Nous Research, the Austin-based open-source AI developer behind the Hermes agent, has raised $90 million in Series B funding at a reported $1.5 billion valuation—the round was led by Robot Ventures, with participation from NVIDIA, Microsoft’s M12, Samsung, Union Square Ventures, Menlo Ventures, Y Combinator, and 1789 Capital. The round brings Nous Research’s total funding to approximately $158 million.
The funding comes as Nous Research launches Hermes for Businesses, an enterprise version of its open-source AI agent that lets companies choose models by price and goal, keep their AI use private, and own their company knowledge. Hermes has been cloned more than 24 million times since its February 2026 release and drives an estimated 2.5% of all global AI token usage. The Wall Street Journal reported the company’s annualized revenue was approximately $36 million as of mid-September 2026, with expectations to surpass $100 million by year-end.
Nous Research Funding Overview By CREDX Media
| Startup Name | Nous Research |
| Founders | Dillon Rolnick (CEO) |
| Founded | 2023 |
| HQ | Austin, Texas, USA |
| Round | Series B Funding |
| Amount Raised | $90M |
| Valuation | $1.5B (reported) |
| Category | Open-Source AI, AI Agents, Decentralized Training |
| Lead Investor | Robot Ventures |
| Key Investors | NVIDIA, Microsoft’s M12, Samsung, Union Square Ventures, Menlo Ventures, Y Combinator, 1789 Capital |
| Total Funding Raised | ~$158M |
| Key Product | Hermes Agent, Hermes for Businesses, Psyche Network |
| Traction | 24M+ Hermes clones, ~2.5% of global token usage |
| Revenue (Annualized) | ~$36M as of mid-Sept 2026 |
💡 Use of Funds
- Hermes for Businesses: Building the enterprise version of the open-source agent.
- Mobile App Development: Expanding Hermes to mobile platforms.
- Decentralized Infrastructure: Advancing the Psyche Network for distributed AI training.
Startup Overview: Nous Research
What Nous Research Is Building in Simple Words
Nous Research develops open-source AI tools that give users control over their data and independence from centralized AI providers. Its flagship product, Hermes, is an AI agent released under the MIT license that remembers projects, writes reusable skills, and lets users choose which models it runs—including local models for privacy.
- Hermes Agent: An open-source AI assistant available on macOS, Windows, and Linux, integrated with Telegram, Slack, and WhatsApp. Users can switch between AI models (Claude, GPT, Gemini, DeepSeek, and 200+ others) without losing work or skills.
- Hermes for Businesses: An enterprise offering that lets companies deploy customized AI agents to handle multi-step tasks without exposing corporate data externally.
- Psyche Network: A decentralized training network that coordinates global GPUs to train large language models without relying on centralized data centers.
- DisTrO Optimizer: A technology that reduces inter-GPU communication requirements by 857x to 10,000x, enabling distributed model training over ordinary internet connections.
Why Nous Research Matters
- Open-Source AI Is Gaining Enterprise Traction The shift toward open-weight models is accelerating, with these models accounting for 61% of top-model token traffic on OpenRouter at an average cost of $0.83 per million tokens. Nous Research is betting that enterprises will seek open-source architecture for more control over their data and less dependence on specific software providers.
- Cost Advantage Is Stark Nous claims a cost of $0.30 per million tokens—roughly 50 times cheaper than the $15 per million tokens charged for GPT-5 Medium. For enterprises managing large AI workloads, the economics are compelling.
- NVIDIA Is Hedging Its Bets NVIDIA has invested billions in proprietary AI labs like Anthropic, but simultaneously backed Nous Research. NVIDIA is indifferent to which model architecture wins, provided the compute remains theirs. The dual bet signals that open-source AI is not a fringe movement—it’s a structural part of the ecosystem.
- Decentralized Training Is a Real Technical Achievement Nous Research has demonstrated that large models can be trained across distributed GPUs connected over ordinary internet connections. Its DisTrO optimizer reduced communication requirements from 74.4 GB to 86.8 MB in a test run—an 857x efficiency gain.
Startup Story: From Open-Source Fine-Tuning to a $1.5B Enterprise Bet
Nous Research was founded in 2023 by Dillon Rolnick and a team of AI researchers who wanted to push back against what they saw as a trend toward centralized, walled-off AI. The company started by fine-tuning Meta’s Llama models so they could follow the views of each user, and published YaRN, a method that gives open models longer context windows at lower cost.
