Nous Research raises 90M dollars at a 1.5B valuation and launches enterprise Hermes agents

Nous Research, the open source artificial intelligence lab that has spent three years arguing against walled garden model development, has moved into the corporate software market. On October 7, 2026, the company confirmed a $90 million Series B round at a $1.5 billion valuation and introduced Hermes for Businesses, an enterprise edition of the Hermes Agent that individuals and developers have been running since February. The round was led by Robot Ventures, with checks from Nvidia, Microsoft's M12, Samsung, Union Square Ventures, Menlo Ventures, Y Combinator and 1789 Capital, a fund where Donald Trump Jr. is a partner.

The startup was formed in 2023 by Jeffrey Quesnelle, Karan Malhotra, Ryan Teknium and Shivani Mitra, and is led by chief executive Dillon Rolnick. Its founding premise was to push back against what it describes as a trend dragging AI toward centralization and walled gardens. That work started with fine tuning of Meta's Llama models so that outputs follow the views of each individual user, and expanded into synthetic data work, open source tool calling and YaRN, a published method that gives open models much longer context windows at far less additional cost.

Hermes Agent arrived in February 2026 under the MIT license. It was built so that the more a user runs it, the better it gets: it keeps persistent memory across sessions, writes its own reusable skills as it works, and carries knowledge from one project to the next. Users decide where it runs and which models it uses, on macOS, Windows and Linux, and through Telegram, Slack and WhatsApp. That distribution has been remarkable. Nous says the agent has been cloned more than 24 million times and, by its own internal estimates, drives approximately 2.5% of all global token usage.

Until now, that reach has produced attention far more reliably than revenue. The new capital is meant to change that, funding an enterprise push that adds the controls corporate technology departments demand while the underlying project stays open source.

Key Facts

TechCrunch reported on October 7, 2026 that Nous Research raised a $90 million Series B at a $1.5 billion valuation, confirming the outlet's own earlier reporting from July 13, 2026. The round was led by Robot Ventures, with participation from Nvidia, Union Square Ventures, Menlo Ventures, Samsung and 1789 Capital, where Donald Trump Jr. is a partner. TechCrunch also reported on October 7, 2026 that the financing brings the three year old startup's total funding to $158 million.

Crypto Briefing reported on October 7, 2026 that the raise brings total capital to approximately $160 million, and that the Series B follows a $50 million Series A led by Paradigm in April 2025, part of an earlier package of equity and Simple Agreement for Future Tokens funding totaling around $70 million. The outlet reported that the money is earmarked for Hermes Agent, a business focused version of it and a planned mobile app.

The Next Web reported on October 7, 2026 that Nvidia, Microsoft's M12, Samsung, Robot Ventures, Union Square Ventures, Y Combinator and Menlo Ventures are among the investors, citing a note on the company's website from Rolnick. In that note, Rolnick wrote: "Somehow, $90 million is modest in modern fundraising. We at Nous have always delivered more with less." He added that the company raised the money "to create AI that serves the people."

Traction figures are sizeable. Nous says Hermes has been cloned more than 24 million times, and The Wall Street Journal reported more than 22 million downloads since February. TechCrunch reported that Nous was at roughly $36 million in annualized revenue by mid September 2026 and expects to pass $100 million before the end of 2026, according to The Wall Street Journal. The company estimates the agent accounts for about 2.5% of global AI token inference usage.

Hermes for Businesses keeps the open source core but adds what corporate buyers ask for: Single Sign On, detailed auditing capabilities and workspace controls. Companies will be able to choose models by price and goal, keep their AI use private and secure, and hold on to their own institutional knowledge. The company says the product will cut spend by letting businesses pick the best models at the best prices while giving them dominion over company knowledge.

Analysis

What this really means is that open source distribution has become the cheapest enterprise sales channel in artificial intelligence. Nous did not spend years building a proprietary model and then court chief information officers. It gave away an agent under the MIT license, watched it accumulate more than 24 million clones and roughly 2.5% of global token usage, and is now selling the governance layer on top of that footprint. Single Sign On, auditing and workspace controls are not glamorous features, but they are precisely the features that convert individual enthusiasm into signed contracts.

The valuation math deserves scrutiny. A $1.5 billion valuation against roughly $36 million in annualized revenue implies a multiple of about 42 times sales, which is rich even for a hot AI market. The counterargument is growth: management expects to pass $100 million in annualized revenue before the end of 2026, which would cut that multiple to roughly 15 times on a forward basis if the target is met. The bigger picture here is that investors are not pricing the current revenue line. They are pricing a distribution channel that already touches a measurable slice of global inference, and a business model that can be layered on top of it without the cost of training frontier models from scratch.

The investor list tells its own story. Nvidia, Microsoft's M12 and Samsung all sell the compute and cloud infrastructure that open agents consume, so a stake in Nous is partly a hedge on demand for inference regardless of which model wins. Robot Ventures, Union Square Ventures, Menlo Ventures and Y Combinator bring earlier stage networks, while 1789 Capital adds a political and financial connection that is unusual for a research lab. The mix suggests the round was supported on strategic as much as financial logic.

There is a real tension, however, in selling enterprise controls around a project that anyone can clone for free under the MIT license. Nous has to convince buyers that the paid tier is worth paying for when the agent itself is not scarce. Its answer is trust, compliance and knowledge ownership rather than model access, which is defensible only as long as the company keeps shipping the self improving capabilities that made Hermes popular. Rolnick's own framing, that $90 million is modest, is a useful reminder that the company sees this round as fuel rather than validation.

Why It Matters

For corporate technology buyers, the arrival of a credible open source agent vendor changes the procurement conversation. Until recently, enterprises that wanted agentic workflows mostly chose between closed platforms that keep data inside a vendor's walls and homegrown projects that are hard to govern. Hermes for Businesses promises a third path: pick whichever model offers the best price for a given task, keep the data private, and retain ownership of the institutional knowledge the agent accumulates.

For the wider open source AI ecosystem, the round is a signal that permissive licensing and commercial success are not mutually exclusive. Nous built its reputation on Llama fine tunes, synthetic data work, open source tool calling and the YaRN context extension method, all of which were published rather than fenced off. If the enterprise tier succeeds, it becomes a template for other labs that want to keep their methods open while still paying for compute and staff.

For investors, the round is a test of whether distribution beats model ownership. Nous does not claim to have the strongest frontier model. It claims to have an agent that people actually use, at a scale measured in tens of millions of clones, and that is a different kind of asset than a benchmark score.

Next Up

The immediate priorities are the enterprise rollout and a mobile app that Crypto Briefing reported is planned alongside it. Nous has also set a public revenue target of more than $100 million annualized before the end of 2026, up from roughly $36 million in mid September, a pace that will require the business tier to convert quickly.

Watch for two things in particular: how many of the 24 million plus clones turn into paying corporate seats, and whether the company can add compliance features fast enough to satisfy regulated industries without alienating the developer community that made Hermes worth cloning in the first place.