Nvidia, Microsoft and Samsung Back Nous Research Series B at $1.5B Valuation
Nous Research Series B raises $90M at a $1.5B valuation, led by Robot Ventures with Nvidia, Microsoft and Samsung backing its Hermes Agent enterprise push.
Nous Research has raised $90 million in a Series B round that values the open-source AI developer at $1.5 billion. Robot Ventures led the Nous Research Series B, which Nous Research confirmed on 7 October 2026, with Nvidia, Microsoft's M12 fund, Samsung and Union Square Ventures among the participants.
Total funding is now roughly $158 million since the company was founded three years ago. The new capital is earmarked for Hermes for Businesses, an enterprise edition of the open-source Hermes Agent, and for a mobile app built on the same technology.
Hermes has been the company's calling card since it arrived on the open-source scene. The agent has been downloaded 22.7 million times since February, a distribution base that most enterprise AI vendors spend heavily to assemble through sales teams and cloud marketplaces.
What Hermes for Businesses Does
Nous Research is positioning the enterprise edition as a deployment-flexible agent that businesses can run where their data already sits. Customers can choose a local installation or a cloud configuration, and the company says the agent can be customized for multi-step workflows while keeping data private and secure.
That flexibility addresses the two questions that stall most enterprise agent evaluations: where the model runs, and who can see the prompts and the outputs. An on-premise option removes a common blocker for regulated sectors such as finance, healthcare and legal services, where data is not permitted to leave a controlled environment.
Security configuration is the other half of the pitch. Customizable agent behavior lets a company restrict which systems an agent can touch, which files it can read and which actions require human approval. That granularity separates a pilot from a production deployment in most large organizations, and it is the part of the product procurement teams test first.
Local deployment shifts costs as well as control. A customer running Hermes on its own hardware pays for compute, power and operations instead of per-token API fees, which changes the economics of high-volume agent workloads. The trade-off is that the customer also owns the uptime, patching and scaling burden that a hosted vendor would normally absorb.
The company has not said which models or runtimes sit under the enterprise edition, or how much of the agent stack a customer can swap out. For buyers weighing an open-source agent against a hosted alternative, that detail decides how portable the deployment stays if requirements change.
The timing lands Nous Research in a market that has moved from demonstrations to procurement. Agent projects inside large companies now pass through security review, budget approval and integration testing, and vendors that cannot answer those reviews lose deals regardless of model quality.
The mobile app points to a second ambition. Nous Research has said the funding will extend Hermes into new territory, and a phone-based agent moves the product from developer tooling toward consumer reach. That is a different business with different distribution economics, and it places the company nearer to the assistant platforms that already ship on handsets.
Inside the Nous Research Series B
The names on the cap table say as much about the round as the dollar figure does. Nvidia, Microsoft and Samsung all run their own agent programs, and all three now hold a stake in an open-source agent lab alongside their positions in closed frontier developers.
Nvidia's interest is easy to trace. Agents generate sustained inference demand, and inference is where the company's data center business is increasingly weighted. Every Hermes deployment, whether on-premise or in the cloud, runs on some accelerator, so wider agent adoption expands the market Nvidia sells into.
Microsoft's M12 gives the company visibility into an agent distribution model that does not depend on a single cloud vendor. Samsung's venture arm has a device-side motive: an agent that runs locally fits a phone-first strategy, and Nous Research has committed to building exactly that. Union Square Ventures, Y Combinator and Menlo Ventures also appear among the backers named by the company.
Strategic money of this kind usually arrives with more than capital attached. Corporate investors in AI infrastructure often negotiate commercial relationships alongside their checks, from compute credits to channel access, and those arrangements can matter more to an early enterprise push than the headline amount.
| Detail | Figure |
|---|---|
| Round | $90 million Series B |
| Valuation | $1.5 billion |
| Lead investor | Robot Ventures |
| Other investors | Nvidia, Microsoft M12, Samsung, Union Square Ventures |
| Total funding | About $158 million |
| Stated use of funds | Hermes for Businesses, mobile app |
| Hermes downloads | 22.7 million since February |
The Nous Research Series B values the company at $1.5 billion, a figure that rests on a specific assumption: that an install base of millions of developers converts into a smaller but far more lucrative base of enterprise contracts. Roughly $158 million raised over three years is modest next to the capital deployed by the largest AI developers, and the next phase will be measured in signed deployments rather than downloads.
That pattern explains why the three corporate investors can back both open and closed approaches without contradiction. The agent stack has several layers: the underlying model, the runtime that executes tasks, the orchestration that decides what runs next, and the deployment environment that holds the data. Holding a position at each layer is cheaper than betting an enterprise strategy on one of them.
Open Source as a Go-to-Market
Nous Research's bet is that free distribution compounds faster than a closed sales motion. Hermes is open source, self-improving and freely installable, and the 22.7 million downloads show that model working. The enterprise product then converts a slice of that install base into paying accounts with support, deployment assistance and security guarantees.
The self-improving design carries weight in procurement. An agent that adjusts its own behavior from past runs is harder to audit than a static model, and buyers in regulated industries will want logs, rollback paths and clear limits on what the agent can change without approval. Those controls have not been detailed publicly, and that gap will shape how quickly regulated buyers move.
Procurement teams will also press on data residency, license terms, upgrade cadence and support commitments. None of those details appeared with the funding announcement, which is typical for an enterprise product that has just been named rather than shipped. Their eventual shape will decide whether Hermes for Businesses lands as a platform decision or a departmental experiment.
The approach echoes the playbook that turned open-weight models into a commercial category, where a free download builds familiarity and a paid tier sells the operational wrapping. It also creates tension. Enterprises want auditability and vendor accountability, while open-source communities expect transparency and permissive licensing. Pricing and licensing terms for Hermes for Businesses have not been published.
What $90 million buys in this model differs from what it buys a closed lab. Closed developers spend heavily on compute for frontier training, while an open-source lab channels capital into enterprise engineering, compliance work and a support organization. Capital intensity is lower. The moat is thinner, because a freely distributed agent can be wrapped and resold by anyone with the same access.
Competition in the enterprise agent market is already dense. Closed labs have spent two years pushing agent frameworks into business workflows, and systems integrators have built practices around them. The differentiator Nous Research is selling is control: local deployment, customization, and a model that customers can inspect rather than rent by the token.
Why this matters
The Nous Research Series B shows the largest technology companies hedging across the agent market rather than picking a single winner. Nvidia, Microsoft and Samsung now hold positions in both closed frontier developers and an open-source alternative, which points to a plural enterprise agent stack: some workloads on proprietary models, others on inspectable agents running inside the customer's own perimeter. For buyers, that optionality is worth more than any one vendor's roadmap, and it makes lock-in a harder argument to sell. For Nous Research, the $90 million buys time to prove that open distribution converts into enterprise revenue before the incumbent platforms close the gap.
Sources
Related Articles
- AI Business Roundup: Compute Capital, Regulation and Agent Adoption Defined the Week
- AI Funding Tracker: Two $3B Rounds and Nvidia's $2.5B Stake Set the Pace
- Nvidia Partners with AlphaGo Architect to Launch Self-Improving Hermes Agent AI
✔Human Verified
Researched and cross-referenced against primary sources by the Bytevyte editorial team. This article was generated with the assistance of artificial intelligence and reviewed by the Bytevyte editorial team.