Modular, the company developing infrastructure that lets developers build AI applications capable of running seamlessly across multiple GPU and CPU vendors, has raised $250 million in fresh funding at a $1.6 billion valuation. The round was led by U.S. Innovative Technology Fund, underscoring Washington’s growing focus on strengthening domestic AI ecosystems and reducing dependency on single-vendor hardware stacks.

At its core, Modular is tackling one of the most pressing problems in the AI industry: fragmentation. Today’s AI software is often tightly coupled to specific hardware vendors, making applications difficult to port, optimize, or scale across diverse environments. By abstracting these differences away, Modular’s runtime allows developers to write AI once and run it anywhere — whether that’s on Nvidia GPUs, AMD hardware, Intel CPUs, or specialized accelerators. This portability not only lowers costs but also gives enterprises resilience against supply chain chokepoints and geopolitical risks tied to single-vendor reliance.

The $250M raise puts Modular in rare company for infrastructure startups, especially given that it is already being compared to a kind of “AI Rosetta Stone” for hardware. Its backers are betting that interoperability will become critical as the AI industry moves beyond Nvidia’s early dominance and toward a more heterogeneous compute landscape. By enabling developers to future-proof their applications, Modular positions itself as a foundational layer for the coming wave of agentic AI platforms, inference services, and enterprise AI adoption.