Digital sovereignty
In shortNot self-sufficiency, but the ability to decide: you can leave if you have to.
Digital sovereignty is routinely confused with independence. That is the most expensive mistake in the field. No company and no state builds the whole chain from grain of sand to language model. Not even the ones that say they do.
Sovereignty means you can decide. You know what you depend on, you know the price of leaving, and you could pay it if it came to that. Whether you actually leave is a separate question.
Three levels that get mixed up
Data sovereignty — you determine who sees, copies and analyses your data. That is not settled by where the server stands, but by who holds the keys and which law the vendor answers to. See Data residency and data sovereignty.
Technological sovereignty — you can understand, inspect and, if necessary, keep running the technology yourself. Open source and open formats are the means here, not the end.
Operational sovereignty — you can keep operating when a vendor fails, terminates, or is legally compelled to cut you off. This is the level that counts in an emergency, and the one least often tested.
The practical test
A usable test is one question. If your largest vendor locked you out tomorrow morning — not out of malice, but because a sanctions list, a payment problem or a government order got in the way — how long would you be unable to operate?
Under a day: sovereign. Under a week: in a position to negotiate. Longer than a quarter: you are not a customer, you are a component.
Why this matters now
The debate used to be academic. It stopped being academic once sanctions, export controls and tariff policy demonstrated that software supply chains can be switched off politically. Contracts and assurances hold for exactly as long as the vendor's government lets them hold.
That is not a reason to panic, and it is not an argument against American technology. It is an argument for knowing the price of your dependency before somebody else sets it.