When a number stops making sense, a certain kind of silence descends upon a venture capital office. One of those figures is eight hundred billion dollars. The figure behind Claude is being whispered, printed, and presented in term sheets to Anthropic, the AI lab in San Francisco. Furthermore, the number itself isn’t the most peculiar aspect. People familiar with the offers claim that Anthropic has been declining.
The company closed a $380 billion round just two months ago. By conventional standards, that was already ridiculous. Investors are now circling with proposals that would more than double it, putting Anthropic on par with OpenAI’s $852 billion valuation. From the outside, it’s difficult to ignore how rapidly the goalposts are changing. There’s a feeling that no one really understands the value of these businesses; they just need to get a piece before someone else does.
| Anthropic — At a Glance | |
|---|---|
| Company | Anthropic |
| Founded | 2021 |
| Headquarters | San Francisco, California |
| Co-founders | Dario Amodei, Daniela Amodei |
| Flagship Product | Claude AI Assistant |
| Latest Model | Mythos (restricted release) |
| Current Reported Valuation | Up to $800 billion |
| Previous Valuation (Feb 2026) | $380 billion |
| Annualized Run-Rate Revenue | Over $30 billion |
| Revenue (End of 2025) | ~$9 billion |
| Key Backers | Amazon (AMZN), Alphabet/Google (GOOG) |
| Primary Rival | OpenAI ($852B valuation) |
| Potential IPO | Reportedly being explored for 2026 |
A portion of it can be explained by the revenue figures. From about $9 billion at the end of 2025, Anthropic’s annualized run-rate revenue has surpassed $30 billion. That is the kind of growth curve that makes spreadsheets seem like fiction—a tripling in less than a year. A large portion of it can be traced back to Claude’s enterprise adoption, especially the coding and agentic tools that have subtly become indispensable within engineering teams at businesses that most people are unaware of. Meanwhile, the Claude assistant frequently appears in unexpected locations, such as independent game studios, legal departments, and hospital administrative offices.
Mythos comes next. When the company recently unveiled its newest model, it practically immediately limited access, citing the model’s capabilities as justification. That choice caused controversy. Anthropic might be acting with genuine caution. It’s also possible that the restriction itself contributes to the narrative that investors are purchasing; the notion that the model is too potent to be released seems like it belongs in a pitch deck.
The chessboard is made more difficult by the fact that Amazon and Alphabet are already deep in the cap table. Microsoft is on the OpenAI side. With GPUs, datacenter leases, and quarterly funding rounds that would have been considered historic events ten years ago, the two ecosystems are now locked in what appears less like competition and more like a slow-motion infrastructure war. For its part, Reuters stated that it was unable to confirm the precise $800 billion offers right away. They are credible because of their trajectory. It’s another matter entirely whether they make them reasonable.
An IPO is being discussed for this year. Retail investors would have a significant appetite if it were to happen because it would be their first opportunity to directly invest in one of these businesses. However, IPOs are scrutinized more than private rounds. Harder questions concerning margins, defensibility, and whether $30 billion in annualized revenue truly translates into long-term profit are raised by public markets. Years ago, Tesla encountered similar skepticism and overcame it. Many others didn’t.
You wouldn’t think any of this was going on outside Anthropic’s office on a weekday. There are no banners or marketing spectacles. Like everywhere else in the city, it’s just engineers entering with coffee. And someone is graciously refusing to accept eight hundred billion dollars somewhere upstairs. For the time being.