Pentagon Scrambles to Fix Compute Crunch Threatening Military AI Rollout

The Pentagon is sounding alarms over a critical bottleneck that could stall its ambitious plans to embed artificial intelligence throughout the military. Defense Secretary Pete Hegseth, Pentagon CTO Emil Michael, and other top officials are pushing hard to accelerate AI adoption, but the department’s compute capacity is struggling to keep pace with demand.

Chief Digital and AI Officer Cameron Stanley laid out the problem starkly at the recent AI+ Expo. He highlighted how AI tools like Palantir’s Maven Smart System played a pivotal role in Iran war operations, enabling strike missions on 13,000 targets in just over a month. The system drove a fourfold increase in network utilization and processed nearly 900 million tokens daily to synthesize battlefield data and speed decision-making.

Yet Stanley’s biggest worry isn’t enemy forces — it’s whether the Pentagon can supply enough “high-octane fuel” in the form of computing power to sustain the AI revolution. “My biggest fear is, can we keep up with the insatiable appetite of American warfighters to accept and leverage technology?” he said. The demand for compute resources is exploding at every level, from frontline operators to senior commanders.

The department has inked deals with tech giants including SpaceX, OpenAI, Google, NVIDIA, Microsoft, Amazon Web Services, and Oracle to bring their AI capabilities onto classified networks. But bureaucratic hurdles and limited compute capacity remain significant barriers to scaling AI tools.

“We’ve handed our warfighters a Ferrari,” Stanley said. “My only sleepless nights come from making sure we never, ever run out of the high-octane fuel that they need, which is compute.” The Pentagon promises announcements soon on how it will expand compute capacity across all domains and classification levels to meet demand.

This compute crunch exposes a broader risk in the military’s AI push: technology alone won’t guarantee battlefield dominance if infrastructure can’t keep pace. As the Pentagon races to harness AI’s promise, it must also confront the hard limits of its own digital backbone — or risk falling behind adversaries who do.