UK Demands Brussels Spare It From 'Made in Europe' Cloud Law
UK Demands Brussels Spare It From 'Made in Europe' Cloud Law
A decade after Brexit, Britain is knocking on Brussels' door asking to be treated as European — this time, for cloud contracts. The European Union's new Cloud and AI Development Act would reserve its most sensitive government data work for EU vendors only, and London is warning that British digital companies will be the collateral damage.
The UK has voiced concerns that the bloc's push for technological autonomy will damage British digital companies, arguing it should be eligible to compete for contracts under the forthcoming "Made in Europe" tech rules, Bloomberg reported on Friday. The story landed at 5:46am London time — before most of Whitehall was awake — and by lunchtime it had reframed the EU–UK summit due in Brussels on November 20.
We read the underlying reporting rather than the aggregator summaries — and the most important thing about this story sits in its sourcing. The debate over limiting CADA contracts to European firms, the UK's claim that excluding it would breach the post-Brexit trade deal, and the Commission's rebuttal of that argument are all attributed to people familiar with the matter. Nobody is saying any of this on the record yet, which tells you this is a negotiation being conducted through leaks — and that both sides consider it too delicate to own publicly.
What Brussels is actually building
The Cloud and AI Development Act, proposed on June 3 as part of the EU's Tech Sovereignty Package, aims to expand Europe's data-centre and cloud capacity and introduces a tiered framework of sovereignty requirements for cloud and AI services — notably in public procurement. That much is official, and documented in techUK's October 9 dispatch from Brussels.
The tiers are the whole story. The framework has four assurance levels, with the first tier applying from February 2028 and the highest tier mandatory by August 2029, according to TechTarget's analysis of the draft. At the top level, European governments would rely exclusively on EU vendors to handle their most sensitive data.
Neil O'Brien, the Conservative MP and former minister who has read the draft closely, spells out what the levels demand in practice: at level 1 a provider must be "established in the Union"; at level 2 its infrastructure, assets and personnel must all sit inside the EU; levels 3 and 4 exclude third-country ownership or influence altogether. Britain, note, is a third country now.
The text is still moving, which is exactly why London is shouting. A non-binding draft opinion from Legal Affairs rapporteur Sergey Lagodinsky of the Greens would cut the four levels to two, reserve at least 80% of new data-centre capacity in "acceleration zones" for EU players, and scrap the route for vetting third-country providers, according to press reports cited by techUK. In the Council, national experts met on October 6 and were reportedly told the proposal is compatible with the EU's WTO commitments — subject to adjustments to its "Buy European" provisions. The lead committees are expected to present their joint draft report on December 1, and techUK says it is preparing its own report on the impact on UK companies. The window to shape this law is open right now.
Brussels' sovereignty ambitions were on full display this month in Dublin, where DeepMind, Nvidia and Mistral pledged AI for science at the RAISE summit — pledges, notably, from two American firms and one European one. The dependence the new law is meant to cure was standing on the stage.
What London is actually asking for
The UK's ask is simple to state and hard to get: eligibility to compete for contracts under rules that were written to prefer Europeans. London's voiced concern, per Bloomberg, is that the autonomy push will damage British digital companies — framed as a warning about harm rather than a demand for special treatment, though it amounts to the same thing.
The objections are specific. According to people familiar with the discussions, the UK side has two: the demand that digital services providers be physically based in the EU, and restrictions on ownership and control. Applied strictly, those two conditions would screen out any British provider with American investors, American infrastructure or an American parent — which is to say, most of them.
The named potential casualties are telling: BT and Nscale. BT is Britain's incumbent telecoms giant; Nscale is the UK's rising AI-cloud champion — the same Nscale whose future sits inside the warnings this month that Britain risks becoming a mere "AI customer" rather than an AI power.
A UK government spokesperson said Britain is "closely engaging with the EU as the Cloud and AI Development Act develops" and will "continue to make the case for close cooperation as a trusted partner", adding that the UK is "already deeply integrated into Europe's digital and technology ecosystem."
There is, in theory, a door. O'Brien points out that the Act's Article 18 provides for accrediting an "associated third country" up to assurance level 3 — the carve-out in legal form. The awkward question, as he frames it, is what Britain would have to agree to in order to walk through it: whether getting accredited means signing up to the same standards, so that Europe's anti-competitive wall simply extends around Britain too.
