How Nvidia and Samsung’s backing fits Nous Research’s enterprise-agent bet

The central business question is therefore not whether investors are willing to finance Nous. They are. It is whether the company can turn Hermes’ reported reach into recurring payments from organizations that use the product over time. The available evidence leaves that conversion unproven.

What does the reported valuation establish?

The $1.5 billion figure is the Journal’s reported valuation, not one Nous stated in its announcement. Without the underlying terms, the figure cannot show how investors priced the company relative to the new capital or what assumptions they used. It is a useful signal of the round’s reported scale, but not a transparent measure of customer demand.

There is also a revenue figure to weigh alongside the valuation. SiliconANGLE, citing the Journal, reported that Nous had $36 million in annualized revenue as of mid-September. That is a reported run rate, not full-year recognized revenue; the available account gives no audited figure or breakdown by product or customer type.

The same reporting said Nous expects annualized revenue to exceed $100 million by the end of 2026. That is a company forecast, not revenue already earned. Without a breakdown, it is not possible to tell how much of the reported revenue comes from individual users, inference services or business contracts.

The $36 million figure, if accurate, is evidence of some monetization and means the business cannot be assessed by usage counts alone. But it does not show whether revenue repeats, which products generate it, or how much comes from enterprise customers. A company can report revenue without having established that its new business offering has a durable customer base.

This distinction matters because the round is associated with an enterprise push, while the headline revenue figure does not identify its source. The available reporting does not connect the reported run rate to corporate Hermes deployments. Nor does it provide the customer or retention data needed to judge whether sales are growing through sustained business use.

Do Hermes’ usage figures show customer demand?

Nous says Hermes has been cloned more than 24 million times and estimates that it accounts for about 2.5% of global token usage. These are company-reported measures. The sources reviewed provide no independent verification of either figure or explanation of how the token-use estimate was calculated.

A clone count is not a count of distinct people, active users or paying customers. The reported token-use share also does not reveal whether the activity generates revenue for Nous. The figures may indicate that Hermes has circulated widely, but they cannot by themselves show how many people use it regularly or how many businesses have bought a service from the company.

The metrics should not be conflated with a separate figure in the Journal’s headline: more than 22 million downloads since Hermes’ February release. That wording and measure differ from Nous’s claim of more than 24 million clones. The available information does not explain how the figures relate, so neither should be used as a substitute for unique-user or customer data.

Nous says it released Hermes Agent under an MIT license in February 2026. The company describes the agent as able to retain project information and create reusable skills. It also says users can choose which models to use and where the system runs. Those product descriptions outline the company’s approach; the available material does not independently evaluate Hermes’ performance or reliability.

An open-source product can attract use without every user paying the developer directly. That makes distribution a possible starting point for a business, not a business result in itself. For investors and potential customers, the unanswered step is whether the use translates into paid services, continued spending and deployments that deliver value over time.

What is Nous selling to businesses?

Nous’s product materials describe a hosted team deployment and an enterprise option for infrastructure controlled by the customer. The listed features include member roles, spending caps, single sign-on and service-level agreements. The company says its new capital will support building and delivering Hermes for businesses.

That offer gives the enterprise strategy more shape than a general promise to sell AI to companies: Nous presents different ways for teams to deploy and manage the agent. The company also frames model choice and control over infrastructure as benefits for organizations. Those are stated product aims, not independently established savings, privacy outcomes or evidence of widespread corporate use.

The strongest case for the investment is that an open-source agent with model and infrastructure choices could appeal to organizations seeking control over how they use AI. Business and enterprise features could give Nous a way to sell services around software that users can also run themselves. That is a plausible commercial proposition, but the available sources do not show that it explains any particular investor’s decision or that customers have adopted Hermes broadly.

The counterpoint is not that an open-source product cannot support a business. It is that the evidence needed to assess this business is not yet established in the reporting: named customer accounts, deployment counts, repeat spending and results from actual use. Product materials show what Nous is offering or intends to offer, not whether businesses are receiving measurable value from it.

The financing therefore marks investor backing for a company pursuing an enterprise market, not proof that Hermes has secured one. A revenue breakdown by product and customer type, alongside independently documented deployments and continued paid use, would make the durability of that bet easier to judge.