Sovereignty — hard to define, harder to pull off (Analyst Angle)

What is sovereignty? Is the purpose of a “Sovereign Cloud” to house the assets locally?

I saw today that Amazon launched its “AWS European Sovereign Cloud,” with its first region in Brandenburg, Germany. I had to chuckle at the absurdity of the focus on local staffing, as if a local technician’s paycheck is the most important factor. The software is controlled in the U.S., and the substantial profits are going back to an American company.

All joking aside, this story is repeating all over the tech markets. Nobody wants China to have a “kill switch” on its train or bus service. Brazil doesn’t want to use an American satellite network. Almost every country is wrestling with the concept of Sovereignty and how they should participate in 21^st century technology.

Every world conflict highlights the importance of tech sovereignty. In Ukraine, Iran, Brazil, and the U.S., we see companies that can defy the government and provide communications services.

What is sovereignty? Is the purpose of a “Sovereign Cloud” to house the assets locally? To own the assets? To create jobs within a country’s borders? To control the software stack?  Or to control where the data can go? Everybody has a different idea of what this means.

Of course, the problem for most countries is that a small economy cannot launch its own LEO satellite network or develop its own gen AI model. These are billion-dollar or trillion-dollar projects that naturally will be dominated by the players with the deepest pockets.   

We’ve already been living in a globalized tech world for the past 80 years. Since the end of World War II, the Americans have dominated the tech markets. During the 1970s through 2000s, in fact, there was a clear trend toward American ownership but with outsourced manufacturing to Japan, China, Taiwan, and elsewhere. You could say that for 40 years the pendulum swung away from “Sovereignty” toward “Globalism,” and the Americans took down our trade barriers to enjoy the benefits of cheaper labor and the talent of other countries.   

Not all countries did the same. Europe protected its mobile infrastructure vendors, so when Lucent, Nortel, and Motorola disappeared, Nokia and Ericsson survived. China is even more obvious, with companies like Huawei and ZTE living under the legal shelter of a friendly government and even receiving cash subsidies.

Let’s look at some specific challenges facing Tech Sovereignty:

1.  Cloud Computing: The success of cloud computing derives from its economies of scale. A big data center, in a location with cheap power and fiber access, can develop a standard server configuration and software stack that is highly reliable and predictable. Imagine a world where every country developed its own computing stack… no developer would ever spend time on the custom stack for Guatemala. Only highly coordinated and highly developed countries can develop their own stack and a menu of applications that use it.   Instead, we see many countries developing national data centers as a sovereign form of Edge Computing.
2. Gen AI models: This takes the Cloud Computing challenge to the next level, because the investment in a useful gen AI model is hundreds of billions of dollars. Even if a small country could develop its own, could it hire the gen AI experts needed? In fact, Apple recently announced that it will use Google’s gen AI models. The big models are so crushingly expensive that even a $4T company decided to buy from its competitor.
3. Networks: Huawei and ZTE have famously been excluded from many countries. It’s easy to imagine a “back door”, hidden in network software with millions of lines of code. Some countries plan to use locally manufactured radios, along with software from Ericsson or Samsung, in an Open RAN configuration. That’s appreciated by the local supplier and their employees, but outside of Sweden and Korea, it really doesn’t solve the issue of sovereignty.
4. Satellites: We’re tracking 14 major satellite constellations that plan to participate in D2D services. Of course, not all of these can succeed. By their nature, LEO satellites travel around the entire globe, so any business model works best as a global business, not a sovereign business. The IRIS2 consortium approach that’s coming together in Europe is simply a gesture toward sovereignty, but it will not achieve any true sovereign control for any nation.

In the end, only China and the United States have any hope of true tech sovereignty. The scale of investment needed to control the entire stack is mind-boggling.  China is working hard to build its own networks, develop its own cloud computing and AI models, and launch its own satellites. They’re still addicted to Western semiconductors for high-performance applications, but we expect that they will solve that challenge in the next five years. In China’s case, the CCP finances the whole show, so its tech sovereignty is tied to the financial health of the entire country as a whole.  

In the United States, investments are made by companies in a Darwinian crucible — bad ideas get flushed, and good ideas thrive. This has led to American control of software layers, hardware development, satellite networks, and the design of semiconductors. We have given away control of our wide-area network software and semiconductor manufacturing, but recent steps are moving the USA back toward more control again.

There’s a rising discussion of “orbital data centers” going on in some chat groups. I have interviewed a few of these companies and listened to their arguments. On the technical merits, I’m not convinced.  But one aspect intrigues me:  Orbital data centers, with direct satellite communications to customers, can operate independently of all Earth governments. Space is considered to be “international waters” from a legal point of view.  We often focus on the cost of electricity, semiconductors, and code. None of those things gets cheaper in space. But the cost of lawyers, negotiating a labyrinth of sovereignty rules in 200 different countries? I wonder how much money would be saved by simply ignoring all government rules about AI/cloud computing — firing all the lawyers.

Imagine that you’re a trillionaire, living on Mars, with a satellite network, orbital data centers, and a social media empire providing information to every human on Earth. And nobody can tax you, sue you, or tell you what to say.