U.S.-China AI Race + Chip Bans Aren’t Working + A Lesson From Nuclear Proliferation | The Spillover

Why It Matters

Superior AI confers a decisive cyber offense‑defense edge, shaping national security and global technology standards. Understanding the limits of export bans informs policymakers on how to balance competition with collaborative risk mitigation.

Key Takeaways

• US chip bans slow but don't stop Chinese AI development.
• China’s models now within months of U.S. performance.
• Compute power, advanced chips remain primary bottleneck.
• Export control loopholes include cloud services and smuggling networks.
• AI safety cooperation needed despite geopolitical rivalry.

Pulse Analysis

The United States has leaned heavily on export restrictions to curb China’s access to cutting‑edge semiconductor technology, hoping to preserve a strategic lead in artificial intelligence. While these measures have introduced friction, the podcast reveals that Chinese firms sidestep barriers through cloud‑based AI platforms, indirect supply chains, and illicit smuggling, eroding the effectiveness of the bans. This reality forces policymakers to reassess whether a hard‑line embargo can meaningfully delay a nation that can rapidly reconstitute its supply network.

Beyond the legal tug‑of‑war, the real constraint on AI advancement lies in raw compute and next‑generation chips. Advanced GPUs and custom AI accelerators are scarce, expensive, and require sophisticated manufacturing ecosystems. China’s recent releases, such as the DeepSeek models, demonstrate that with sufficient compute resources, they can produce capabilities only months behind U.S. offerings. This underscores that the bottleneck is not talent alone but the physical hardware that powers large‑scale model training, a factor that will shape the tempo of the AI race for years.

Strategically, the episode draws a parallel to nuclear proliferation: just as containment without dialogue proved insufficient, AI governance demands a blend of competition and cooperation. The hosts argue that whoever masters superior AI will dominate cyber‑offense and defense, making global AI safety standards a matter of shared security. Engaging China in transparent safety protocols, while maintaining vigilant export policies, could mitigate the risk of an unchecked intelligence explosion and align the two powers toward a more stable technological future.

Original Description

This episode unpacks the evolving U.S.-China AI rivalry, the limits of technological export controls, and what’s really at stake as both countries race to shape the future of intelligence.

Submit Your Question For a Chance to Win a Copy of Sebastian Mallaby’s Book The Infinity Machine at https://cfr.org/giveaway.

Host:

Sebastian Mallaby, Paul A. Volcker Senior Fellow for International Economics, CFR - https://www.cfr.org/experts/sebastian-mallaby

Guest:

Chris McGuire, Senior Fellow for China and Emerging Technologies, CFR - https://www.cfr.org/experts/chris-mcguire

We discuss:

1. How U.S. export controls on chips are slowing China’s AI progress, but not stopping it, as loopholes, smuggling and cloud access weaken enforcement.

2. Why China’s progress is stronger than expected, with competing models only months behind the U.S.

3. As Chris McGuire puts it: “Whoever has the better AI is going to have the offense-defense advantage in the cyber realm.”

4. Why compute and advanced chips are the real bottleneck.

5. Why the “AI intelligence explosion” is overstated, with real-world deployment slowed by infrastructure, regulation, and human constraints.

6. The tension between containing China and working with it on global AI safety and governance.

00:00 - Introduction to The Spillover

00:00:30 - Today’s Topics: AI Geopolitics, US-China Race, & AI Safety

00:01:40 - Chris Maguire: White House Tech Background

00:03:48 - 2022: ChatGPT and Global AI Awareness

00:05:46 - AI Risks in National Security & Cyber

00:06:50 - Chip Export Controls: Strategy & Timing

00:10:45 - 2026 Update: DeepSeek & China’s Progress

00:17:15 - Enforceability of Tech Embargoes

00:20:05 - Policy Shifts in the Trump Administration

00:23:48 - Huawei’s Resilience Under Sanctions

00:29:20 - Recursive Self-Improvement & AI Race

00:35:55 - Global Risks: Rogue States & Terrorism

00:41:40 - Nuclear Arms Control as an AI Model

00:54:10 - Superhexa: AI Eyewear & Industry Tech

Mentioned on the Episode:

Sebastian Mallaby, The Infinity Machine: Demis Hassabis, DeepMind, and the Quest for Superintelligence, Penguin Random House - https://www.penguinrandomhouse.com/books/752231/the-infinity-machine-by-sebastian-mallaby/

Chris McGuire, “The New AI Chip Export Policy to China: Strategically Incoherent and Unenforceable,” CFR.org - https://www.cfr.org/articles/new-ai-chip-export-policy-china-strategically-incoherent-and-unenforceable

Chris McGuire, “Trump’s Reversal on AI Chips is a Historic Blunder,” The Washington Post - https://www.washingtonpost.com/opinions/2025/08/27/trump-nvidia-chips-deal-china/

Want to keep up with The Spillover? Sign up to receive an email alert when new episodes are released at https://cfr.org/newsletters.

The Spillover is a production of the Council on Foreign Relations. The opinions expressed on the show are solely those of the hosts and guests, not of the Council, which takes no institutional positions on matters of policy.