Fridays, 2.30 pm (Lisbon) · Artificial intelligence on NOW →Fridays, 2.30 pm · AI on NOW →
Miguel Sousa SoaresMSS
← Writing
LinkedIn · Essay · 16 June 2026

The Washington Switch

The switch exists, it sits outside Europe and it can be thrown by political decision. No vital sovereignty should depend on a switch on another continent.

It was 5:21 p.m. on a Friday in Washington. One letter was all it took.

In that letter, the US government ordered Anthropic to suspend access to its two most advanced models for any foreign national, inside or outside US territory. To comply, the company switched everything off for everyone, everywhere in the world.

This was not a commercial cutoff. It was an export-control directive, an instrument normally reserved for weapons, nuclear technology and satellites. As far as is known, this is the first time that framework has been applied to access to a language model. Nor did it distinguish adversaries from allies. Britons, Canadians, Australians and Europeans all found themselves on the same side of the border: the outside.

It is worth understanding how we got here. For years, part of the industry and its advocates presented these systems as strategic assets: too powerful to circulate without safeguards, capable of functioning as dual-use technology. It was marketing, and it was conviction, and it worked. The problem is that the state listened. It took the language literally. If you call AI a weapon, you cannot be surprised when the state keeps it in an arsenal. The directive did not come out of nowhere. It was the logical conclusion of a narrative the industry helped write about itself.

And this is where everything changes. Dependence is no longer merely a weakness to be fixed with better contracts. It becomes a political condition. Once AI is a strategic asset, it answers to a flag, not a market. The idea of neutral, global infrastructure was a commercial fiction from the start: convenient for those selling it, invisible to those buying it.

Portugal risks becoming a digital trading post: it powers other people’s AI with energy, data and demand, and receives in return access it does not control. A few weeks ago, this was a thesis. This Friday, it became a precedent. Even if access is restored tomorrow, the lesson will not disappear. The switch exists, it sits outside Europe and it can be thrown by political decision.

And this is not even about trusting whoever happens to govern in Washington today. Administrations change; the architecture of dependence remains. Whoever hands access to a single decision-making centre hands it to everyone who will sit there in the future. The risk is not who signs the letter. The risk is that one letter is enough.

We are building public services, banks, hospitals, courts and newsrooms on AI infrastructure that rents out capacity and revokes access according to a foreign policy we do not vote on. We call it the cloud. It is territory. Someone else’s territory. Europe talks about strategic autonomy and still treats AI like electricity, something that will always come through the socket. It does not always. It stopped coming at 5:21 p.m. on a Friday.

The answer is neither naïve protectionism nor the illusion that we can copy the American frontier within a decade. It is more modest and more serious: decide that vital functions must never depend on rented access. Keep our own capacity where the risk is existential. Test continuity. Know what happens when the socket stops supplying power.

Portugal and Europe will not build the next frontier model tomorrow. But they can decide today what to do when access disappears. No vital sovereignty should depend on a switch on another continent. It is time to stop confusing convenience with sovereignty.

Friday’s letter was addressed to a company. The warning was meant for us.

Translation of the text in the author’s archive. Original publication: O Interruptor de Washington · LinkedIn · Essay · 16 June 2026.
Original bibliographic details (Portuguese)

Ensaio de Miguel Sousa Soares, publicado originalmente no LinkedIn a 16 de junho de 2026. Esta página é o arquivo pessoal do autor.

Miguel Sousa Soares is an economist and analyst of geoeconomics and artificial intelligence, with more than twenty years’ experience in strategy and corporate finance. He writes for Visão, Jornal de Notícias, Expresso and Correio da Manhã. He comments on artificial intelligence on NOW on Fridays.