Corporate contracts seize open-source code as NUS Research builds its billion-dollar valuation on privacy

The real battle in the race for AI agents was not about model conversational skill, but about who holds the operational keys within complex work environments. While open-source agent code has been treated as experimental tools on developers’ machines, NUS Research disclosed a different path by moving its open-source agent directly into enterprise infrastructure. This shift not only gave the company a valuation of $1.5 billion, but also reshaped the competitive terms in a sector that has long hesitated to hand its sensitive data over to closed cloud interfaces.

The startup, which is less than three years old, confirmed the closing of a Series B financing round of $90 million, raising its total capital raised to $108.5 million. The round was led by Robot Ventures, with notable participation from Nvidia, Samsung, Menlo Ventures and Union Square Ventures, alongside 1789 Capital, where Donald Trump Jr. serves as a partner.The investment clearly bets on open-source code’s ability to secure profitable enterprise contracts, away from software showmanship.The numbers reflect this shift: the company’s annual revenue reached about $36 million by mid-September 2026, with forecasts reported by the Wall Street Journal that it will exceed $100 million before the end of that year.

The financial rise was driven by a broad deployment base built around the company’s open-source agent, Hermes Agent, which has been forked more than 24 million times by developers and individuals. According to the company, this agent alone accounts for roughly 2.5 percent of global AI token consumption. This exceptional usage volume provided a strategic foothold that allowed the company to understand developer behavior and needs precisely, before deciding to turn this technical momentum into a commercial product for enterprises called Hermes for Business.

The Hermes for Business strategy hinges on decoupling workflow automation from data risk. Large companies that seek to employ smart agents to manage multi-step workflows face a genuine security dilemma when using proprietary models that require sharing their sensitive records with external servers.The solution offered by the company enables the deployment of custom agents that handle complex workflow chains while keeping data fully private and protected.Here the value proposition shifts from a ready-made chat model to full control over the agent’s path within the enterprise firewall.

This development has a direct implication for enterprises and tech teams in the Gulf, Egypt and the Levant, where digital sovereignty and data-protection compliance are critical requirements in banking, telecommunications and government sectors. In markets that pour massive investments into computing infrastructure, the model forces a shift in purchasing decisions, freeing organizations from the tough trade-off between advanced agent capabilities and the confidentiality of internal records. For CTOs and engineers in the region, the required skill set is changing in practice; the focus moves from issuing commands to external APIs to building advanced expertise in deploying locally independent agents and integrating them into enterprise systems while ensuring decision autonomy and record confidentiality.

The multibillion-dollar bet does not purchase a new language model, but buys companies’ confidence in delegating their complex operations.And when open-source code succeeds in generating record revenue and securing enterprise privacy, it reshapes the power map of the AI industry, confirming that the ultimate winner is the one capable of enabling organizations to run swarms of agents with complete safety and total autonomy.