- A U.S. order forcing Anthropic to suspend foreign access to its Fable 5 and Mythos 5 models validated long-standing European fears
- European and U.K. officials seized on the move as proof that sovereign AI needs to be a top priority
- The fallout could create a major opening for telcos outside the U.S. to support sovereign ambitions
Turns out the folks in Europe worried about an overdependence on foreign tech companies and the prospect of a critical infrastructure kill switch weren’t overreacting. The U.S. just proved them right, though the infrastructure in question wasn’t cloud access but rather the ability to tap one of world’s most advanced AI models.
On June 12, Anthropic revealed the U.S. government ordered it to “suspend all access to fable 5 and Mythos 5 by any foreign national,” effectively cutting the globe off from both frontier models. The government citied evidence of a jailbreak capability as the reason for its request.
Anthropic publicly disagreed with the government’s directive, stating “we disagree that the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people. If this standard was applied across the industry, we believe it would essentially halt all new model deployments for all frontier model providers.”
The thing is, no matter how fast Fable 5 and Mythos 5 come back online (which some have speculated will happen by July 1), the cat’s out of the bag. There’s now demonstrable evidence that the U.S. government can order companies to revoke access to AI models – and those companies will comply.
Bursting a different AI bubble
The whole scenario is the realization of fears that just weeks earlier had led to the introduction of the EU’s Cloud and AI Development Act (CADA) in an attempt to claw back some control over critical infrastructure. And, of course, the U.S. decision didn’t go over well with European officials.
“The issue here is not whether the US is right to take national security concerns seriously. Of course it is. The issue is that Europe cannot keep building its tech stack on access that can be switched off overnight by a foreign government,” Aura Salla, a Member of the European Parliament (MEP) from Finland, wrote on LinkedIn. “There have been enough wake-up calls like this for Europe. Now it’s high time for action to make European tech sovereignty a reality.”
German MEP Sergey Lagodinsky called the U.S. order “discriminating and ineffective” and urged so-called middle powers like Canada, India, Singapore and Australia to form an alliance similar to Project Tapestry to further the training of new frontier foundation models.
U.K. officials, though no longer part of the EU, sang the same tune.
“We treat every other threat to our sovereignty with deadly seriousness, but we haven’t learned to treat this one in the same way,” U.K. Minister for AI and Online Safety Kanishka Narayan said in a LinkedIn post. “I care about sovereign AI because it now decides our security…The most pressing, the most significant question for Britain’s national security is the future of our AI.”
Fierce’s take on what’s next
As Arthur D. Little Senior Advisor Klaus Schmitz put it, “This is a turning point. It shows that technological and economic sovereignty must be redefined.” He noted, however, that ending reliance on foreign technology has the potential to be a “massive challenge for Europe and for every company depending on global AI innovation.”
But it also presents a massive opportunity for telcos outside the U.S.
Countries pursuing digital sovereignty (that is, across both cloud and AI) will be looking for alternatives to hyperscaler infrastructure. While they’re not necessarily experts in running data centers at hyperscale volumes, telcos already know what it’s like to manage regulated national infrastructure. They’re also already in possession of the networks that would be necessary to connect data center facilities and customers. If they're smart, they'll lean into their status as trusted providers to capitalize on the moment.
The biggest issues – as in the U.S. – will be securing land and access to power.
Some operators, including Germany’s Deutsche Telekom and Korea’s SK Telecom, are already blazing this trail. Both are looking to build gigawatt-scale AI factories geared toward sovereign compute.
The biggest hurdle on this front will perhaps be capex and, more specifically, figuring out how to balance spending for network upgrades (like the 6G upgrades coming just a few years down the line) and on compute facilities that can take years to build. But with foreign governments stepping up with funding and private capital waiting in the wings, solutions are in sight.
Then there are the AI models. Telecoms may not be the best positioned to build frontier models but given their extensive enterprise relationships, they can offer one heck of a distribution channel for up-and-coming model makers like Alibaba, Mistral AI, LG AI Research, Cohere and others.
Then again, (and since we’re not talking about the U.S.) maybe some entrepreneurial telcos will opt to train up their own versions of DeepSeek or Kimi AI’s open-source models to offer directly to businesses and consumers.
However things play out, one thing is certain: the era of “what if” is over. Sovereign efforts are now a top priority for countries outside the U.S. and a huge door has definitively opened for telcos. The question is whether they’ll step through before someone else does.