Table of Contents
Nous Research has raised a $90 million Series B at a $1.5 billion valuation, giving the open-source AI lab new capital to expand Hermes Agent from a developer-driven product into an enterprise AI assistant.
The round was led by Robot Ventures, with reported participation from Nvidia, Microsoft’s M12, Samsung, Union Square Ventures, and other investors. The financing brings Nous Research’s total funding to about $160 million, with the Series B intended to support enterprise deployments with features such as single sign-on, employee permissions, workspace isolation, auditing, cost management, and Slack or Microsoft Teams integrations.
Those controls will help turn Hermes Agent into a commercial platform for companies that want more control over models, data, deployment environments, and agent behavior than closed AI stacks typically provide.
Hermes Moves From Developer Tool to Enterprise Platform
Hermes Agent is a self-improving AI agent that runs across desktop, terminal, cloud, and communication surfaces. It can retain memory, generate skills from prior work, schedule tasks, browse the web, run isolated subagents, use multiple sandboxing backends, and connect through apps such as Slack, Discord, Telegram, WhatsApp, Signal, email, and CLI environments.
That broad surface area is central to Nous’s enterprise pitch. Instead of selling only a hosted chatbot, the company is pushing Hermes as an agent layer that can be deployed across different infrastructure setups. Hermes supports local use, cloud deployment, and access to more than 200 models through Nous Portal.
Nous keeps Hermes free and MIT-licensed while monetizing around model licensing, tools, cloud hosting, and enterprise management services. That structure gives Nous an open-core path. Developers can inspect, run, and modify the core agent, while larger organizations pay for operational controls and hosted infrastructure.
CEO Dillon Rolnick on Keeping AI Under Users’ Control
In a note on the fundraise, Nous Research CEO Dillon Rolnick framed the round as funding for an alternative to centralized AI and closed platforms. Nous was founded in 2023 to challenge that direction, he wrote, arguing that access to powerful AI should not be controlled by a few companies.
He also criticized calls for sweeping regulation that he sees as aligned with financial interests, making the case for systems whose safety comes from auditability rather than gatekeepers.
“We worry that the conversation is being dominated by fearmongering pleas for sweeping regulation, pleas that appear aligned with, if not motivated by, financial interests. This capital will go towards building the alternative: open, competitive, ‘safer’ because of its auditability and not because of its gate-keepers.”
That philosophy shaped Nous’s early work fine-tuning Meta’s Llama models around “user-alignment”—models that follow the individual user’s worldview instead of one imposed by a provider. Rolnick pointed to the company’s work on synthetic data, open-source tool calling, and YaRN, a method for extending model context windows at lower additional cost, as foundations for its later agent development.
Hermes carries that approach into software that learns from use, remembers projects, and creates reusable skills across sessions. Rolnick said Nous released the agent under an MIT license because it believes future competitive advantages will come from “execution, trust, and extensibility,” not locking down the underlying technology.
Rolnick shared updated adoption figures in the note: Hermes has been cloned more than 24 million times and, by the company's internal estimates, now drives approximately 2.5% of all global token usage. He said thousands of contributors have helped build the agent, with thousands more building on top of it.
A Larger Bet on Open AI Infrastructure
Nous Research’s broader positioning goes beyond Hermes as a single assistant. The company describes itself as a pioneer in open AI training and research, with work spanning agents, model architecture, data synthesis, fine-tuning, and reasoning. On its website, Nous says it is building open intelligence infrastructure and calls Hermes “the most widely used open source agent harness in the world.”
The Series B is a bet that open-source AI agents can become enterprise infrastructure, software that companies can run, inspect, adapt, and connect to their own systems without giving up all control to a closed provider.
Instead of selling only a hosted chatbot, Nous is pushing Hermes as an agent layer that can be deployed across different infrastructure setups. Hermes supports local use, cloud deployment, and access to more than 200 models through Nous Portal.
Furthermore, Hermes can retain memory, generate skills from prior work, schedule tasks, browse the web, run isolated subagents, use multiple sandboxing backends, and connect through apps such as Slack, Discord, Telegram, WhatsApp, Signal, email, and CLI environments.
That’s a wealth of flexibility for enterprises that, assuming the Nous team can properly add the necessary workplace controls, can become one of the key winners in the race to turn AI agent adoption into enterprise revenue.
Hermes has been downloaded more than 22 million times since February 2026, and reportedly has reached roughly $36 million in annualized revenue, with the company expecting to exceed $100 million before the end of the year.
