The news
Nous Research, the open-source AI lab behind the Hermes Agent assistant, said on October 7, 2026 that it raised a $90 million Series B at a $1.5 billion valuation. Nvidia (NVDA) and Samsung Electronics are among the backers, and the lab plans to use the money to sell Hermes to businesses.
SiliconANGLE reported that Robot Ventures led the round, with Nvidia, Samsung, Y Combinator and several others participating. The Next Web, citing The Wall Street Journal, listed Microsoft's venture arm M12, Union Square Ventures and Menlo Ventures among the investors as well.
Hermes Agent, released in February 2026, is an AI agent: software that carries out multi-step tasks on a user's behalf rather than only answering questions. It runs as a desktop application and a cloud service and can work with a user's files, programs and web browser, according to SiliconANGLE. The Next Web reported that it runs on macOS, Windows and Linux and connects to Telegram, Slack and WhatsApp.
Usage figures vary by source. Nous says Hermes has been cloned more than 24 million times and drives about 2.5% of global AI token usage, an internal estimate, according to The Next Web. The same outlet said The Wall Street Journal put downloads at more than 22 million.
SiliconANGLE reported that Nous had $36 million in annualized revenue in mid-September and projected more than $100 million by year-end. The lab sells three tiers: Hermes Cloud for consumers, Hermes Business with team-level controls, and Hermes Enterprise with organization-wide administration.
The new money is aimed at Hermes for Businesses, which The Next Web said lets companies choose models by price and objective, keep data private and retain ownership of what the organization learns.
The numbers
- Series B
- $90 million
- Valuation
- $1.5 billion
- Annualized revenue, mid-September (SiliconANGLE)
- $36 million
- Hermes clones (company figure)
- 24 million+
- Share of global token usage (company estimate)
- About 2.5%
Why CEOs should care
For technology buyers, Nous represents a different bet from the closed agents sold by the largest labs. An open-source agent that can switch between models lets a company route routine work to cheaper models and keep sensitive data in-house. Before piloting it, ask who supports the software in production, how model routing decisions are logged, and what access the agent gets to files and browsers on employee machines.
For CISOs, the download numbers are the warning sign. A free desktop agent with tens of millions of downloads is likely already on some company laptops. Check endpoint inventories for Hermes and similar agents, and set a policy on which agents may touch corporate files, chat tools such as Slack, and credentials stored in browsers.
For CFOs and boards, the investor list matters. When Nvidia and Samsung fund software that consumes large volumes of tokens, they are also funding future demand for their chips and memory. Leaders weighing AI budgets should expect hardware suppliers to keep subsidizing tools that raise usage, and should track consumption-based costs closely as agents run longer tasks.
The bigger picture
The round shows hardware makers spreading bets beyond the biggest closed-model labs. Open-source agents that run many model calls per task create steady demand for compute whichever model wins, which gives chip and memory makers a reason to back them.
It also shows how fast agent startups are moving to enterprise sales. Nous is following a familiar open-source path: build a large free user base, then charge companies for controls, administration and support.
What’s next
Watch whether Nous reaches the more than $100 million year-end revenue figure SiliconANGLE reported, how quickly Hermes for Businesses wins named corporate customers, and whether its Business and Enterprise tiers give IT teams enough control over an agent that already runs on employee laptops.
What “Fact-checked” means
Fact-checking means testing a story’s facts against the evidence before it is published. This story went through at least two separate checks before this version was published.
- What we checked
- Its names, figures, dates, job titles, quotes and who said what were checked against the story’s sources, including its main source where it could be opened. The headline was checked for accuracy and overstatement.
- How
- A first check reviewed the whole story. If it passed, a second, skeptical check went back to the sources to look for mistakes in the most important facts. If a check flagged the story, it was edited to fix the problems found, and a separate re-check then reviewed the whole story again.
- Who
- The checks are made by our newsroom, as steps kept separate from the writing, under rules set by our editor, Hussein Mukhtar. A story the checks still flag is held for the editor, who decides whether it is fixed, published or dropped.
- If something is wrong
- “Fact-checked” does not mean error-free. If a material error is found after publication, we correct the story and add a note saying what changed. Report an error
