Europe Rejects Calls to Slow the AI Race as Mistral Pushes Back Against US Tech Giants

A new divide is emerging in the global artificial-intelligence race — not simply over who can build the most powerful AI, but over whether development should deliberately slow down at all.

European AI companies and officials are pushing back against proposals from leading American AI developers for greater coordination around advanced-model development and safety, arguing that restrictions introduced now could make it even harder for Europe to compete with the United States.

Reuters reported on September 18 that European companies including French AI developer Mistral are challenging calls associated with major US developers such as Anthropic, OpenAI and xAI for stronger coordination and oversight as increasingly capable AI systems emerge.

The disagreement exposes one of the biggest questions facing the technology industry: Can governments make artificial intelligence safer without freezing today's competitive hierarchy in place?

Why Some AI Leaders Want Stronger Safeguards

Concern about increasingly capable AI systems has grown alongside improvements in reasoning, coding, autonomous agents and scientific applications.

The debate is no longer limited to hypothetical superintelligence.

AI systems are already becoming capable of performing longer sequences of work with less human intervention. They can write and execute code, analyse complex research, operate software tools and increasingly participate in scientific workflows.

That has encouraged some leading developers to argue that the industry needs stronger evaluation and safety mechanisms before capabilities advance substantially further.

Anthropic CEO Dario Amodei has repeatedly warned about the speed at which advanced AI capabilities are developing, while US AI companies have discussed mechanisms for coordinating safety standards and providing greater independent oversight.

Europe Sees a Different Risk

European technology companies do not necessarily reject AI safety.

Their concern is who benefits if development slows now.

American companies already occupy many of the strongest positions in frontier artificial intelligence. They have access to enormous computing infrastructure, large research teams and billions of dollars in investment.

Europe has considerably fewer companies operating at that scale.

Mistral is among the continent's most prominent frontier-model developers, but it competes against US companies with significantly larger financial and computing resources.

Some European industry figures therefore argue that introducing restrictive coordination at this stage could preserve the existing gap rather than create genuinely neutral safety rules.

Mistral Has Become Central to Europe's AI Ambitions

Mistral occupies an unusual position in the debate.

The Paris-based company has become one of Europe's most visible attempts to build an independent competitor to American frontier AI companies.

Its significance therefore extends beyond individual models.

For European policymakers concerned about technological sovereignty, having competitive domestic AI companies reduces dependence on technology developed and controlled elsewhere.

That makes any proposal to restrict frontier development strategically sensitive.

If European developers slow down while already trailing the largest American laboratories, catching up could become even more difficult.

Is AI Safety Becoming a Competitive Weapon?

This is where the argument becomes complicated.

Safety proposals can serve a legitimate purpose while simultaneously producing competitive consequences.

Suppose regulators require extremely expensive model evaluations, specialised compliance teams and extensive external audits before an advanced AI system can be released.

A company worth hundreds of billions of dollars may be able to absorb those costs relatively easily.

A smaller startup may not.

The regulation might improve safety while unintentionally creating a larger barrier to entry.

That possibility is why some academics, startups and European industry representatives have raised concerns about what they see as the risk of regulatory capture — rules that appear neutral but disproportionately benefit companies already powerful enough to comply with them.

Europe Is Already Behind in Computing Power

The disagreement comes as Europe confronts another major weakness in the AI race: infrastructure.

Training and operating frontier AI models requires enormous computing resources.

Those computers require data centres, advanced chips, high-capacity electricity networks and enormous amounts of investment.

Europe has struggled to match the scale of infrastructure expansion occurring in the United States and China.

A Reuters commentary this week highlighted Europe's fragmented electricity market and energy constraints as major obstacles to competing effectively in artificial intelligence.

That means Europe's AI problem is not only about algorithms.

It is also about electricity.

AI Data Centres Are Becoming an Economic Issue

The infrastructure behind AI is increasingly visible to ordinary consumers.

Data centres require huge amounts of electricity, and governments are confronting questions about who should pay for the grid upgrades needed to support them.

Spain announced plans this week to raise planned electricity-transmission investment to more than €17 billion through 2030, partly because electrification and large energy users including data centres are increasing demand on the grid.

Similar debates are occurring in the United States.

The AI race is therefore becoming inseparable from energy policy, industrial policy and national infrastructure.

