Treasury's Scott Bessent Warns AI Race with China Could End in 18 Months

Why It Matters

Bessent’s remarks elevate AI from a technical curiosity to a national security priority, framing the technology as a decisive factor in global power dynamics. By attaching a concrete timeline, he forces policymakers to confront the speed at which AI could reshape economies, defense systems, and critical infrastructure. The focus on computing power and a specific model (Mythos) underscores the growing belief that control over AI hardware and safety tools will determine which nation sets the rules of the digital future. If the United States fails to act within the cited 12‑18 month window, it could cede strategic advantage to China, affecting everything from supply chains to geopolitical influence.

Key Takeaways

• Bessent warned the U.S. lead over China in AI is only three to six months.
• He projected the U.S. share of global AI computing power could reach 70‑80% within a few years.
• Timeline for AI to dominate daily life: "a year, maybe 18 months," per Bloomberg.
• Anthropic’s Mythos model highlighted as a strategic asset for U.S. AI security.
• Comments followed a joint Treasury‑Fed meeting with major banks on AI‑related cyber risk.

Pulse Analysis

Bessent’s alarm reflects a broader shift in U.S. policy circles where AI is no longer a peripheral issue but a core component of economic and security strategy. Historically, technology races—such as the semiconductor battle of the 1980s—have hinged on both raw capacity and the ability to translate that capacity into proprietary applications. The deputy secretary’s focus on computing power mirrors past efforts to secure supply chains for chips, suggesting a similar playbook will be applied to AI.

The emphasis on a narrow temporal lead is a tactical move to accelerate legislative and executive action. In practice, this could translate into increased funding for AI research through agencies like NSF and DARPA, tighter export controls on high‑end GPUs, and incentives for domestic data center expansion. The mention of Anthropic’s Mythos model also hints at a potential public‑private partnership model where the government backs select AI safety tools, creating a controlled ecosystem that can be rapidly deployed for national defense.

From an industry perspective, the warning is a double‑edged sword. Companies that already invest heavily in AI talent and infrastructure stand to benefit from any policy boost, while laggards may face heightened regulatory scrutiny and competitive pressure. The message is clear: the AI race is not a long‑term gamble but a near‑term sprint, and the United States must marshal resources now to avoid a strategic deficit that could reverberate across all sectors of the economy.