The U2 spy plane's success wasn't just the aircraft; it required new acquisition methods, a new intelligence analysis center (NPIC), and a direct line to the president. Similarly, AI requires a full ecosystem overhaul, not just API access to models.
The distinction between "mission systems" and "business systems" is obsolete. To support warfighters, the Pentagon must modernize its core logistics, personnel, and finance systems to provide data at an operational tempo—a task they were never designed for.
The U2 project created a specific, new data source that agencies could build processes around. In contrast, AI presents an overwhelming flood of data and externally developed tools, making integration and focus—not capability creation—the primary challenge for government.
Without market competition, government agencies must create their own forcing functions. This means using personnel policy to reward risk-takers with unconventional careers, while actively stifling the careers of leaders who fail to adapt and move quickly.
Since AI is developing commercially first, the government has a second-mover advantage. It can learn from private sector blunders, like uncontrolled spending on AI tokens, and act as a powerful buyer to negotiate better terms, rather than leading development itself.
Equipping soldiers with GPUs is key, but innovation will be stifled by bureaucracy. Leadership must change property accountability policies, explicitly accepting the financial risk of lost high-value equipment to encourage real-world use and experimentation.
The primary cyber threat from AI is not new attack vectors, but the speed at which models can discover existing vulnerabilities. The only effective defense is to use AI to find, prioritize, and patch these flaws faster than adversaries can exploit them.
Contrary to the belief that a draft pulls in top talent, the modern volunteer military and civil service are, on average, more educated than the general population. This results in a smaller but surprisingly skilled and professional force capable of leveraging complex technology.
Government contracting often pays for billable hours, creating a perverse incentive against using AI, which allows smaller teams to deliver results faster. This "status quo bias" in procurement actively discourages the adoption of productivity-enhancing technology.