The argument nobody will make on the record
Behind the polite language sits a genuine legal fight. The UK argues that any move to exclude it breaches the terms of the post-Brexit trade deal governing EU–UK relations; the Commission refutes that reading, according to people familiar with the matter. Expect this one to be litigated in communiqués for years.
London is also irked by the wider pattern: European preference rules are now appearing across legislative files, from cars to public procurement. The tech fight is one front of a broader drift — and Britain, outside the room where the rules are written, can only watch it happen.
And the drift has a price tag attached. Britain is under political pressure to align with Brussels on tariffs against subsidised Chinese car imports in exchange for access to the EU-wide procurement model. France and Germany agreed a car deal last week that opens the door for the UK to qualify for European subsidies under "Made in Europe" rules in the forthcoming Industrial Accelerator Act — but only if Britain mirrors EU rules on foreign direct investment and trade defence, which Bloomberg describes as a thinly veiled reference to measures the bloc has taken against China. Business Secretary Jonathan Reynolds signalled earlier this week that he is open to hitting Chinese electric vehicles with tariffs to win enhanced access to European markets — but, he added, the call would be made on UK needs, not EU demands.
Burnham's other bet: the R-word is back on the table
None of this happens in a political vacuum. Last month, Prime Minister Andy Burnham used his Labour conference speech to declare that "Brexit has done more harm than good" and floated options for Britain's long-term relationship with Europe — including, he told the BBC's Today programme, going "all the way" to rejoining the EU. He said Brexit had failed to deliver the Leave campaign's promise of "taking back control" and had instead left the country weaker, a framing confirmed across Reuters, Bloomberg and South African wire coverage of the speech.
Burnham, who took office in July after Labour MPs ousted Keir Starmer, has promised to lay out the options around the November summit with the bloc. Britain's economic output is 6–8% lower than it would have been had the country stayed in the EU, according to a National Bureau of Economic Research paper cited by Reuters — the kind of number that turns a taboo into a talking point.
The summit itself — November 20 in Brussels, per Bloomberg — has a packed agenda beyond tech: aligning rules on transporting agriculture and food products, linking the two sides' cap-and-trade emissions programmes, and the perennially contentious question of youth mobility. The cloud carve-out now joins that list, and it may be the item with the longest half-life.
Burnham, for his part, has not been shy about intervening in tech markets: he recently torched DeepMind's year-long paid exile clauses for AI talent. A prime minister willing to take on Google over non-competes is unlikely to accept a quiet freeze-out of British cloud firms.
The critic's case: this cuts Europe off from the best AI
Update, Oct. 24: the €115 billion question Brussels didn't price in
Update, Oct. 24: the cost argument against the Act has hardened into a number. A study by the European Centre for International Political Economy — funded by CCIA Europe, which counts Amazon, Google and Meta among its members — puts the 10-year bill for the Cloud and AI Development Act at €115 billion in its middle case, against the European Commission's upper estimate of €18 billion (WebProNews). Direct costs to public budgets would reach €33.3 billion, the study says, with software rewriting accounting for 69% of the total; CCIA Europe's Mitchell Rutledge said the official assessment 'missed 70 percent of CADA's true impact.' The Commission counters that the Act is not law yet — the Council and Parliament still have to finalise it — and says it will bring 'more efficient public procurement and faster infrastructure deployment' (Raconteur). For British firms, Pinsent Masons warns of 'legal uncertainty and supply-chain challenges,' while techUK notes the Act has picked up its first amendments in Parliament with 'Made in Europe firmly on the agenda' for the Nov. 20 EU–UK summit.
The sovereignty argument has a formidable critic in O'Brien, who calls CADA "EU policymaking at its very worst." His core charge is blunt: "My view is that CADA is likely to be extremely bad for European public services, because European governments will effectively be cutting off their access to the world's best AI services."
