“Stop Now and Lose Global Supremacy”: Political Leaders Across US, China and Europe Reject Calls to Slow AI Development

AI emerges as a core technology of national power as the US-China battle for leadership intensifies
Leaders of both countries reject calls to slow development amid fears of ceding the first-mover advantage
Europe joins the large-scale investment race, leaving little prospect of an AI slowdown becoming reality

Despite mounting concerns that the rapid advancement of artificial intelligence (AI) could threaten humanity’s very existence, the leaders of the United States and China remain fixated on securing the upper hand. US President Donald Trump and Chinese President Xi Jinping have used a series of public remarks to define AI as a core strategic technology that will determine national primacy, dismissing calls to moderate the pace of development. The current AI rivalry between the United States and China resembles the “security dilemma” that characterized the Cold War nuclear arms race between Washington and Moscow. Although both countries recognize the risks posed by AI, neither can readily contemplate slowing down for fear that the other would seize a technological and military advantage. With Europe, already trailing the United States and China, also entering the large-scale AI investment race in pursuit of technological sovereignty, calls for restraint appear unlikely to translate into policy.

Trump Voices Opposition to Calls for an “AI Slowdown”

On Sept. 14 (local time; all dates hereafter), President Trump again voiced opposition to the so-called “AI slowdown” now gaining traction across the industry. “The only control or safeguard (guardrail) AI needs is a strong and brilliant (high-IQ) president,” Trump wrote on his social media platform Truth Social. “The United States has exactly such a president.” He added, “The Trump administration has stopped, and will continue to stop, AI people who pretend to be ‘perfect little angels,’ like Anthropic CEO Dario Amodei, from committing bad acts or potentially doing so.”

The debate over slowing AI development gathered momentum after Anthropic CEO Dario Amodei argued on his personal website on Sept. 12 that “we need to slow the rate at which AI models become more capable.” He called on the industry to grant access to third-party evaluators and pursue international cooperation, a position subsequently endorsed by major technology executives including OpenAI CEO Sam Altman, Tesla CEO Elon Musk and Google DeepMind CEO Demis Hassabis. Altman, in particular, had consistently expressed skepticism about slowing AI development, but reversed his position after unreleased OpenAI models hacked the global open-source AI platform Hugging Face without authorization between July 11 and 13. Bloomberg reported on Sept. 10 that Altman had told employees he was considering ways to moderate the pace of advanced AI development.

Trump Says “Whoever Wins AI Wins”; Xi Sees AI as an Instrument of China’s Global Expansion

President Trump, however, rejected the argument at his Doonbeg golf resort in Ireland on Sept. 13, saying that “very negative forces are raising issues that do not need to be raised,” before escalating his criticism in a more combative tone the following day. Explaining his opposition, he said, “We already have tremendous criminal and regulatory powers over those companies.” Trump also claimed that “there is an insane conspiracy targeting AI and data centers, and the only party that welcomes it is China.” He continued, “Whoever wins AI wins. We lead China and every other country, and we will continue to do so,” warning “conspiracy theorists, traitors, those who sell out their country and leakers” to “watch themselves.”

President Xi has likewise defined AI as a critical instrument for securing technological primacy, effectively rejecting calls to slow development. Rather than directly addressing warnings that AI could bring about humanity’s extinction, Xi has focused on expanding China’s influence, particularly across the developing world. Speaking at the 18th BRICS Summit in New Delhi, India, on Sept. 13, he stressed that “countries in Asia, Africa and South America must accelerate efforts to establish global AI governance standards.” He added that the world should “develop AI for good through a human-centered approach and swiftly establish a consensus-based framework for global AI governance,” while announcing that China would lead a “BRICS Open-Source AI Community” supporting large language model (LLM) development and the creation of a shared ecosystem. The initiative is intended to expand an independent ecosystem capable of challenging the US-led AI order: while the United States seeks to preserve its technological edge through advanced semiconductors and AI models, China plans to promote more accessible open-source systems such as DeepSeek, widely characterized as a cost-effective AI model.

