As the European Union (EU) prepares to respond to Trump’s 25% tariff on aluminium and steel, a widely reported paper from Goldman Sachs has suggested that applying pressure on US technology providers is among the levers the EU could use to negotiate with the US.
Under the Digital Markets Act, the EU Competitions Authority has numerous ongoing investigations related to US big tech, which some experts believe could be deployed as levers to negotiate a more favourable trade agreement with the US. At the same time, the EU has recognised its shortcomings as regards to heavy reliance on US tech and has ambitions to develop sovereign AI capabilities analogous to Trump’s Project Stargate initiative.
Former Google CEO Eric Schmidt told the BBC that there should be government oversight on private tech companies developing AI models, but he warned that over-regulation would stifle innovation. Schmidt also saw US export controls of semiconductors as a way to prevent adversaries getting hold of the technology to develop powerful AI systems.
Given the US’ ongoing trade war with its allies, policymakers are wary of how far will Trump go and whether the US would impose export controls to curb the UK and Europe’s ambitions to develop sovereign cloud and sovereign AI capabilities. In retaliation, would the EU impose tariffs on US tech?
On ING, global head of macro Carsten Brzeski, and senior economist for Germany and global trade Inga Fechner recently co-wrote an article discussing the trade-war, noting that while Europe will try to prepare for an upcoming possible trade war with the US, trade wars are not be won by the trade-surplus country. “It is always the surplus countries that have more to lose. Therefore, Europe might want to consider another route: the strengthening of the domestic economy.
“Think of reducing dependency on the US by increasing domestic military industries, including reducing too many technological standards of weapons systems and pooling of defence purchases, and deregulation of the tech sector, including significant investments,” they said.
EU ramps up AI investment
At the end January, the European Commission presented the Competitiveness Compass, which sets out an industrial strategy for new tech in Europe. The Competitive Compass points to a Europe where future technologies, services and clean products are invented, manufactured and put on the market, while being the first continent to become climate neutral.
The Competitive Compass is based on findings from the Draghi report, from Mario Draghi, the former European Central Bank President. To reignite the EU innovation engine, the European Commission said: “We want to create a habitat for young innovative startups, promote industrial leadership in high-growth sectors based on deep technologies, and promote the diffusion of technologies across established companies and SMEs.”
The strategy will involve so-called AI gigafactories and will apply AI initiatives to drive development and industrial adoption of AI in key sectors. There is also a dedicated EU startup and scaleup strategy to address the obstacles that are preventing new companies from emerging and scaling up.
The EU is proposing a 28th legal regime to simplify applicable rules, including relevant aspects of corporate law, insolvency, labour and tax law, and to reduce the costs of failure. It claims that this will make it possible for innovative companies to benefit from one single set of rules wherever they invest and operate in the Single Market.
This week, European Commission president Ursula von der Leyen launched InvestAI, an initiative to mobilise €200bn for investment in AI, including a new European fund of €20bn for AI gigafactories. She described the public-private partnership, as akin to “CERN for AI”, which would enable scientists and companies of all sizes to develop advanced very large models needed to make Europe an AI continent.
The EU’s InvestAI fund is supporting our future AI gigafactories across the EU. Each will have around 100,000 “last-generation AI chips”. There will inevitably be questions over what exactly is meant by “last-generation AI chips”, but previous generations of AI acceleration hardware are cheaper, and DeepSeek has shown that it is not always necessary to run state-of-the-art technology to achieve comparable results to US AI firms such as OpenAI that do have access to these chips.
