AnswersPublished15 sources
Why the U.S.–China AI Race Depends on the Rest of the World
Bank of America’s central verdict is that neither the U.S. nor China can secure AI leadership on its own: advanced chips, memory, materials, manufacturing equipment and power infrastructure remain globally distributed.
Published byEdited with GPT-5.6 TerraImages generated with GPT Image 2
Research answer
Key takeaways
- Bank of America’s central verdict is that neither the U.S. nor China can secure AI leadership on its own: advanced chips, memory, materials, manufacturing equipment and power infrastructure remain globally distributed.
- South Korea’s memory industry and Japan’s semiconductor materials and equipment make both countries pivotal to the AI buildout; China’s transformer exports also show why power hardware is becoming a strategic part of...
- The planned September 2026 Trump–Xi meeting could create limited channels for AI safety and trade discussions, but it is unlikely to resolve the deeper conflict over export controls and frontier AI capabilities.
AI Prompt
openai.comCreate a landscape editorial hero image for this Studio Global article: How do Bank of America analysts assess the U.S.–China AI race—including each country’s respective strengths and supply-chain, energy-grid, c. Article summary: Bank of America’s core assessment is that neither the U.S. nor China can “win” AI independently: AI leadership depends on an international stack of electricity, equipment, chips, memory, materials, manufacturing and capi. Topic tags: general, news, general web. Style: premium digital editorial illustration, source-backed research mood, clean composition, high detail, modern web publication hero. Use reference image context only for broad subject, composition, and topical grounding; do not copy the exact image. Avoid: logos, brand marks, copyrighted characters, real person likenesses, fake screenshots, UI text, readable text, watermarks, charts with fake numbers
The competition between the United States and China is often presented as a contest over the best AI models or the fastest accelerators. Bank of America’s Matty Zhao offers a broader—and more practical—reading: neither country can dominate AI without international partners. The decisive stack includes electricity, grid equipment, semiconductor manufacturing, advanced chips, memory, materials and capital, all of which are spread across borders. 8
AI leadership is a supply-chain problem, not just a model race
China has developed relative strengths in power-grid buildout and manufacturing equipment, according to Zhao. But it remains dependent on overseas suppliers for advanced chips, memory and some high-end materials. 8
The U.S., meanwhile, retains major strengths in the frontier AI ecosystem and leading technologies. Its constraint is increasingly physical infrastructure: power availability, grid capacity and the ability to build and connect large data centers. Zhao specifically highlighted U.S. grid limitations as a counterweight to America’s technological edge. 8
That means access to one critical layer—such as advanced AI accelerators—does not deliver control over the entire system. A country also needs memory, industrial equipment, reliable electricity, transmission infrastructure and the materials that feed chipmaking.
South Korea and Japan are essential links
BofA’s argument places key U.S. allies at the center of the AI supply chain rather than on its periphery.
- South Korea is crucial because of its position in memory production, a component required alongside AI processors in data-center systems.
6
8
- Japan supplies high-tech semiconductor materials and manufacturing equipment.
These dependencies complicate any attempt at complete technological separation. The AI race may be led by Washington and Beijing, but its supply chain runs through multiple economies.
Why transformers and grids matter to AI
The growth of AI data centers is making electricity infrastructure a strategic constraint. More computing capacity requires not only chips and servers, but also generation, transmission, substations and grid hardware.
China is becoming a notable supplier of this equipment. Its transformer exports reached 64.6 billion yuan, or about $9.6 billion, in 2025, up roughly 30% year over year, according to reporting on Zhao’s assessment. 6
The figure does not by itself determine AI leadership. It does illustrate BofA’s wider point: as data-center construction expands, the bottleneck can move from GPUs to the physical systems that deliver power. Investors and policymakers assessing AI capacity therefore need to look beyond model releases and chip shipments.
Competition will persist alongside interdependence
The policy conflict remains substantial. Washington restricts China’s access to top-end chips and chipmaking technology, and it has also limited certain U.S. investments in sensitive Chinese AI-related sectors. 2
Those restrictions sit alongside efforts to prevent the rivalry from becoming more dangerous. Ahead of the planned September 23–25 Washington visit by Chinese President Xi Jinping, U.S. and Chinese officials discussed potential AI guardrails and a possible bilateral AI dialogue, including notifications related to shared goals and threats. 1
2
4
The near-term objective appears to be risk management, not an AI détente. Reporting ahead of the September 24 leaders’ meeting indicated that extending or preserving the existing trade truce was a more likely outcome than a broad breakthrough on technology policy. 1
The practical takeaway
BofA’s thesis is not that the U.S.–China AI rivalry is unimportant. It is that the rivalry cannot be understood—or won—at the level of models alone.
The U.S. has a powerful AI ecosystem but infrastructure constraints. China has advantages in grid development and manufacturing equipment but relies on foreign technology at critical semiconductor layers. South Korea and Japan provide indispensable memory, materials and equipment. 6
8
For anyone tracking the AI economy, the most useful lens is the whole stack: compute, memory, materials, equipment, electricity and data-center connectivity. That interdependence is likely to endure even as export controls, trade negotiations and AI-safety talks intensify. 2
8
Studio Global AI
Continue your research
This page includes a source-backed answer you can continue inside Studio Global.
People also ask
What is the short answer to "Why the U.S.–China AI Race Depends on the Rest of the World"?
Bank of America’s central verdict is that neither the U.S. nor China can secure AI leadership on its own: advanced chips, memory, materials, manufacturing equipment and power infrastructure remain globally distributed.
What are the key points to validate first?
Bank of America’s central verdict is that neither the U.S. nor China can secure AI leadership on its own: advanced chips, memory, materials, manufacturing equipment and power infrastructure remain globally distributed. South Korea’s memory industry and Japan’s semiconductor materials and equipment make both countries pivotal to the AI buildout; China’s transformer exports also show why power hardware is becoming a strategic part of...
What should I do next in practice?
The planned September 2026 Trump–Xi meeting could create limited channels for AI safety and trade discussions, but it is unlikely to resolve the deeper conflict over export controls and frontier AI capabilities.