What would have to happen for us to be wrong

Investment theses are almost never written down in a form that could turn out to be false. They are written as directions of travel, which means they can absorb any outcome and be restated afterwards as having been right about something.

This one is narrow enough to be wrong, so here is what being wrong would look like — written now, while admitting it costs nothing.

If buyers keep choosing convenience even when leaving is easy

The thesis assumes buyers will use portability once its cost falls far enough. If migration becomes genuinely cheap and organisations still stay with vendors they dislike, then switching cost was never the binding constraint and the moat was something else — habit, relationships, risk aversion.

How we would see it: portable alternatives reaching feature parity in a category, and market share not moving over several years.

If the regulatory direction reverses

Current European rules push toward portability, transparency and documented obligations. That direction is an assumption, not a law of nature. A serious deregulatory turn, or enforcement that never materialises, removes a large part of the pressure.

How we would see it: portability provisions quietly unenforced, and procurement questions about exit disappearing from tenders rather than spreading.

If open source stops producing competitive products in these categories

The thesis needs sovereign alternatives to be genuinely good, not merely available. If the capital intensity of the next generation of systems means only very large organisations can build competitive versions, then the choice becomes a worse product for more independence, and most buyers will correctly decline.

How we would see it: the gap between open and proprietary widening rather than narrowing over several years in the categories we care about.

If the studio produces worse companies than a fund would

Structural, and the one most within our control. If ventures built in-house consistently stall where founder-led companies push through, the model is wrong regardless of the thesis.

How we would see it: our own companies failing at the commitment threshold rather than at the market one.

If concentration turns out to be efficient

The uncomfortable one. It is possible that a small number of platforms operating most of the world’s computing is simply better — cheaper, more reliable, more secure than the alternative — and that the independence being argued for is a preference somebody pays for rather than a risk being mispriced.

How we would see it: sustained cost and reliability advantages large enough that portable alternatives are only chosen by organisations with regulatory obligations forcing them.

Why publish this

Partly because a thesis nobody can check is marketing. Mostly because writing down the falsifying conditions in advance is the only defence against retrofitting the argument later, and the second year of any thesis is when that temptation starts.

If one of these turns out to be true, the useful thing is to have said so beforehand, so that noticing it does not require also admitting the argument was never testable.