The team’s early work on distributed training—DisTrO—demonstrated that frontier-class models could be trained without expensive superclusters. In a December 2024 run, Nous pre-trained a 15-billion parameter model using machines distributed around the world, livestreaming the process and showing that bandwidth requirements could be reduced by orders of magnitude.
Hermes was released in February 2026 under the MIT license. The agent’s popularity exploded among developers and individual users, with more than 24 million clones by October. CEO Dillon Rolnick framed the company’s mission in a note on the fundraise: “We brought together some of the strongest investors in technology… to advance our mission, and to bring that mission to new domains. Somehow, $90 million is modest in modern fundraising.”
The Series B now funds the next phase: bringing Hermes to enterprises through Hermes for Businesses, while continuing to develop the open-source ecosystem that made the company’s reputation.
Our Take on Nous Research
Nous Research is solving the most consequential problem in AI—who controls the technology—and doing it with a combination of open-source distribution and enterprise monetization that’s unusual in the industry.
The company’s bet is that enterprises will pay for control, privacy, and model flexibility rather than accepting vendor lock-in. Hermes for Businesses directly addresses the concerns that prevent companies from adopting AI agents at scale: data exposure, model dependency, and knowledge ownership. The 24 million clones and 2.5% token usage share demonstrate that the open-source foundation has genuine traction.
The $1.5 billion valuation is justified on multiple fronts: a massive open-source user base, a clear enterprise monetization path, and strategic backing from NVIDIA, Microsoft, and Samsung. The revenue trajectory—from $36 million annualized in mid-September to a projected $100 million by year-end—suggests commercial momentum is building.
The real risk isn’t technical—Nous has demonstrated both model quality and distributed training capabilities. The risk is enterprise sales execution. Selling to businesses requires different muscles than building open-source communities. Hermes for Businesses faces competition from established enterprise AI platforms, and the company will need to prove it can support the compliance, security, and integration requirements that large buyers demand.
For now, Nous Research has capital, traction, and a clear thesis: AI that serves the people, not just the platforms.
Founder Background: The Team Behind the Open-Source AI Movement
Dillon Rolnick (CEO) founded Nous Research in 2023 with a mission to ensure that people who use AI systems can manage their own AI. His framing of the company’s purpose has remained consistent: “The principle that the people who use AI systems should be able to manage their own AI has remained unchanged since the initial release of Hermes.”
The broader Nous Research team includes contributors to some of the most important open-source AI projects, including the DisTrO distributed training optimizer and the Psyche Network. The company’s technical credibility in the open-source community is a core asset.
The Open-Source AI Enterprise Race Nobody Expected
Nous Research operates at the intersection of two markets: open-source AI tools and enterprise AI platforms.
Open-source competitors include Meta’s Llama ecosystem and Mistral. But Nous differentiates through its agent-first approach—Hermes isn’t just a model, it’s an agent that remembers, learns skills, and works across platforms.
Enterprise AI platforms including OpenAI, Anthropic, and Google offer closed-source alternatives with strong compliance and support. Nous’s bet is that open-source architecture and model flexibility will win over enterprises that want control.
The competitive moat is ecosystem lock-in through openness. Hermes supports 200+ models, meaning users don’t get locked into any single provider. The company’s selection by NVIDIA as the reference runtime for the Nemotron 3 Ultra 550B further cements its position as a neutral infrastructure layer.
Founder Intelligence: What Founders Can Learn from Nous Research
- Build the community first, monetize second: Hermes had 24 million clones before Nous launched its enterprise product. Open-source distribution creates awareness and trust that traditional sales can’t match.
- Solve a technical problem that matters: DisTrO wasn’t a marketing gimmick—it solved the bandwidth constraint that prevents distributed training. Deep technical work attracts deep technical users.
- Strategic investors can be platform partners: NVIDIA, Microsoft, and Samsung aren’t just writing checks—they’re validating Nous’s position in the ecosystem. Choose investors who open doors.
- Open-source doesn’t mean anti-business: The MIT license on Hermes doesn’t prevent Nous from building a commercial enterprise offering. Open-source can be the foundation, not the business model.
Investor Intelligence: Why This Round Gets Premium Capital
Robot Ventures led the round, bringing deep expertise in crypto and decentralized infrastructure. The investor lineup includes some of the most strategic names in technology: NVIDIA, Microsoft’s M12, and Samsung.
The composition signals conviction across multiple dimensions: NVIDIA’s investment validates Nous’s position as infrastructure for open-weight models; Microsoft and Samsung bring enterprise credibility and potential distribution channels; Union Square Ventures and Menlo Ventures provide early-stage conviction from experienced VCs.
