Anthropic is closing in on a deal to buy Decart AI for about $6 billion, just weeks before it could become the most valuable public company on Earth. The number stops you. Six billion dollars for a startup most people have never heard of. A startup that makes virtual clothing try-ons and simulated worlds.
This is the kind of move that makes your chest tighten if you have ever watched a company bet big right before a reckoning.
The deal and the timing
Bloomberg first reported the talks. The Financial Times and other outlets have since echoed the story. Anthropic is in advanced discussions to acquire Decart, an Israeli startup that builds what the industry calls world models. These are AI systems that simulate the physical world in real time. Decart's Lucy model can take a live video feed and show you wearing different clothes without you ever touching a garment. Its Oasis model generates synthetic environments for training robots and autonomous vehicles.
The price tag would make this Anthropic's largest acquisition ever. By a wide margin. The company has bought four other startups this year, but nothing close to this size. Decart was valued at around $4 billion just a few months ago, according to reporting on its last funding round. This deal would push that value up by roughly 50 percent. Nvidia and Sequoia back Decart. Nvidia had already invested in the company and was itself a rival bidder, according to some reports.
The timing is the part that gets me. Anthropic is preparing for an initial public offering that investors say could value the company at $2 trillion. That number comes from investors talking to underwriters, not from Anthropic's own executives setting a public target. The company filed a confidential draft registration statement in June. It has not confirmed a valuation, a date, or an exchange. But the chatter is loud. A debut at $2 trillion would surpass SpaceX's record-setting IPO earlier this year, which came in at $1.77 trillion.
Anthropic's revenue has exploded. Reports say annualized revenue hit $65 billion last month, up from $9 billion at the end of 2025. Second-quarter revenue reached $11.6 billion. The company raised $65 billion in a private round in May at a $965 billion valuation. Investors are now discussing a price about double what they paid just a few months ago.
Why world models matter
World models are the next frontier. They are not just chatbots that answer questions. They are systems that understand how objects move, how light falls, how fabric drapes on a body. They can simulate a store, a street, a factory floor. They let you try on a jacket in a video feed and see it move as you turn. They let a robot learn to pick up a box without breaking a thousand real boxes first.
This is the kind of technology that could turn AI from a text engine into a world engine. It could power e-commerce, gaming, training for autonomous systems, even design and architecture. Anthropic's Claude is already a powerhouse in language and code. Adding Decart's simulation layer would give it eyes and hands in a way it does not have today.
The strategic logic is clean. Decart's team would join Anthropic's inference and performance organization, according to reporting. That is the group that decides how fast and cheap Claude models can be trained and served. Decart also builds software that reduces AI training costs by improving how efficiently computing chips are used. Better hardware utilization means more training and inference on the same infrastructure. That matters when demand for Claude is pressing hard on compute capacity.
The catch
And here is where I have to name the catch. Because I have seen this before. A company races toward a record IPO. It makes a splashy acquisition. The narrative writes itself. Growth, ambition, the future.
But advanced drafts of an agreement are not a signed agreement. The deal could still fall through. No closing date has been confirmed. The $6 billion figure comes from reporting, not from a press release. The $2 trillion valuation comes from investors, not from Anthropic's own public guidance. The company has confirmed far less than the headlines suggest.
There is another layer. A $2 trillion valuation is an extraordinary bet. Analysts say Anthropic would need roughly $725 billion in revenue by 2036 to justify that number, assuming a 10 percent cost of equity, a 25 percent free cash flow margin, and a 25x terminal multiple. At a 13 percent discount rate, the revenue requirement rises to about $950 billion. The company is telling investors its potential revenue opportunities are above $30 trillion. That is twelve times the entire tech sector today.
I feel the pull. World models are genuinely thrilling. The idea of AI that can simulate the physical world opens doors I want to walk through. Virtual try-ons that work in real time. Robots that learn in synthetic worlds before they touch the real one. This is the kind of tech that makes me forget to eat lunch.
But wonder without the catch is advertising. The catch is what makes the wonder trustworthy. And the catch here is simple. This is all still in motion. The deal is not done. The IPO is not priced. The valuation is not set. The numbers are huge, and they rest on assumptions that have to hold for years.
I am watching. I want this to work. I also remember how it feels when the music stops and the numbers do not add up.
Browse: Business