Nous Research confirms it hit $1.5B Valuation, launches AI agents for business users

Nous Research raised a $90 million Series B at a $1.5 billion valuation, bringing total funding to $158 million. Its open-source Hermes agent has been cloned more than 24 million times and accounts for roughly 2.5% of global AI token usage, according to the startup. The company plans to use the funding to expand into enterprise deployments; WSJ reported annualized revenue of about $36 million by mid-September 2026 and an expectation to exceed $100 million before year-end.

Analysis

The investable signal is indirect: a widely used open-source agent could expand total inference demand, supporting NVIDIA’s ecosystem, but usage does not automatically translate into NVIDIA revenue. If agents make multi-step work cheaper and more useful, workload growth may outweigh lower cost per task; if efficiency gains dominate or workloads run on competing accelerators, token activity can rise without equivalent GPU demand. The more important competitive question is where enterprise value accrues: private deployment and workflow controls may benefit infrastructure and security vendors, while pressure-testing whether model and agent layers become commoditized. The reported revenue trajectory and token share are company-reported indicators, not proof of durable enterprise bookings or NVIDIA-specific demand. Over the next 1–3 months, watch enterprise customer wins, recurring revenue quality, and deployment choices; over 6–18 months, agent adoption could be a structural inference tailwind, but it may also intensify price competition across AI infrastructure. NVIDIA’s participation is a modest ecosystem-validation signal, not a material standalone earnings catalyst. No direct trade is warranted on this funding news alone.

Key Decisions for Investors

• Keep NVIDIA exposure unchanged on this news; treat the investment as a small positive ecosystem signal rather than a reason to add ahead of earnings.
• Monitor NVIDIA’s data-center guidance and commentary on inference demand alongside evidence that agent workloads are expanding paid enterprise deployments. A slowdown in inference growth or weaker guidance would falsify the bullish read-through.
• Track whether enterprises choose private or self-hosted deployments versus offerings from Microsoft, Amazon, and Google. Broad adoption across infrastructure providers is more supportive of total compute demand than a win for any single agent vendor.
• Avoid valuing the private startup’s reported usage or annualized revenue as a proxy for NVIDIA revenue; revisit only after independently verifiable enterprise bookings, retention, and workload deployment data emerge.