The reported $1.5 billion valuation places Nous Research among the most valuable open-source AI companies, reflecting both the traction of Hermes and the strategic value of its decentralized training technology.
The Road Ahead for Nous Research
Expansion Targets
With $90 million in capital, Nous Research will:
- Scale Hermes for Businesses to enterprise customers
- Build a mobile app for Hermes Agent
- Continue developing the Psyche Network for decentralized training
- Expand its open-source ecosystem and model support
Competitive Response
Expect:
- OpenAI, Anthropic, and Google to deepen enterprise AI agent offerings
- Other open-source AI companies to pursue enterprise monetization
- NVIDIA to continue investing across both open and closed AI ecosystems
- Nous Research’s moat: open-source community + model flexibility + decentralized infrastructure
Conclusion
Nous Research’s $90 million Series B at a $1.5 billion valuation validates the thesis that open-source AI can compete for enterprise customers. The company’s 24 million Hermes clones, projected $100 million revenue run rate, and strategic backing from NVIDIA, Microsoft, and Samsung position it as a credible alternative to proprietary AI platforms.
Market tailwinds favor Nous Research: enterprises seeking control over their AI infrastructure, the cost advantage of open-weight models, and growing demand for model flexibility. The company’s Hermes for Businesses offering directly addresses the data privacy and vendor lock-in concerns that limit enterprise AI adoption.
For deep analysis of AI funding, open-source innovation, and companies reshaping enterprise AI infrastructure, explore CREDX Media for comprehensive coverage of startup funding trends, valuation intelligence, and founder playbooks.
Want the latest funding news and market insights delivered straight to your inbox? Subscribe to CredX Letters for verified deal terms, valuation benchmarks, and daily macro signals.
Frequently Asked Questions
1. What is Nous Research’s total funding and latest round size?
Nous Research has raised approximately $158 million total. Its $90 million Series B was led by Robot Ventures in October 2026, at a reported $1.5 billion valuation.
2. How much did Nous Research raise in its Series B round and who led the investment?
Nous Research raised $90 million in Series B funding led by Robot Ventures, with participation from NVIDIA, Microsoft’s M12, Samsung, Union Square Ventures, Menlo Ventures, Y Combinator, and 1789 Capital.
3. What is the valuation of Nous Research?
Nous Research is valued at $1.5 billion following its Series B round, as reported by The Wall Street Journal and confirmed by the company.
4. Who are the major investors in Nous Research?
Major investors include Robot Ventures (lead), NVIDIA, Microsoft’s M12, Samsung, Union Square Ventures, Menlo Ventures, Y Combinator, and 1789 Capital.
5. What is Hermes and how does it work?
Hermes is an open-source AI agent released under the MIT license. It remembers projects, writes reusable skills, and lets users choose which models it runs—including local models for privacy. It works on macOS, Windows, and Linux, and integrates with Telegram, Slack, and WhatsApp.
6. What is Hermes for Businesses?
Hermes for Businesses is an enterprise version of the Hermes agent that lets companies choose models by price and goal, keep their AI use private, and own their company knowledge.
7. What traction justifies Nous Research’s $1.5 billion valuation?
Hermes has been cloned more than 24 million times since February 2026 and drives an estimated 2.5% of global AI token usage. The company’s annualized revenue was approximately $36 million as of mid-September 2026, with expectations to surpass $100 million by year-end.
8. Who are Nous Research’s founders?
Nous Research was founded in 2023 by Dillon Rolnick (CEO) and a team of AI researchers focused on open-source and decentralized AI.
9. What is the Psyche Network?
Psyche Network is a decentralized training network that coordinates global GPUs to train large language models without relying on centralized data centers. It uses the DisTrO optimizer to reduce communication bandwidth requirements by up to 10,000x.
10. What are Nous Research’s plans for the $90 million Series B capital?
Nous Research will scale Hermes for Businesses to enterprise customers, build a mobile app for Hermes Agent, and continue developing the Psyche Network for decentralized training.
11. What is DisTrO and why does it matter?
DisTrO is a family of distributed optimizers that reduces inter-GPU communication requirements by three to four orders of magnitude. In a test run, it reduced bandwidth from 74.4 GB to 86.8 MB—an 857x efficiency gain—enabling model training over ordinary internet connections.
Join the Discussion
No comments on this briefing yet
Join the conversation above. Share your thesis, legal nuance, or empirical observations.