A country cannot become an AI superpower simply by producing talented programmers. It needs enough chips, data centres and electricity to run the systems they build.

The US Still Holds a Major Advantage

The most prominent frontier AI companies remain overwhelmingly American.

OpenAI, Anthropic, Google, Meta and xAI have access to enormous pools of investment and computing infrastructure.

That creates a powerful feedback loop.

Better models attract more customers and investment. More investment pays for additional computing infrastructure. More computing capacity supports development of the next generation of models.

Breaking into that cycle becomes increasingly expensive.

European policymakers therefore worry about reaching a point where dependence on foreign AI technology becomes structurally difficult to reverse.

Europe Wants AI Sovereignty

The phrase AI sovereignty is increasingly appearing in European technology discussions.

It refers to the ability of a country or region to develop, operate and control critical AI infrastructure without depending entirely on foreign companies.

The concept includes models, chips, cloud computing, data centres, energy infrastructure and access to training data.

French and German officials have been particularly vocal about strengthening Europe's ability to develop its own AI ecosystem.

Reuters reports that some European officials view accelerating domestic development as strategically important precisely because the continent currently has so few frontier developers.

But AI Safety Concerns Are Not Imaginary

The European argument does not make the underlying safety problem disappear.

Advanced AI systems are becoming capable of performing tasks that would have seemed unrealistic only a few years ago.

AI-assisted coding can accelerate software development but could also potentially help discover vulnerabilities.

AI biology tools can assist legitimate scientific research while creating questions about access to dangerous biological knowledge.

Autonomous agents could increase productivity while also creating new cybersecurity and reliability risks.

The challenge is therefore genuine.

The disagreement concerns how those risks should be managed and who should write the rules.

Regulation Could Determine Who Wins the AI Race

Technology competition has historically focused on engineering.

Artificial intelligence is different because regulation may become almost as important as technical performance.

Governments can determine what models companies are allowed to deploy, what testing is required, what data can be used and what liabilities developers face when systems cause harm.

Those decisions influence the cost of building AI.

If regulations become extremely expensive to satisfy, large incumbents may gain an advantage.

If regulation remains too weak, dangerous or unreliable systems could reach the market without sufficient safeguards.

Finding the middle ground will be difficult.

Europe Has Already Chosen a More Regulatory Approach

The European Union has generally taken a more interventionist approach to digital regulation than the United States.

European rules covering privacy, online platforms and artificial intelligence place significant responsibilities on technology companies.

Supporters argue those protections are necessary to protect citizens and create trustworthy technology.

Critics argue Europe has sometimes become better at regulating successful technology companies than producing them.

The new disagreement over slowing advanced AI development brings that long-running debate directly into the frontier-model race.

China Adds Another Complication

Any international slowdown would also need to consider China.

Artificial intelligence is increasingly treated as strategically important technology rather than simply another software industry.

Governments view advanced AI as relevant to economic productivity, scientific research, cybersecurity and national security.

That creates a coordination problem.

Even if American and European developers agreed to slow certain forms of development, policymakers would need confidence that competitors elsewhere were following comparable restrictions.

Without that confidence, countries may fear that restraint simply gives a strategic advantage to someone else.

The AI Race Is Becoming a Geopolitical Race

This is why discussions about AI safety increasingly sound like discussions about nuclear technology, semiconductor manufacturing or energy security.

Artificial intelligence has become a strategic asset.

Countries want the economic benefits of automation, but they also want control over the infrastructure producing those benefits.

Europe's resistance to slowing development reflects that reality.

For European companies, the question is not simply whether advanced AI presents risks.

It is whether the rules designed to address those risks leave Europe permanently dependent on American technology.

What Happens Next?

The disagreement is unlikely to disappear.

As AI capabilities improve, pressure for stronger safeguards will probably increase. At the same time, countries and companies that believe they are behind will have powerful incentives to accelerate development.

That creates two competing forces.

Safety pushes towards caution. Competition pushes towards speed.

Europe now finds itself directly between them.

Its policymakers want stronger AI protections, but its technology industry also wants enough freedom and infrastructure to compete with Silicon Valley.

The outcome could determine whether the next generation of artificial intelligence remains concentrated among a handful of American companies — or whether Europe develops a genuine frontier AI industry of its own.