Nous Research valued at $1.5 billion after $90 million in new capital
Open-source-driven AI lab Nous Research announces a $90 million Series B at a $1.5 billion valuation — eight months after the MIT-licensed Hermes Agent launched. The capital will now fund an enterprise version, and the question is whether "open source plus enterprise controls" can become a durable business model.
Nous Research has raised $90 million in a Series B round valuing the company at $1.5 billion, according to the company's own announcement, published October 7, 2026 and reported by Crypto Briefing. The round is led by Robot Ventures, with NVIDIA, Microsoft's venture arm M12, Samsung, Y Combinator, Union Square Ventures and Menlo Ventures among the other investors — but the list is, per the source, not exhaustive, so the shareholder roster remains incomplete.
The capital history — and a digital-asset thread
The new round brings the company's total capital raised to roughly $160 million. The previous major round was a $50 million Series A led by Paradigm in April 2025, which Crypto Briefing reports was part of a package of around $70 million consisting of equity and a so-called SAFT. A SAFT (Simple Agreement for Future Tokens) is a contract in which investors purchase the right to tokens a project expects to issue later, rather than buying equity today. The fact that a SAFT is part of the capital history gives Nous Research's funding a digital-asset dimension beyond a conventional VC story — but the source does not specify how the Series B itself is split between equity and any token component.
What Hermes Agent is
The core of the push is Hermes Agent, launched in February 2026 under the MIT license — one of the most permissive open-source licenses available. Developers can use, modify and redistribute the code with very few restrictions. According to the company's description, the agent is built to be autonomous and self-improving, with persistent memory across sessions and the ability to generate its own reusable skills.
The company says Hermes Agent has been downloaded or cloned more than 24 million times since launch, and that the agent accounts for roughly 2.5 percent of global token usage in AI inference. Both figures are the company's own claims relayed via Crypto Briefing, and they are not independently verified.
The turn toward the enterprise market
The primary use of the new capital is an offering the company calls "Hermes for Businesses." The plan is to add the features corporate IT departments require before anything is allowed near a company machine — including SSO (single sign-on), audit trails and workspace controls. The company also says a mobile app is planned, and that Hermes will remain open source.
No launch date or pricing has been confirmed for either the business product or the mobile app — the plans currently exist only as company statements.
The founders and the open question
Nous Research was founded in 2023 by Jeffrey Quesnelle, Karan Malhotra, Ryan Teknium and Shivani Mitra, and has positioned itself from the start as an open-source AI lab. CEO Dillon Rolnick signed the announcement note on the funding.
The obvious tension in the story is the business model: how do you make money on a product anyone can copy for free? The answer Nous Research is betting on is the classic "open core, closed enterprise" pattern — the community version builds reach and credibility, while enterprise features generate revenue. Investors including NVIDIA, Microsoft's M12 and Samsung evidently believe the math works.
But several matters remain unresolved. All the key figures — downloads, inference share, valuation — originate from the company itself, relayed through a single secondary source, with no primary documentation. The relationship between the earlier SAFT agreements and today's equity ownership is unclear, and it is not known whether the Series B contains a token component. And the biggest question remains unanswered: whether an enterprise version of an MIT-licensed tool can generate revenue justifying a billion-dollar valuation — or whether competitors can copy both the agent and the enterprise features without paying anything at all.