US businesses, Anthropic went from around 10% of combined OpenAI-plus-Anthropic business subscription spend in early 2025 to over 65% by February 2026. And the more mature an industry's AI adoption is, the more it prefers Claude. That is the pattern you want to see if you are betting on which product has the deeper work integration.
Claude Code alone is generating $2.5B in annualized revenue. A year ago, developer tooling was considered the least defensible slice of the AI market. Now it is one of the largest standalone AI revenue lines in existence.
None of this was hidden. Anyone paying attention to developer forums, enterprise procurement conversations, or simply which tool their own technical friends stopped complaining about could see it happening. The valuation market is reflecting it now. Builders saw it first because they always do.
My Own Arc Mirrors The Macro Arc
I am one data point. But the shape of my usage matches the shape of the broader curve, and I think that is worth naming.
I have been using Claude for AI-assisted coding since Sonnet 3.7. For coding, it is where I spend roughly 90% of my time. The rest goes to whatever model a specific tool happens to ship with.
For general purpose queries - research, writing, quick answers, planning - I stayed with ChatGPT. Habit, mostly. Plus a few use cases where it still felt stronger or faster. I also kept Gemini and Grok in rotation for specific tasks where each has a real edge.
Yesterday I started a Claude subscription outside of coding for the first time.
That switch is small on its own. What is interesting is how often I hear the same arc from other people now. Developers adopted Claude first because the coding difference was obvious and measurable. General-purpose usage followed later, quietly, once people realized they were already opening Claude tabs all day for work and the friction of switching back to ChatGPT for everything else stopped making sense.
This is what an adoption wedge looks like. Coding was the wedge. General use is the widening crack behind it.
Why Opus 4.6 And Sonnet 4.6 Changed The Calculation
I want to be careful here, because "Claude has no competition in coding" is the kind of sentence that invites a fight I do not need to have. GPT-5-Codex is genuinely strong. Gemini has loyal users. People have their preferences and most of them are defensible.
What I will say is this: for my actual workflow, nothing else comes close right now. Opus 4.6 for the hard problems and Sonnet 4.6 for the volume work have quietly become the default. The failure rate on non-trivial code tasks dropped to a point where I stopped double-checking with other models out of habit.
And this is where the moat question gets interesting.
The model gap alone is not a moat. Models catch up. What is harder to catch is the surrounding product layer - Claude Code as a real development tool, Skills, MCP integrations, the agentic workflows that turn the model into something you actually ship with. That is product work, not just research work, and Anthropic has been quietly out-shipping OpenAI on that axis for most of 2026.
A competitor can match the model. Matching the product takes longer. That is what a moat starts to look like.
What This Means If You Are Building Something Right Now
If you run a small business, an agency, a product team, or you are a solo builder like me, the practical read is this.
The "just use ChatGPT" default is no longer automatic. It still works fine for a lot of things. But if you have not seriously evaluated Claude for coding, technical writing, long-context work, or any workflow involving structured thinking, you are probably leaving quality on the table.
Multi-model is still the right posture. I would not tell anyone to cancel ChatGPT or Gemini. The right question is not "which AI wins" but "which tool fits which job in my stack." Claude for coding and structured work. ChatGPT for certain kinds of fast general queries and its ecosystem features. Gemini where its integrations help. Grok when you want something closer to live web behavior.
But the default slot - the tool you open first, the one that does the most load-bearing work - is now a real contest. Twelve months ago it was not.
The Honest Caveat
Anthropic could stumble. OpenAI has deeper pockets and a much larger consumer footprint. Google has distribution most labs dream about. Any of them could ship the next thing that resets the race.
But for the first time since this wave started, one lab has pulled ahead on a combination of model quality, enterprise adoption, developer preference, and revenue velocity at the same time. That combination is what a moat actually looks like when it starts forming - not any single lead, but several reinforcing leads happening at once.
Four months ago I said there were no moats. I was right at the time. I am calling it differently now, and the speculative market is starting to agree.
Takeaways
Builders are leading indicators. Funding-round valuations are lagging ones. The gap between the two is where the real signal lives.
Coding was the adoption wedge for Claude. General use is following, quietly, one user at a time.
The model gap alone is not a moat. The product layer around the model is. That is where Anthropic has been quietly pulling ahead.
Multi-model is still the right stack posture. But the default slot is genuinely up for grabs now, and more of my own default minutes keep landing on Claude.