No deal in Strasbourg. YC in Stockholm. Ireland leads on AI talent.

Tuesday’s session in Strasbourg was the last political trilogue on the Digital Omnibus on AI scheduled before formal adoption. After roughly twelve hours, the institutions left without a deal. The single unresolved question that blocked the entire package: whether AI systems embedded in products already regulated under EU product safety law — machinery, medical devices, toys, radio equipment — should be exempted from the AI Act’s conformity assessment requirements or remain subject to both frameworks simultaneously.

What this means immediately: The proposed deadline shifts — Annex III to December 2, 2027 and Annex I to August 2, 2028 — have no legal effect until a final Omnibus text is adopted and published in the Official Journal. August 2, 2026 remains the operative date under existing law. Every organisation that suspended compliance preparation on the assumption of a postponement is now carrying live legal risk. The rapporteurs for the file — Arba Kokalari (EPP, Sweden) and Michael McNamara (Renew, Ireland) — briefed press in Strasbourg on Wednesday morning.

The next trilogue is expected around May 13, 2026, under the Cypriot presidency which runs until June 30. If no deal is reached before July, the Lithuanian presidency takes over from July 1. Former AI Act negotiator Laura Caroli called the collapse damaging: the risk is not just regulatory but structural — undermining the harmonised standards ecosystem that European companies need to demonstrate AI Act compliance regardless of which deadline ultimately applies.

The study by Germany-based think tank Interface, drawing on a Revelio Labs dataset of 1.6 million AI professionals, finds that Ireland is the second-largest AI talent market in the world on a per-capita basis, with 4.19 AI professionals per 1,000 inhabitants — behind only Singapore. Switzerland (3.25), Luxembourg (3.18), the Netherlands (2.56), and Denmark (2.33) also rank in the global top 10 per capita.

In absolute terms, Germany is the standout: 17,000 AI engineers — the fourth-largest national total worldwide, behind only the US, India, and China — consolidating its position as Europe’s frontier AI engineering hub. The overall picture is one of genuine strength in concentrated pockets, with the gap between Northern and Western Europe and Southern and Eastern Europe remaining substantial.

The geopolitical dimension: tightening US immigration rules under the Trump administration are redirecting talent toward Europe. Indian AI professionals now make up 8.3% of the EU AI workforce, up from 7.7% in 2024. In Ireland, Indian employees account for nearly 30% of the AI talent pool, up from 21% in 2024. The Netherlands is emerging as a specific magnet for American AI professionals relocating to Europe. France is the outlier: strong AI system, losing international talent, and the only major EU country with more women in senior AI positions — a function of hiring locally rather than internationally, the researchers note, which does not indicate improvement in the broader gender picture.

Y Combinator hosted a special free event in Stockholm on Tuesday, bringing its co-founders Paul Graham and Jessica Livingston to Sweden alongside Gustaf Alströmer, General Partner at YC, and Christina Gilbert, YC Visiting Partner. The event was organised by SWENODE.AI in partnership with Wilson Sonsini, Dell Technologies, Legora, Inception Fund, and Google. The framing was explicit: this is not a networking event. It is a recruitment exercise. YC is looking for engineers, builders, and anyone curious about startups — “whether you’re a college CS student, an industry professional, or a senior in high school.”

The anchor speaker, beyond the YC co-founders themselves, was Max Junestrand, CEO and co-founder of Legora (YC W24). Legora is perhaps the most compelling exhibit of what the Stockholm builder ecosystem can produce: founded in 2023 by Junestrand at 23, it is now valued at $5.5 billion after a $550 million Series D in March 2026, serves over 800 law firms across 50+ markets, and made two acquisitions in April alone — Canadian legal AI startup Walter and Stockholm-based legal research startup Qura, whose CEO described Legora’s architecture as “rethinking legal research from first principles.”

The context behind the Stockholm visit is one data point: 60% of YC’s Winter 2026 batch founders were from outside the United States. As every major US investor increases their time in Stockholm — a pattern confirmed by multiple Nordic VC observers — and as tightening US immigration policy pushes talent toward Europe, the gravitational pull of Silicon Valley is weakening precisely as the quality of European builders is rising. Paul Graham coming to Stockholm to recruit is a leading indicator, not a lagging one.

The European Parliament votes today, Thursday April 30, on a resolution arising from its DMA review debate held Monday in Strasbourg. The resolution calls for quicker compliance proceedings under the Digital Markets Act, closer scrutiny of AI-driven search tools and cloud services, and stronger enforcement coordination between the Commission and national competition authorities.

The vote arrives in the same Strasbourg session as the Omnibus rapporteurs briefed press on the failed trilogue — creating an unusual week in which Parliament was simultaneously failing to ease AI regulation and voting to tighten platform regulation. The two are connected. The DMA debate was shaped in part by Monday’s Google Android preliminary findings from the Commission, which concluded that Google must grant rival AI services — including ChatGPT and Claude — the same Android system access it gives its own Gemini assistant. The Parliament’s resolution reinforces that direction, pushing the Commission to proceed faster on open platform access for AI services.

The CDT Europe April AI Bulletin also flagged a related enforcement action: in the context of Meta’s exclusion of third-party AI assistants from WhatsApp, the Commission sent a supplementary statement of objections to Meta on April 15 — arguing that Meta’s revised fee-based access model produces the same exclusionary effect as the original ban. Interim measures are being considered. The week’s pattern is consistent: while the Omnibus stalls on compliance timelines, the enforcement machinery on AI platform access is accelerating independently.