In this testimony, Dr. Smith argues that the AI revolution is significantly underhyped and poses a profound, misunderstood challenge to national security. He describes the rapid development of AI from simple models to advanced reasoning and planning systems that will soon surpass top-level human experts in fields like mathematics and software programming. The core of his argument centers on the unprecedented demand for energy required to power this growth, with data centers requiring gigawatts of power on the scale of multiple nuclear plants. He urges lawmakers to address this impending energy crisis by supporting an "all of the above" energy strategy and enacting federal measures to accelerate the 18-year average timeline for building new power transmission infrastructure. Finally, he frames the issue as a critical race against China, warning that if China achieves "superintelligence" first—a milestone predicted within a decade—it would fundamentally alter the global balance of power in catastrophic and unpredictable ways, a threat he considers more significant than traditional geopolitical conflicts.
Dr. Smith's Urgent Warning on the AI Revolution and National Security
0:00:11 The AI Revolution is Underhyped: Dr. Smith asserts that the country is unprepared for the profound changes AI will bring, as it is evolving into a new form of intelligence whose implications are not fully understood.
0:00:30 AI's Rapidly Advancing Capabilities: Beyond tools like ChatGPT, AI is becoming a powerful reasoning and planning system.
Important detail: AI programmers are being developed to replace traditional ones, and AI mathematicians as skilled as top graduate students are expected within the next year.
0:01:15 Critical Needs for the AI Industry: To maintain its lead, the US industry urgently requires high-skilled immigration, light-touch regulation focused on cyber and bio-threats, and, most importantly, a massive increase in energy supply.
0:01:27 An Unprecedented Energy Crisis: The demand for computation and energy is the single biggest challenge.
Key takeaway: The industry is planning 10-gigawatt data centers, with each gigawatt being the equivalent output of an average nuclear power plant.
Important detail: The tech industry's energy consumption is projected to grow from 3% to 9% of total US generation, requiring an additional 29 gigawatts of power by 2027 and 67 more by 2030.
0:02:56 Regulatory Hurdles Stifling Growth: It currently takes an average of 18 years to get power transmission infrastructure approved and built. Dr. Smith calls for federal preemption to accelerate this process.
0:03:26 The Imminent Arrival of Superintelligence: The speaker outlines a timeline where AI reaches human-level intelligence within three to four years and "superintelligence" (intelligence greater than human) within a decade.
0:03:48 A National Security Race Against China: It is crucial for America to achieve superintelligence first.
Important detail: Despite US chip restrictions, China is closing the gap rapidly with industrial programs and has already produced competitive models like DeepSeek.
Key takeaway: Dr. Smith is more concerned about China winning the AI race than a potential conflict over Taiwan, stating that a Chinese lead in superintelligence would shift the global balance of power in ways the US cannot predict or counter.