Donald Trump's visit to Beijing in mid-May 2026 wasn't just diplomatic theater; it was a window into how the US and China are competing for dominance in artificial intelligence through fundamentally different strategies. Trump arrived with an unusual delegation: not just diplomats, but top executives from technology, semiconductors, finance, and aviation, including Elon Musk, Tim Cook, Jensen Huang (Nvidia's CEO), and leaders from Goldman Sachs, Citigroup, Qualcomm, and Intel. This wasn't accidental. The composition of the delegation signaled what the US government sees as the real stakes in the AI race: access to Chinese markets, protection of supply chains, easing of chip export restrictions, and repositioning American capital's influence in China.

What's Really at Stake in the US-China AI Competition?

The competition between the US and China can no longer be explained through tariffs or trade deficits alone. Instead, it rests on three interconnected strategic pillars that form the backbone of AI dominance:

• AI Computing Power: The raw computational capacity needed to train and run large language models, which requires massive data centers and energy infrastructure.
• Advanced Chips: Specialized processors like graphics processing units (GPUs) that power AI systems, which depend on cutting-edge manufacturing and design.
• Rare Earth Elements: Critical minerals and specialty metals used in semiconductors, magnets, turbines, and defense systems, where China holds significant processing advantages.

These three areas are inseparable. AI requires chips to function; chip production needs rare earth elements and advanced manufacturing; and all of this infrastructure demands energy, data centers, engineering talent, and coordinated state planning.

The US still possesses clear advantages in certain areas. American companies dominate the world's most powerful AI systems, control an advanced GPU ecosystem, command large pools of private capital, and operate cutting-edge research labs. A Brookings assessment from April 2026 emphasized that the US maintains overall performance superiority in frontier AI models, while China remains under pressure due to restrictions on accessing advanced chips. By many metrics, American AI models still outperform Chinese competitors.

How Is China Approaching AI Differently Than the US?

But China's strategy operates on a different plane entirely. Rather than viewing AI as a digital service sold by private companies, Beijing treats computing power as a matter of national infrastructure, comparable to electricity grids, railways, and ports. This distinction is fundamental to understanding the modern AI race.

In the US, AI is largely marketed as subscriptions, API (application programming interface) access, private cloud packages, and company-scale efficiency tools. China, by contrast, first establishes the infrastructure, then scales it, and finally enables widespread application across society. The real question, according to the analysis, is not which company produces the flashiest AI demo, but rather who can turn computing power into a public utility.

China released a computing infrastructure action plan in late 2023 aimed at establishing an integrated computing power network by the end of 2025, increasing the capacity of national computing centers, and making this capacity more accessible and affordable. By May 2026, discussions regarding China's national computing power network had advanced further. Beijing began positioning computing power alongside basic infrastructure categories such as water, energy, transportation, and communications. The government is framing computing capacity using tokens as a measurable, priceable, distributable element that can be opened to the mass market, much like mobile data packages. This approach takes AI out of the realm of corporate chatbots and positions it as a fundamental input for production, public services, research, and industry.

A 2025 RAND assessment noted that China's National Integrated Computing Network is attempting to pool public and private data centers, with state funds, local AI labs, and pilot zones supporting this broader ecosystem. This represents a civilization-scale distinction between the two powers: the American model relies on competition among capital markets and private tech giants, while the Chinese model progresses through state planning, infrastructure investment, and national industrial policy.

How Are Chip Restrictions Reshaping the Competition?

The chip issue represents the harshest front of this competition. The US has pursued a strategy of restricting China's access to advanced AI chips to slow its model training capacity. The Trump administration repealed the Biden-era AI Diffusion Rule in May 2025, but simultaneously issued industry warnings regarding Chinese-made advanced processors and the use of American chips in training Chinese models. Even the permission granted at the end of 2025 for Nvidia H200 chips to be sold to China under certain conditions reveals Washington's internal contradiction: the US wants to restrict China while simultaneously preventing American chip giants from completely disengaging from the Chinese market.

China is using this pressure as motivation to strengthen its own semiconductor ecosystem. According to a November 2025 Reuters report cited in the source, China issued a new directive encouraging the use of domestic AI chips in state-backed data centers, opening space for domestic manufacturers like Huawei, Cambricon, MetaX, and Moore Threads. In May 2026, Alibaba's T-Head unit introduced the Zhenwu M890 AI chip, signaling China's determination to reduce dependence on Nvidia. Alibaba also announced a three-year, $53 billion investment plan for AI and cloud infrastructure.

Huawei's recent announcements reinforce this trend. According to a Reuters report from May 25, 2026, despite restrictions on access to advanced manufacturing equipment, Huawei announced a new chip design approach based on improving system-level efficiency and set a goal of developing a high-end chip equivalent to 1.4 nanometer density by 2031. This shows that China is not merely trying to catch up with the same technology, but is seeking alternative engineering paths to overcome restrictions.

Why Are Rare Earth Elements Becoming the Decisive Battleground?

Rare earth elements represent the most decisive front of the struggle. On April 4, 2025, China added products related to samarium, gadolinium, terbium, dysprosium, lutetium, scandium, and yttrium to its export control list. This decision was widely seen as a response to Trump's tariffs and created direct pressure on defense, electronics, turbine blade, magnet, and semiconductor supply chains.

The International Energy Agency (IEA) recorded this regulation as a significant development regarding critical mineral supply security. According to the IEA, the controls implemented by China in April 2025 subject the export of certain medium and heavy rare earth elements to a licensing mechanism. This move is not merely a commercial decision; it is a strategic one. Rare earth elements are fundamental inputs for production areas ranging from defense and electric vehicles to wind turbines, semiconductors, magnets, and aviation engines.

China's decisive advantage lies not so much in mining these elements, but at a far more critical stage: processing and refining capacity. The West's dependence on China is therefore not just about raw material supply, but at the level of high-value intermediate inputs. The European Union's preparation of joint stockpiling mechanisms for tungsten, rare earth elements, and gallium demonstrates that the West sees this dependence as a strategic risk. China, meanwhile, is now openly turning this advantage into a technological bargaining chip.

The composition of Trump's Beijing delegation, the timing of the visit, and the strategic focus on computing infrastructure, chips, and rare earth supply chains all point to a fundamental shift in how the US and China are competing for AI dominance. It is no longer a race defined by which company builds the best chatbot, but rather which nation can control the infrastructure, supply chains, and resources that make advanced AI possible at scale.