Podcast Summary: "China's EUV Manhattan Project and Export Control Mythbusting with Chris McGuire"

Podcast: The AI Policy Podcast
Host: Gregory C. Allen (CSIS, Wadhwani Center for AI and Advanced Technologies)
Guest: Chris McGuire (Council on Foreign Relations, fmr. National Security Council & State Department)
Date: December 22, 2025
Episode Theme:
A mythbusters-style deep dive on the realities and misconceptions surrounding U.S. and allied export controls on advanced semiconductors and equipment, with special focus on China's efforts to develop extreme ultraviolet (EUV) lithography, advances in Chinese AI hardware/software, and the effects, loopholes, and future of export control policy.

Episode Overview

Greg Allen welcomes Chris McGuire, a veteran of U.S. tech policy and national security, for a candid, technically detailed conversation. The episode sets out to separate fact from fiction regarding export controls’ effectiveness, China’s technological progress, and broader U.S.-China AI competition. Special focus is given to the recent Reuters report about China’s prototype EUV lithography machine—a potential “Manhattan Project” for China’s chip ambitions—and how this plays into strategic balance and propaganda efforts.

Key Discussion Points & Insights

1. Guest Background & Policy Trajectory

• McGuire describes shifting from nuclear policy (U.S.-Russia arms control) to emerging tech due to the strategic future of tech competition, especially AI and chips. His role has centered on bridging technical complexity and policy strategy.

  • Quote: "My niche has sort of been being able to explain, understand and explain kind of highly technical topics to policymakers and then seeing the strategic implications of that." [02:40]

• He outlines the role of the National Security Commission on AI (NSCAI) in anticipating policy needs, prefiguring initiatives like the CHIPS Act and hardware-based competition with China.

2. Genesis and Evolution of U.S. Export Controls

• Allen and McGuire chart the path from early Trump-era controls (e.g., EUV) through Biden’s sweeping October 7, 2022 rules and their lineage in NSCAI recommendations.
• The U.S. focus has been on "hardware leverage" as its strongest competitive edge over China, especially in semiconductor manufacturing equipment (SME).

3. China’s EUV Breakthrough? Parsing the Reuters Report

• What’s real: China has pursued a secretive EUV program for years. It’s logical China would attempt to assemble/test an EUV machine, and the story exposes ongoing efforts to evade export controls.

  • Quote (McGuire): "The fact that China can’t make the most advanced chips themselves—which it can't because they can't get the equipment to do it—is the lever that we have." [11:36]
  • Quote (Allen): “I think a more accurate characterization would be assembled an EUV machine... It’s mostly a story of export control evasion and getting around our existing export controls by identifying component suppliers.” [15:05]

• What’s not real / meaningful: The prototype is mostly assembled from foreign (ASML supply chain) parts, not domestically made. It hasn't fabricated any chips, and robust, repeatable commercial production is still years away—likely 2030 or beyond.

  • Quote: "It is very, very difficult to do it well. And then to do it well at scale is even a level beyond that." [16:41]
  • Quote: “If we're in the mid-2030s, we're probably in a decent spot. If we're in the late 2020s, then that’s going to pose pretty significant challenges.” [17:30]
  • Reuters sources say: government targets working chip production by 2028, more likely 2030.

• On Chinese propaganda: Stories like this are part of a coordinated campaign to sap Western policy willpower by claiming controls are ineffective.

  • Quote: "There is a Chinese propaganda campaign around the idea of export controls and trying to push the narrative that export controls don't work... we have seen zero chips come to market in any of these stories." [20:55]

4. Effectiveness of Export Controls (AI Hardware)

• On Huawei’s chips: McGuire’s CFR analysis shows a widening performance gap between Huawei and Nvidia, with Huawei’s pipeline chips getting worse, not better. Even under generous assumptions about production quantity, Huawei will have only 4% of Nvidia’s compute production by 2026—likely less.

  • Quote: “The chip that Huawei is going to release next year, the Ascend 950, has lower processing power and lower memory bandwidth than the best chip that they make today. So they're going backwards.” [27:21]
  • Quote: "Aggregate AI computing power they're making is going to be next year... 4% of what Nvidia is making, and it’s likely lower." [28:38]

• Stockpiling & loopholes: Huawei’s perceived earlier parity derived from TSMC manufacturing violations and stockpiled HBM memory, not domestic breakthroughs. The U.S. has since closed some gaps (e.g., HBM controls), and Chinese AI compute is now structurally falling further behind.

