As the US and China build rival AI blocs, Europe is scrambling to reduce its dependence on both. But despite its ambitions to become a technological power in its own right, the EU still lacks the chips, infrastructure and homegrown AI models to compete on equal terms.
With the two leading AI powers, the United States and China, forming their own AI blocs, other countries are manoeuvring to harness the strengths of both groupings and avoid losing out.
ADVERTISEMENT
ADVERTISEMENT
Europe is trying to become a third force in AI, but sovereignty is harder to achieve than simply building a home-grown model. Its prospects depend on whether it can also secure the computing power, chips and infrastructure needed to develop and run AI at scale.
As the rivalry intensifies, experts say the window for countries to be able to choose freely between the AI superpowers may be narrowing, forcing them to at one point pick between the US-led Pax Silica, which the EU is already a part of, or the China-inspired World AI Cooperation Organization. But could Europe eventually offer an alternative?
US and China form separate AI blocs
In the artificial intelligence technology race, the United States and China have formed blocs with other nations seeking to set standards and reap AI's potential benefits.
Europe and many other countries are in a dilemma: Do they go with the US’s most advanced chips and AI technologies, or join China’s more affordable, open-source AI systems?
A European alternative
Views in Europe have changed sharply this year as fears mount over how vulnerable the European Union is to exploitation.
Trump’s forced shutdown of Anthropic’s Fable 5 and Mythos 5 models in June, as well as the more recent switching off of Elon Musk’s military-grade satellite network Starshield in Ukraine, have raised alarm bells in European capitals.
In June, the EU unveiled its European Technological Sovereignty Package, a wide-ranging set of proposals aimed at boosting the continent’s chip production capacity, expanding data centre capacity, supporting sustainable AI, promoting open-source software and integrating data centres into the energy grid to develop sovereign AI for energy management.
The EU has already made progress in boosting chip manufacturing capacity through its Chips Act. This includes a recent €319 million state aid package for Austrian firm Infineon Technologies to develop next-generation semiconductors, as well as €5 billion in Italy and €1.6 billion in Czechia for similar projects.
Europe also has established strengths in the semiconductor supply chain, thanks to firms such as ASML, which produces advanced lithography systems needed to manufacture chips, and Soitec, which supplies semiconductor substrates used to make them.
But the bloc’s ambitions extend beyond chipmaking. The EU is also building a federated cloud and artificial intelligence infrastructure that its architects say could fundamentally change the continent’s relationship with major US and Chinese technology providers.
Rather than building a single European cloud platform from scratch, the project connects existing national infrastructure into a federated network of nodes designed to operate across borders.
Yet Europe remains heavily dependent on foreign technology, particularly in AI. The EU has no dominant frontier AI model comparable to those developed by OpenAI, Anthropic or China’s DeepSeek, and has limited leverage over the companies building the technology. France’s Mistral is one of the few prominent European contenders.
The EU is also trying to rein in its €1 billion-a-day trade deficit with China, driven in part by the bloc’s reliance on Beijing for critical raw materials and electrical components needed for AI infrastructure. Up to 90% of Europe’s rare earth elements come from China, according to Euronews reporting on the cost of reducing that dependence.
Beijing has since agreed to create a “green channel” to streamline and expedite licence approvals for rare earths and permanent magnets destined for the EU, as part of an initial trade deal with the bloc.
Alongside efforts to build up its industrial capacity, Brussels has sought to position Europe as a global rule-setter for artificial intelligence.
In September, European Commission President Ursula von der Leyen vowed that Europe would help write the rules for AI, positioning the bloc as a rule-setter on a par with Washington and Beijing. The EU’s AI Act, adopted in 2024, is billed by Brussels as the “world’s first comprehensive rules on artificial intelligence”.
The Act regulates AI according to a four-tier risk framework, with obligations increasing according to the risks posed by a system. While limited-risk applications such as spam filters face relatively light requirements, certain practices deemed to pose unacceptable risks, including social scoring, are prohibited.
But European companies, concerned that the rules could stifle innovation as they race to catch up with US and Chinese rivals, lobbied hard and secured a delay to rules covering high-risk systems through the AI Omnibus bill, amid a broader shift in the Commission’s political priorities following the 2024 EU elections.
Amid growing global calls for greater control over AI, von der Leyen said the EU’s rules put “Europe in the position to shape global efforts”.
She also said she would invite the main frontier labs “for a discussion on how we can support ongoing industry efforts to pace the frontier”. European firms have been more wary of slowing down, fearing this could be an attempt by US giants to lock in their dominance.
Kazakhstan among countries hoping to harness the best of both blocs
Kazakhstan, a Central Asian nation of about 21 million people, is a rare example of a country that has sought to bridge that gap.
Just three weeks after its AI minister Zhaslan Madiyev visited Washington in June to join the US. group working to strengthen AI supply chains, he headed to Shanghai for a signing ceremony making the Central Asian country a founding member of China’s initiative to shape the global AI agenda.
Being in Pax Silica means more technological partnerships, high-tech manufacturing jobs and a place in global AI supply chains. At the same time, as part of China’s World AI Cooperation Organization alongside 28 other initial founding countries including Pakistan and Russia, Kazakhstan hopes to also have a voice in setting regional and global AI standards.