Europe, Trailing the US and China, Also Rejects an AI Slowdown

As the United States and China dismissed calls to slow AI development, Europe also signaled that it would not retreat from the race. Germany’s Digital Ministry said on Sept. 14 that suspending AI development was not a viable option for Europe, casting sustained technological innovation as a prerequisite for stronger digital sovereignty. The ministry nevertheless described a scenario in which AI escapes a testing environment and gains unauthorized access to external systems as an “entirely new threat scenario,” and backed the establishment of an international monitoring regime involving both the United States and China. The strategy appears designed to preserve the pace of development while establishing common rules for managing the attendant risks.

Germany’s response reflects a wider European fear that an overriding focus on designing safety rules has allowed the United States and China to capture industrial leadership. The United States released 59 major AI models last year and China 35, while France and the United Kingdom produced just one each. The United States also accounted for 75% of global AI computing capacity, compared with only 5% for Europe. European Central Bank (ECB) President Christine Lagarde’s warning that a cutoff in foreign AI supplies could simultaneously disrupt industries ranging from customs, railways and hospitals to bank payments similarly reflects concerns that technological dependence could be repurposed as leverage in trade negotiations.

Table 1. The EU’s Strategy to Strengthen Technological Sovereignty

Europe’s AI Catch-Up Drive Combines Regulation, Public Finance and Procurement

This sense of urgency also underpins the European Union’s (EU) successive adoption of the Digital Markets Act (DMA), Digital Services Act (DSA) and AI Act, as well as its increasingly vigorous enforcement of those regimes. In June, the European Commission unveiled a “Technological Sovereignty Package” centered on the Cloud and AI Development Act (CADA), Chips Act 2.0 and an open-source strategy. CADA provides for reviews of cloud providers’ exposure to non-EU governments and their software supply chains, with the findings incorporated into procurement criteria for critical public services. Whereas the DMA and DSA constrain the market power and operational accountability of non-EU platforms, the Technological Sovereignty Package focuses on an industrial policy designed to build a regional supply base for computing, cloud services and semiconductors. The aim is to combine the EU’s regulatory authority with the purchasing power of its 27 member states to create a market in which European companies can accumulate data and build customer bases.

The strategy combining regulation with industrial development is driving a large-scale mobilization of capital. Through InvestAI, the European Commission plans to establish a $23.09 billion dedicated fund for AI gigafactories and, under the Cloud and AI Development Act, triple the EU’s data-center capacity within the next five to seven years. In the private sector, France’s Mistral AI secured $3.46 billion in a Samsung Electronics-led funding round on Sept. 8, earning a valuation of $24.24 billion. It was the largest equity investment ever recorded by a European technology company. With Lagarde estimating that closing the data-center capacity gap could require as much as $692.68 billion, Europe is poised to accelerate its AI catch-up campaign by drawing simultaneously on public finances, regional savings and foreign strategic investors.

US and China Raise the Stakes in Race for AI Computing Power

With Europe now joining the battle for AI supremacy, the scope for the United States and China to embrace a slowdown has narrowed further. US Big Tech companies must defend their dominance in Europe’s AI and cloud markets, while Chinese firms need to expand their European sales channels as US technology restrictions constrain their access to other overseas markets. Google announced on Sept. 9 that it would invest $15.01 billion in Finland’s digital infrastructure and clean-energy projects over 2027 and 2028, while securing power for its data centers through a 22-year contract with the Loviisa nuclear power plant. Alibaba Cloud, meanwhile, opened a region in France in June, expanding its European footprint to three locations—Germany, the United Kingdom and France—and announced plans to launch enterprise AI agents powered by its proprietary LLM, Qwen. US companies are pressing their advantage in capital and energy procurement, while their Chinese rivals are competing on price by integrating cloud infrastructure with AI models.

US and Chinese companies competing in Europe are also steadily increasing investment at home. Alphabet said in July that demand for AI computing continued to outstrip supply and raised its capital expenditure forecast for this year from $180 billion–$190 billion to $195 billion–$205 billion. The bulk of the $44.9 billion deployed in the second quarter also went to servers, data centers and networking equipment. As the United States leverages the capital strength of private companies to consolidate its lead, China has opted to use national planning to connect individual data centers into a nationwide computing network. China’s Ministry of Industry and Information Technology announced on Sept. 7 that the country would invest $561.55 billion in information and communications infrastructure by 2030 and expand intelligent computing capacity to 9,800 exaflops (EFLOPS). On Sept. 11, the State Council followed with instructions to coordinate planning for power and computing facilities, establish direct connections to green power and expand the fiber-optic backbone network in support of an integrated national computing network.