He reaches for history to make the point. Europe once tried to build its own Google, a project called Quaero — "reader, it did not go well," he writes. Mandating European vendors, in his telling, does not conjure European capability into existence; it just means governments pay more for worse tools. The Commission's own impact assessment concedes the costs — compliance, certification, EU-based infrastructure and staff — would be passed on to customers at higher prices. Protectionism, the assessment even argues, delivers "domestic value added through import substitution." O'Brien's dry response: banning the most technologically advanced tools in human history will supposedly be good for the European economy.
The anxiety behind the law is real, though. In June, Anthropic cut non-US users off two of its models for three weeks to comply with American export controls — European teams simply lost access. And this month, the White House's cyber office asked OpenAI and Anthropic to keep new frontier models away from the UK's AI Safety Institute until Washington finishes its own review. When a foreign government can switch off your AI, "sovereignty" stops being an abstraction.
Which is why the sharpest framing comes from TechTarget's Chris Gilmour, a CTO with two decades in networking, who proposes a simple test: "Could your business keep running its AI if any one model, provider or government changed the terms tomorrow?" His answer — sovereignty as portability, not postcode — is the standard Brussels should be judged against.
Our take: protectionism wearing sovereignty's clothes
Let's call this what it is. The "Made in Europe" rules are a protectionist instinct dressed in security-policy clothing. Nothing in a four-tier accreditation bureaucracy builds a single new data centre or trains a single new frontier model; Europe's great AI hope remains essentially one company, Mistral, and no procurement preference changes that. What the tiers will reliably produce is higher bills for European taxpayers — the Commission admits it — and a slower, worse AI stack inside the governments that most need to modernise.
That does not make the underlying fear foolish. The June cutoff proved that dependence on American providers is a live vulnerability, and Washington's pre-release review demands prove it cuts both ways. But the honest answer to that vulnerability is portability — open weights, multi-provider strategies, funded domestic compute — not a wall with a small, slow, humiliating door marked "associated third country."
Britain's ask is reasonable and should be granted quickly — with reciprocity. The UK is, as its own spokesperson says, deeply integrated into Europe's digital ecosystem; pretending otherwise for procurement purposes is fiction. But London should be careful what it wishes for: walking through the Article 18 door likely means adopting the very standards it is complaining about. Ten years after Brexit, the choice is the same one it has always been — a say in the rules, or an exemption from them. You cannot have both forever.
UK 'Made in Europe' cloud law FAQ
What is the EU's Cloud and AI Development Act?
Proposed on June 3, 2026 as part of the EU's Tech Sovereignty Package, CADA aims to expand European data-centre and cloud capacity and sets tiered "sovereignty requirements" for cloud and AI services — especially in public procurement. It has four assurance levels: the first applies from February 2028, the highest becomes mandatory in August 2029.
Would CADA actually ban British companies from EU contracts?
Not yet — nothing is decided. The EU is debating whether to limit CADA contracts to European companies, and the UK is lobbying for a carve-out. One legal route already exists in the draft: Article 18 allows an "associated third country" to be accredited up to assurance level 3. The fight is over what Britain would have to concede to get that status.
Why does the UK say exclusion would breach the trade deal?
London argues that freezing British firms out of EU procurement breaches the post-Brexit EU–UK trade agreement. The European Commission disagrees. Both positions are reported via people familiar with the discussions — neither side has made the argument on the record.
When do the rules take effect?
On the current draft's timetable, the first tier applies from February 2028 and the top tier becomes mandatory in August 2029. But Parliament and the Council are still amending the text — the lead committees' joint draft report is expected December 1 — so those dates could move.
What's on the agenda at the November 20 EU–UK summit?
According to Bloomberg: the tech carve-out, aligning rules on transporting agriculture and food products, linking the two sides' cap-and-trade emissions programmes, and youth mobility — plus Burnham's promised outline of options for Britain's long-term relationship with the EU.
Sources: Bloomberg (via Bloomberg Law and syndicated wire); techUK's Dispatch from Brussels (Oct 9); TechTarget (Chris Gilmour, Oct 19); Reuters (Burnham's Labour conference speech, Sept 29; the rejoin roadmap, Oct 4); Neil O'Brien's Substack (Oct 19); Sowetan (conference report, Sept 30).
By Daniel Rao. Daniel Rao covers the global AI industry for TechPulse USA, reporting on the models, money and regulation shaping artificial intelligence in the US, Europe and Asia.
Comments
Post a Comment