Competing Views of Sovereignty: The Haves vs. The Have Nots in AI (and Steel)

It's interesting to watch how political leaders craft narratives around economic competition that reflect their own country's strengths and weaknesses. For those countries that dominate a particular sector, the concerns of others about sovereignty and security are not that important; for those who are currently losing, however, sovereignty and security are serious considerations. And a country's position on these issues may vary depending on the industry and its particular capabilities.

For example, in steel, where the U.S. industry has been weakened over the years by competition from abroad, President Trump famously said, "If you don't have steel, you don't have a country." And then putting that declaration into practice, the U.S. steel industry has been the beneficiary of Section 232 national security tariffs, and national security was invoked in the case of the purchase by Nippon Steel of U.S. Steel.

In the tech sector (including AI), by contrast, the U.S. is currently dominant and therefore downplays concerns about the sovereignty and security of others. As an example, here's a recent post from Jacob Helberg, the Under Secretary of State for Economic Affairs, arguing that by pushing for sovereignty in the digital arena, other countries will be choosing to leave themselves behind, and that they should rely on products from the more advanced U.S. industry instead:

These days, few words flatter a government like “digital sovereignty.” It carries the music of independence, the dignity of self-rule, the promise that a nation holds its own destiny in its hands. So it is no surprise that the expression has been pressed into the service of a fashionable, fast-spreading policy debate.

... a growing number of governments, persuaded that independence requires duplication, are drafting national AI strategies to match [a UN proposal]—each resolved to rebuild, inside its own borders, a stack that already exists somewhere else.

It is a seductive vision. It is also backward and counterproductive.

... Picture the conference—there is always a conference—where forty governments rise in turn to pledge a sovereign cloud, a sovereign model, a national champion of their very own. They will applaud one another’s independence. Then they will go home and pour billions into companies built to do precisely what thirty-nine others are doing, in markets too small to sustain even one of them, chasing margins that thin asymptotically toward nothing the moment the next champion is announced. They will have achieved not so-called “digital sovereignty” but a kind of synchronized mediocrity—a planet of subscale clones, each heroically reconstructing last year’s breakthrough while the breakthrough itself moves on without them.

While others rebuild the present, American firms will be inventing the future. ...

...

The champions of digital sovereignty believe they are arming their nations for the future. In truth they are marching them, in perfect and well-funded formation, into the past. Digital sovereignty was never a wall, and it was never a copy. It was always a frontier—and the only nations that will be digitally sovereign in the age of intelligence are the ones bold enough to keep pushing it outward, into the territory no one has built yet.

As this shows, the "haves" in a particular sector (in this case, the U.S. in AI) tend to take a position that is, in essence, "don't worry about sovereignty, it's not that big a deal, and you will hurt yourself by trying to achieve it."

By contrast, the "have nots" will say something like "actually, we are very worried about sovereignty in this area, in part due to concerns about security." As a reflection of this view, take a look at this conversation between Soumaya Keynes of the FT and UK Minister for AI and Online Safety Kanishka Narayan:

Soumaya Keynes

Yeah, so I suppose there is this narrative that the UK is particularly good at this application layer when it comes to AI. But there’s also a concern about sovereignty. Because, you know, if what we’re doing is using the AI and applying it well, that potentially leaves us vulnerable to another actor whose name rhymes with Shmerica, shutting off our access to the AI stack, the AI supply chain overall, right?

And so, there was obviously this event a few weeks ago that freaked everyone out when the US restricted access to Anthropic’s frontier models, Mythos and Fable, to non-US citizens. How do you think about the kind of vulnerability that the UK faces there?

Kanishka Narayan

Well, I think that the diagnosis has to be that we are not in a terrific position from a strategic leverage point of view. The frontier models are developed either in the United States or slightly behind them by a few months in China. And so Britain does not have companies when it comes to frontier language models, and we don’t have the core chip architecture, the Nvidia chips, as well as the memory and packaging.

...

And so the nature of the problem is very stark, and I think it’s a serious question for the future of our economic and national security. And so what do you do about it? And there are two ways I think about this. Broadly, what you’re trying to do is have ... chips on the table so that you can access others as critical inputs.

And you can do that either by having chips on the table in the AI stack, or you can also think about it as a broader international negotiation bundle. We might not just offer AI things, we might offer other things that are important to, say, the United States or to China to be able to secure access. On AI, what we’re trying to do is effectively say we think it’s really critical we build more leverage, and that’s why we are focused on things that give us leverage.

For the UK (and others), then, there are important sovereignty and security concerns related to AI – mirroring to some extent the current U.S. concerns about steel – that can't be ignored.

A few decades ago, similar concerns led to a push by many countries around the world to promote a domestic steel industry. Will we see something similar now in AI? Will countries ignore the U.S. downplaying of sovereignty on this issue, and adopt interventionist policies that aim to develop a domestic AI sector?

Finally, it's worth emphasizing that the varying positions that governments adopt on sovereignty are tied to the state of specific industries, and are somewhat predictable and understandable as governments try to promote what they see as in their interest for each sector at a given moment. As mentioned at the outset, sovereignty is good and important if you are weak in a particular sector and need to justify protection from foreign competitors, but it can be dismissed if you are the dominant player. It would be interesting to try to press them on these inconsistencies, although I suspect that in most cases you will not get much of an acknowledgement of this.