Bezos-backed Taurus, a secretive AI startup founded by former Google DeepMind researchers, has raised $10 billion in one of the largest venture rounds in AI history, landing a $38 billion valuation and instantly vaulting into the same conversation as OpenAI and Anthropic.
The round, led by T. Rowe Price and Fidelity with participation from NVIDIA and Amazon, brings the company's total raised equity past $14 billion. That's a 50% premium over where the company was valued in late 2024, a jump that tells you everything about how fast investor sentiment has shifted toward agentic AI over the past eighteen months.
Taurus has operated almost entirely under the radar since its 2022 founding by David Dohan and Mihir Parmar, two researchers who left Google DeepMind to pursue a specific and ambitious thesis: that the next frontier in AI isn't smarter language models, it's autonomous agents capable of executing complex, multi-step workflows in real enterprise environments. While OpenAI and Anthropic dominated headlines with ever-larger LLMs, Taurus was quietly building in a direction the market is only now fully pricing in.
The scale of this round is a signal, not just a transaction. A $10 billion capital infusion at this stage, with institutional players like Fidelity alongside strategic backers NVIDIA and Amazon, suggests the investors aren't taking a speculative bet. They're locking in early infrastructure access. A significant portion of the capital is expected to go toward securing NVIDIA's H100 and next-generation Blackwell GPUs, the compute bedrock needed to train Taurus' forthcoming flagship model. In an environment where GPU allocation has become a competitive moat in itself, that calculus matters enormously.
Bezos' direct involvement here carries its own weight. Amazon is already a backer, and Bezos has been positioning himself aggressively in the AI stack since stepping back from day-to-day operations at Amazon. With this round, he's effectively established a third pole in the US AI ecosystem, one that sits outside the Microsoft-OpenAI and Google-Anthropic gravitational fields. That's a structural shift worth watching: the AI arms race, once a two-horse competition between tech giants, now has a well-capitalized independent player with serious enterprise ambitions and a founder whose instinct for long infrastructure bets has been validated repeatedly.
For the AI sector, a $38 billion valuation achieved almost entirely in stealth is a credibility moment for the agentic AI category. Enterprise automation and autonomous workflow tools have been generating buzz at the product level, but this capital event confirms that institutional money has made its conviction clear. Competitors in the same space, particularly mid-size startups without comparable compute access or investor backing, now face a materially harder landscape.
Merger and acquisition activity in AI is likely to accelerate as a direct result. Smaller players developing adjacent tooling, whether in orchestration, memory systems, or vertical-specific agents, become acquisition targets not just for the big labs but now for Taurus as it builds out its stack. The company's willingness to raise at this scale suggests it's not optimizing for a near-term exit but positioning for a decade-long infrastructure play.
The more immediate question is what Taurus actually ships. Stealth mode has served it well for fundraising, but the company will need to demonstrate model performance and enterprise traction to justify its valuation as the market matures. With this round closed, the clock starts. Watch for a flagship model announcement before year-end and, likely, the first major enterprise partnership disclosures that will give the market something concrete to evaluate against that $38 billion number.
Also read: Amazon commits $25 billion more to Anthropic in a bet that could reshape the cloud and chip landscape • Lovable's API flaw exposed private project data from the $6.6 billion AI app builder used by Nvidia and Microsoft teams • GPT-Image-2 reviews and corrects its own output before you ever see it
