Nvidia backs Reactor’s $74M Series A for real-time world models

Nvidia and Sapphire Ventures have joined Reactor’s Series A round, bringing the real-time world model startup’s total funding to $74 million. The investment was first reported by Crypto Briefing.

Reactor builds simulation platforms that generate dynamic, physics-accurate environments for robotics, autonomous systems, and industrial applications. The round follows a string of high-profile AI and robotics deals, including Nvidia’s $12.9 billion acquisition of Hugging Face last month—a move that expanded its stake in open-source AI tooling. Unlike Hugging Face, Reactor operates further downstream, focusing on the infrastructure that trains and tests AI agents in virtual worlds.

The funding arrives as Nvidia faces pressure from startups like Wafer, which targets its hardware dominance with AI-optimized alternatives. Reactor’s work, however, aligns with Nvidia’s broader strategy: enabling the simulations its customers rely on. The startup’s real-time capabilities could also complement Nvidia’s own efforts in digital twins and robotics training.

What remains unclear is how Reactor’s technology will differentiate itself in a crowded field. UK-based Zenithon, which raised $10 million last month, is building world models for extreme physics, while XDOF, negotiating a Series B at a $1.2 billion valuation, focuses on robotics data infrastructure. Reactor’s edge may lie in its real-time performance, but the startup has yet to disclose benchmarks or enterprise adoption.

For now, Nvidia’s backing gives Reactor a high-profile validator—and a reminder that the AI stack is being built in layers, from data to simulation. The next test will be whether Reactor can turn its funding into a platform others can’t ignore.

Sources: cryptobriefing.com