“Everyone is hedging,” said Dana Masalimova, a former Kazakhstan diplomat and independent scholar and center associate at the Kathryn W. and Shelby Cullom Davis Center for Russian and Eurasian Studies at Harvard University.
“We don’t want this binary choice,” she said.
Kazakhstan is benefiting from both US and Chinese investments, but is in a “very precarious spot,” Masalimova said.
It has rich critical mineral resources essential to AI development and has cheap electricity to power data centers, which have become politically fraught in the US.
Earlier this year, Kazakhstan’s government signed an investment deal worth up to $10 billion involving Nvidia to build a “data center valley” in the country.
Meanwhile, more Chinese companies are investing in Kazakhstan. Local officials say Chinese foreign direct investment in Kazakhstan reached a record $2.8 billion in 2025. Chinese leader Xi Jinping met with Kazakh President Kassym-Jomart Tokayev in July to expand cooperation under China’s global infrastructure building Belt and Road initiative.
Washington’s pressure campaign versus Beijing’s courtship
Both China and the US are leveraging access to AI to increase their global influence.
President Donald Trump insisted last month that “whoever wins AI, wins.” US officials and AI companies have accused China of large-scale “distillation,” or extraction of capabilities, from frontier American models.
Beijing has rejected those claims and says the US is pursuing an AI monopoly while restricting China’s access to some of the world’s most cutting-edge AI chips and chipmaking machinery.
Washington has signalled that Pax Silica — with China as its implicit target — is only for the trusted.
“Countries can’t be trusted partners in Pax Silica while substantively engaging in initiatives that advance a competing vision for super intelligence,” the US State Department said, using the Trump administration’s new preferred term for AI.
For its part, Beijing touts the openness and inclusiveness of its World AI Cooperation Organization. “As both major countries in AI, China and the United States should leverage respective strengths in AI rather than guarding against each other,” the Chinese embassy in Washington said in a statement. “AI should not be owned by major countries.”
The two AI powers have different messaging largely because Beijing needs to court third countries in its bid to build an alternative coalition, whereas Washington already has a powerful network of allies, said Kyle Chan, a fellow at The Brookings Institution’s John L. Thornton China Center.
That lends power to what he described as “swing countries.”
Indonesia, Singapore may also be caught between China and the US
Like Kazakhstan, Indonesia has resources vital for AI and other technologies, such as nickel and copper. It’s also seeking to make the most of both AI powers, having joined China’s World AI Cooperation Organization.
Indonesia wants to benefit from both, Mari Pangestu, the Indonesian National Economic Council’s vice chair, told reporters at a University of Hong Kong Business School event. “We should be able to be part of both, put it that way. And choose the best system.”
US AI firms are bringing investments to other countries like Singapore, where Anthropic is setting up its latest Asia office. The city-state says it is also considering joining China’s AI organization.
In Saudi Arabia, the AI startup Humain unveiled in September a platform built on an open model by a Chinese AI company. The kingdom also has a deal with Nvidia to expand AI computing with its advanced chips.
France, one of the most ambitious AI developers aside from the US and China, is attempting to go its own way with its startup Mistral, whose new model released this month outperforms some advanced Chinese models.
Japan is closely aligned with the US on strategic and security issues, including advanced semiconductors, trusted AI and standards. But it will not reject the more affordable Chinese AI models for “lower-risk applications,” said Takahiro Tsuchiya, a professor at Kyoto’s University of Foreign Studies.
Difficult to challenge the 2 leading powers
For countries which have substantial economic and military capacity but no self-developed powerful frontier AI models of their own, aligning with the US may be vital for their access to data centres and chip deals with US companies, said Anton Leicht and Sam Winter-Levy at the Carnegie Endowment for International Peace.
Essentially, their access to advanced AI systems “depends on the whims of policymakers in Washington and Beijing,” they said, while those countries may have limited influence over AI’s future.
China, meanwhile, has been closing the technology gap with the US when it comes to chipmaking and advanced AI capabilities, but it still trails top US technological capabilities, at least for now.
Also, Beijing’s open-source approach allows countries to be less reliant on the US, and China has been providing funding for developing countries to develop their own AI ecosystems, said Alberto Lemma, a research fellow at the think tank ODI Global.
To date, China has provided official sector grants and loans for projects and activities that use or enable AI technologies in at least 77 low- and middle-income countries, according to the AidData research lab at William & Mary, a public university in Virginia.
Some top recipients include Nigeria, Zambia, Ghana, Bangladesh, Senegal, Cameroon and Kenya.
Public perceptions are also changing. Recent study results from Pew Research Center have shown that people in middle-income countries are becoming more positive and trusting of China when it comes to AI and the technology’s regulation.
All things considered, many countries “don’t want to pick a side ... because there are advantages and disadvantages to each technology stack,” said Aalok Mehta, director of the Wadhwani AI Center at the Center for Strategic and International Studies.
Most countries may eventually lack a good option since they cannot compete on building AI frontier models, but “every country in the world in some way or another is thinking about AI sovereignty,” he said. After all, “maybe no other country than the US and China can really build these very large, powerful models.”
