Issue No. 01 · The Brief
Brussels spent a decade regulating American clouds. On 3 June it started writing European startups onto the purchase order. The founders building the sovereign supply, and the rounds that prove it.
The news
On 3 June 2026 the European Commission did the thing it has been threatening to do for a decade: it put technological sovereignty into legislation. The package bundles a Cloud and AI Development Act that restricts US hyperscalers from handling sensitive public data, a Chips Act 2.0 aimed at building advanced semiconductor capacity at home, an open-source strategy, and an energy roadmap. The framing was not subtle. “We want to be sure nobody has a kill switch,” Executive Vice-President Henna Virkkunen told reporters.
The reflexive read is that this is Brussels regulating again, a tax on Amazon, Microsoft, and Google dressed up as strategy. It is the read most of the American tech press reached for within the hour, and it is not entirely wrong. Europe has spent fifteen years writing rules for an industry it does not own, and a fourth acronym does not obviously change the ownership.
The more useful read is that a continent that imports more than 80% of its digital products and services just wrote itself a purchase order. CADA does not merely encourage European alternatives. It defines four tiers of cloud sovereignty and requires the highest ones to be structurally independent of third-country law, which is something AWS and Azure cannot offer as they are currently built. That is not a ban. That is demand, with a legal deadline attached. The distinction matters, because demand is the one input European deep tech has always been short of.
Consider what the highest tier actually asks for. EU ownership. EU-national personnel. Immunity from the US Cloud Act, the 2018 statute that lets American authorities reach data held by American companies wherever in the world it physically sits. A hospital network, a finance ministry, a court system: each now has a compliance reason, not merely a patriotic one, to find a supplier that can satisfy those conditions. The private sector is untouched, so nobody is forcing a bank off Azure. But every public buyer on the continent just acquired a checklist that the incumbents fail on structurally, not for lack of effort or spend.
Translation: the EU has converted a political grievance into a procurement spec. Procurement specs are boring. They are also how money moves. A speech about strategic autonomy funds nothing and ages badly. A tiering requirement attached to public healthcare, finance, and judicial workloads funds whoever can meet tier four, and the list of companies that can is short, European, and in several cases already raising at prices that assume this exact tailwind.
Either this is the moment European tech sovereignty stops being a speech and starts being a budget line, or it is the same dependency, relabeled and re-funded at home. Possibly both. Either way, the companies that have to deliver the supply are already raising, and we have been counting them.
What the package actually does
CADA's top sovereignty tiers require EU ownership, EU-national personnel, and immunity from the US Cloud Act. The rules apply to sensitive public-sector workloads in healthcare, finance, and the judiciary. Private-sector cloud is untouched, but every European public buyer now has a compliance reason, not just a flag-waving one, to look local. The incumbents can build EU regions all they like; structural independence from US law is the one box they cannot tick.
The successor act shifts from building fabs to stimulating demand for European-made chips and securing the design capability the continent mostly lacks. The Commission says it will prioritise an advanced foundry; reports put a 3nm facility near €30B. The headline target is €120B of investment by 2035, which is to say a decade out, which is to say this is industrial policy measured in elections rather than quarters.
CADA also sets out to triple EU datacentre capacity within five to seven years, roughly €200B in mostly private investment, so European AI workloads can run on European infrastructure rather than route through US hyperscalers governed by US law. “Mostly private” is the load-bearing phrase: the Commission is naming a number it hopes the market will fund, not one it has appropriated.
Tripling compute means tripling the demand for power, so the package pairs the cloud and chips ambitions with an energy roadmap. It reads like a footnote and is not one. Sovereign infrastructure that cannot be powered, cheaply and on the continent, is a press release. The countries that solve cheap clean energy will host the sovereign datacentres; the rest will keep renting.
What we've already filed
Cloud sovereignty · CADA
Comand AI · €32M Series A · Paris
Deep tech · EIC Fund + NRW.BANK
Semiconductors · imec + Invest-NL
Foundation models · €1.7B Series C
Here is the part the believers should sit with before they uncork anything. A purchase order is not a supply chain. CADA can require tier-four independence, but the GPUs that train the models still come from Nvidia, the lithography that prints the chips comes from ASML and then ships to fabs mostly in Taiwan and Korea, and the €200B to triple datacentre capacity is described, carefully, as “mostly private investment,” which is Brussels for “we are hoping someone else pays.” The chip target runs to 2035. Sovereignty on a ten-year amortisation schedule is still sovereignty, but it is not a switch you flip this quarter, and the people most excited about it tend to talk as if it were.
There is also the protectionism problem, which the doomers will reach for and which is not nothing. A rule that only EU-owned, EU-staffed suppliers may touch sensitive workloads is a moat, and moats protect whoever is already inside them. Done badly, CADA funds a generation of mediocre national champions who win contracts because they are local rather than because they are good, and the continent ends up paying more for worse infrastructure in the name of owning it. That failure mode has a long and well-documented European pedigree, and pretending otherwise is how you get burned.
Two things are true at once. The policy is real demand, and the demand can be wasted. What decides which way it breaks is not the legislation, which can only ever write the cheque. It is whether the companies on the receiving end are any good. And this is the part where, the doom budget duly spent, we get to be believers, because the evidence is annoyingly specific and it predates the press conference.
eleQtron is building trapped-ion quantum computers in Siegen, not Silicon Valley, on €57M from the EIC Fund and NRW.BANK. Eyeo raised €40M, backed by imec and Invest-NL, to rebuild the camera sensor from the light up. Comand AI took €32M to put the command layer of defense software on a laptop a French officer can actually own. Mistral closed a €1.7B Series C led by ASML, the most European cap table you could draw up, at an €11.7B valuation. None of these is a press release waiting for a subsidy. They are companies that were already solving the supply problem before the Commission got around to naming it.
Trapped ions in Siegen, a sensor rebuilt from the photon, a defense stack on a French laptop, a foundation model funded by a lithography giant. A decade of speeches, and then a purchase order. The doomers will want the contracts signed first. The contracts are being written.
Filed by eu-tailwind Exclusive. Independent reporting on the rise of European tech.
Back to The Brief →Read by founders, investors, and operators across Europe
One considered email a week: the funding, the founders, and the arguments worth reading. No noise.
See a recent dispatchFree. One email a week. Unsubscribe anytime, we never share your address.