TL;DR
A Thorsten Meyer AI analysis argues that Europe’s digital policy focus has produced visible interfaces such as cookie banners while the leading AI models, compute networks and funding pools developed elsewhere. The piece says Brussels is now trying to respond through InvestAI, but claims the plan does not yet match the scale of U.S. and Chinese AI investment.
Europe is trying to regain influence in artificial intelligence through new public funding and digital policy changes, but a new Thorsten Meyer AI analysis says the bloc still lacks the frontier labs, compute scale, energy base and private capital needed to compete with the United States and China.
The analysis points to Brussels’ InvestAI plan, described in the source material as a €200 billion effort made up of €50 billion in public money and €150 billion in hoped-for private funding. It says €20 billion has been set aside for AI gigafactories, with EU funds covering no more than 17% and operational compute expected in 2027 or 2028.
The report contrasts that target with claims that four major U.S. hyperscalers are set to spend about $700 billion in capital expenditure in 2026 alone, with Amazon and Microsoft each near $200 billion. It also cites Stargate, a U.S. AI infrastructure project, at $500 billion. Those figures are presented in the source as scale comparisons, not as independent verification by this article.
On model capability, the analysis identifies Mistral as Europe’s only major lab in the frontier large language model discussion, while saying the firm remains behind leading U.S. systems and low-cost Chinese open-weight competitors. It says Europe has no entrant in the most restricted frontier tier described in the source material as export-controlled and geopolitically sensitive.
Europe regulated the interface and forgot the engine
The cookie banner is the most-used European software of the decade. While Brussels perfected the consent pop-up, the frontier was built elsewhere — and now, in H2 2026, Europe wants to buy back in without changing what put it on the outside.
This isn’t about whether privacy or safety matter — they do. It’s that Europe mistook regulating the interface for having a seat at the table. You can’t grant your way out of a structural problem while keeping the structure — the laws, the capital gaps, the energy costs, the talent drain all left untouched. The fix isn’t another framework: it’s open weights as a product, sovereign compute on affordable power, real capital plumbing — and to stop mistaking a check for a strategy.
Europe’s AI Leverage Problem
The argument matters because AI infrastructure is becoming a source of economic power, security policy and industrial dependence. If European companies, governments and public services rely on non-European models, clouds and chips, Europe may have less control over pricing, data flows, availability and technical standards.
The source material says the EU spends about €264 billion a year importing non-EU digital products, relies on non-EU digital stack components for more than 80% of its needs, and has about 70% of its cloud market held by AWS, Google and Microsoft. Those figures frame the AI debate as part of a wider digital dependence issue rather than a narrow model race.
The report’s central claim is that regulation can shape markets only when paired with production capacity. Its interpretation is that Europe has built strong rule-making institutions but has not matched them with enough compute, cheap power, risk capital or talent retention to shape the underlying technology.
SQL Server 2025 Unveiled: The AI-Ready Enterprise Database with Microsoft Fabric Integration
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The analysis uses cookie banners as the emblem of Europe’s digital policy record. It cites Legiscope, a consent-management vendor, estimating that EU users collectively spend about 575 million hours a year dismissing cookie banners. The source itself flags that estimate as a back-of-the-envelope figure from an interested party, not a settled measurement.
The report also says studies of real-world consent banners have found widespread compliance problems, citing one analysis of roughly 400 banners that found about 89% breached rules through dark patterns, vague purposes or related issues. It says the legal trigger for many cookie prompts comes from the ePrivacy Directive’s Article 5(3), rather than GDPR alone.
Brussels has since proposed changes through its Digital Omnibus package, which the source says would move toward one-click choices and browser-level preferences. According to the source material, the European Commission has claimed the changes could save businesses €800 million a year.
AI compute infrastructure equipment
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Funding And Capability Gaps
Several parts of the picture remain unsettled. It is not yet clear how much of the €150 billion in hoped-for private InvestAI funding will materialize, which companies will take part, how quickly gigafactories will come online, or whether they will provide enough compute at competitive prices.
The model rankings and benchmark claims in the source material are fast-moving. The analysis cites systems including GPT-5.5, Claude Opus 4.8, Gemini 3.1, GLM 5.2, DeepSeek V4 and Kimi as of late June 2026. Capability comparisons can shift quickly as vendors release new models, update inference systems or change pricing.
It is also unclear whether regulatory changes to cookie consent will reduce user friction in practice. Browser-level preferences and one-click choices may lower the number of prompts, but implementation details, enforcement and industry response will determine the effect users actually see.
Fine-Tuning with Python: Train, Align, and Deploy Custom LLMs Using LoRA, QLoRA, PEFT, Instruction Tuning, and DPO on Consumer Hardware (Python Series – Learn. Build. Master. Book 15)
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
InvestAI Moves Into Execution
The next test is whether Brussels can turn InvestAI from a funding announcement into operational compute and private investment. The source material says the gigafactory capacity is expected in 2027 or 2028, making procurement, site selection, grid access and energy pricing near-term markers to watch.
Policy changes to cookie consent will also show whether the EU can reduce digital friction while keeping privacy protections in place. For AI, the larger question is whether Europe pairs regulation with build-out: open-weight model development, affordable power, scale-up capital and talent incentives.
AI gigafactory equipment
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
What is the actual news development?
A Thorsten Meyer AI analysis published in late June 2026 argues that Europe’s AI response remains too small and structurally weak despite Brussels’ InvestAI plan and efforts to simplify digital regulation.
Is this a breaking news story?
No. This is an analysis piece based on reported policy plans, investment comparisons and claims about the AI market as of late June 2026.
What is confirmed and what is claimed?
Confirmed in the source material are the EU policy references, including InvestAI and Digital Omnibus proposals. The broader judgments about Europe’s lack of AI leverage, benchmark standing and investment adequacy are claims and interpretations attributed to the Thorsten Meyer AI analysis.
Why do cookie banners matter in this story?
The report uses cookie banners as a symbol of Europe’s focus on regulating user-facing digital interfaces. It argues that this policy reflex did not create the infrastructure, capital base or frontier AI firms needed to compete globally.
What should readers watch next?
Readers should watch whether InvestAI attracts the expected private money, whether AI gigafactories arrive on the stated 2027 to 2028 timeline, and whether European labs close the gap with U.S. and Chinese AI systems.
Source: Thorsten Meyer AI
