THE LONG VIEW

INTELLIGENCE BRIEFING: The Rise of Watt’s Law and the Strategic Energy Ceiling on American AI Dominance Executive Summary: A fundamental shift is redefining the trajectory of American technological leadership: the decline of Moore’s Law and the rise of 'Watt’s Law,' where AI capability is increasingly constrained by power availability and system-wide efficiency rather than transistor density. Modern AI systems operate like industrial plants, consuming megawatts of electricity and demanding sustained throughput over peak performance. This new reality positions energy infrastructure as a critical determinant of national competitiveness. Without urgent policy alignment between AI expansion and power delivery—spanning grid resilience, permitting reform, and strategic partnerships—the U.S. risks ceding long-term advantage to rivals capable of mobilizing energy and compute at scale. The window to act is narrow, and the stakes extend across economic productivity, national security, and global influence. Primary Indicators: - Energy utilization has replaced transistor scaling as the primary constraint on AI progress - AI workloads require continuous, high-power operation akin to industrial processes - system efficiency—measured in tokens per joule—is now the key metric of real-world AI capability - ASML and other core semiconductor players warn that energy use could cap AI training capacity - Aurora supercomputer consumes ~38.7 megawatts, equivalent to 30,000–40,000 homes - FLOP/s per dollar doubles only every ~2.46 years, signaling slowing returns from hardware alone Recommended Actions: - Establish an all-of-government emergency task force to integrate AI and power infrastructure planning - direct PCAST to issue annual assessments on national energy-to-AI capacity progress - create national test beds for AI deployment under real-world power and cooling constraints - treat export controls as time-buying measures, not standalone strategies - prioritize partnerships with energy-rich allies to co-develop AI infrastructure Risk Assessment: The United States stands at the edge of a silent crisis—one not of code, but of current. Should the nation fail to recognize that the next phase of innovation will be powered not by nanometers but by kilowatt-hours, it will watch its technological supremacy erode in plain sight. Adversaries do not wait for readiness; they exploit inertia. A future where China or another power leverages scale, state-directed energy access, and full-stack optimization to surpass U.S. AI output is not speculative—it is mathematically inevitable under current trajectories. The ceiling is not technological. It is organizational. And the cost of delay will be measured in lost decades of productivity, weakened defense capabilities, and a diminished role in shaping the global digital order. —Sir Edward Pemberton