  • Quote: “If you take the TSMC violation out of that data, China’s AI compute really starts to plummet in 23, 24, and then completely bottoms out in 25.” [33:12]

5. Real-World Challenges in Policy Implementation

• Maintenance of deterrence: McGuire contends U.S. policy should have “zero risk tolerance” on advanced SME controls—the foundation must not be compromised.
• Policy lag: Controls have been slow to adapt (e.g., HBM delay), while China circumvents daily.
  • Quote: “The Chinese were not waiting until once a year...they were doing that every single day and we were updating them once a year. That inequity resulted in some pretty big exploitations.” [37:41]

6. Mythbusting: Fact, Fiction, or Middle Ground?

A selection of myths discussed:

Myth 1: Export Controls Just Accelerate Indigenization

• McGuire: Fiction. China’s pedal-to-the-floor indigenization predates U.S. controls by years; marginal impact from controls themselves.
• Allen: Somewhere in-between. Badly crafted controls can accelerate; well-crafted controls can slow China, and regardless China’s indigenization drive will persist.

Myth 2: Export Controls Fatally Harm U.S. and Allied Industry

• McGuire: Somewhere in the middle. Some impact, but strategic value outweighs short-term pain. Markets often reallocate, and data suggest little lasting harm.
• Allen: Fiction (in this instance). Controls have not hindered Nvidia or equipment makers significantly due to global demand, shifting composition, and supply chain effects. Export controls are behind the only two areas (EUV, jet engines) where China’s indigenization has truly failed.

Myth 3: Smuggling Chips into China is a Non-Issue

• Both: Fiction, even "embarrassingly fiction" ([Allen, 73:19]). Major evidence of substantial smuggling and “compute smuggling” via offshore data centers. Malaysia’s outsized imports and DOJ busts underscore the reality.

Myth 4: Chinese Progress on Open-Source AI Proves Controls Don't Work

• Both: Fiction. Progress continues, but only at a fraction of what would be possible absent controls. The key question is the counterfactual: “Where would they be without them?” Both believe controls remain essential for keeping China behind.

Myth 5: Getting China 'Addicted' to Nvidia Chips Maintains U.S. Leverage

• Both: Fiction. China’s willingness to break ecosystem lock-in is unique—the “addiction” argument doesn’t work under their state-driven urgency and resources.

Notable Quotes & Memorable Moments

• On China’s EUV work:
  • "It's the most complicated machine that humans have ever made by a pretty significant margin." [13:50, McGuire]
• On risk to U.S. posture:
  • "If this regime, the equipment regime fails, then our ability to regulate technology globally, constrain China at all in anything, becomes zero." [13:07, McGuire]
• On the damage myth:
  • "There are basically two industries where we have had pretty significant successes holding back Chinese indigenization efforts. One is advanced chip manufacturing and the other is jet engines. And both of those are industries where we have imposed export controls." [62:46, McGuire]
• On data center and “compute smuggling”:
  • "You don't need to ship the chips to China because all you do is you ship the data to China. You're just going to send the files over..." [66:20, McGuire]
• On policy lag like HBM loophole:
  • "We had to move faster than we did because the Chinese were not waiting...and we were updating them once a year." [37:41, Allen paraphrasing]

Policy Recommendations and the Road Ahead

• Biggest Mistake: Returning to a sliding-scale approach on controls. Instead, the line must remain firm, with no allowance for China acquiring even “slightly behind” chips in large volume.
  • Quote: "Moving to an idea where China can have as many of the second or third best chips as they want would dramatically decrease the U.S. advantages in compute." [85:01, McGuire]
• Biggest Gap: Regulating remote access (“compute renting”)—the U.S. lacks any barriers to China leveraging U.S. cloud compute abroad.
  • Quote: "On remote access, we have nothing. It’s completely unregulated. China can just use, you know, US cloud to power anything and everything they want with basically no restrictions." [86:18]

Timestamps for Key Segments

• 10:30–17:00: Dissection of China's EUV Manhattan Project reporting
• 25:38–34:45: Deep dive into Huawei chip competitiveness and real-world AI compute numbers
• 40:45–54:00: Mythbusting: Indigenization, economic harm, and smuggling
• 64:02–73:19: How chip and compute smuggling actually works; rental loopholes
• 85:01–87:27: Mistakes to avoid and policy opportunities ahead

Tone & Language

The conversation is intellectually rigorous, technical, and often wryly candid, with both speakers quick to dispel wishful thinking or policy spin. The mythbuster format ensures honest, straightforward assessments.

• Example, Allen (on smuggling myth):
  "It’s like embarrassingly fiction. I would be embarrassed to have my name associated with the argument that like chip smuggling does not occur." [73:19]

Takeaways for Listeners

• Reports of Chinese breakthroughs in chip manufacturing must be evaluated against the realities of scale, supply chains, and yield—not headlines or propaganda.
• Export controls, while imperfect and with inevitable loopholes, have been critical in slowing Chinese progress and keeping the U.S. and allies ahead in semiconductor and AI competition.
• Policy must now evolve to treat “remote access” and global cloud compute as essential fronts in the tech contest, not just physical exports of chips and equipment.
• The goalpost is not to stop all progress in China, but to maximize the speed and breadth of Western advantage in strategic technology.

For further reading:

• Chris McGuire’s testimony to the House Foreign Affairs Committee
• CFR reports on Huawei and the Chinese chip ecosystem
• CNAS studies on AI chip smuggling

(End of summary)