Shield AI raises $2B at $12.7B valuation and acquires simulation leader Echelon Technologies to accelerate AI pilots

Summary

Shield AI raised $2B at a $12.7B valuation and acquired Echelon Technologies, the simulation company behind the U.S. military's Joint Simulation Environment. The two moves target the same problem. Shield AI is one of two companies selected by the U.S. Air Force to build AI pilots for the Collaborative Combat Aircraft program, and Echelon provides the simulation infrastructure to train and validate those systems at scale.

Ryan Tseng, Shield AI's co-founder, frames the core engineering challenge as hitting high performance and high assurance simultaneously. Consumer AI models iterate fast and perform well, but hallucinations are tolerable when the output is an essay. They are not tolerable when the system is distinguishing blue force from red force or keeping an aircraft out of the ground. Shield AI layers assurance frameworks on top of the same underlying model advances coming out of large labs, benefiting from that research across the full stack—from development tools to deployed edge systems—while adding the reliability guarantees military applications require.

The U.S. military operates across an unusually wide spectrum of platforms. Some aircraft are 80 years old; others came off the line recently. The focus for advanced autonomy is primarily on new systems, where integrating software is straightforward, though there is growing interest in retrofitting legacy platforms as well.

Echelon acquisition

The military has long struggled with a rehearsal problem. Training ranges in the U.S. are far smaller than the actual Pacific theater, making it impossible to rehearse the scenarios that matter most. The military's answer was simulation, but early efforts produced fragmented results. Every vendor, platform, domain, and allied nation had its own simulator that couldn't talk to the others.

Echelon broke that pattern. It is the only multinational, multi-domain, multi-vendor simulation environment in existence, covering air, sea, space, and ground with realistic physics and sensor models including radar, night vision, and electro-optical/infrared. Tseng describes it as the authoritative environment for training against high-end threats.

Owning that environment allows Shield AI to close the feedback loop between simulation and deployed AI pilots. It can test autonomous systems against authoritative threat models, generate training data at scale, and prepare human pilots to operate alongside AI wingmen. The goal is building AI pilots capable enough to be integrated into real combat formations alongside U.S. and allied forces, validated in simulation before they ever fly in the field.

Shield AI's selection for the Collaborative Combat Aircraft program and the Echelon acquisition together position the company as the dominant vertical integrator for AI-driven air autonomy, controlling both the pilot software and the simulation environment used to train and certify it.