Superintelligence Europe — No. 018

Péter Magyar’s Tisza party won Hungary’s parliamentary election on Sunday April 12 in a landslide. With 97.35% of precincts counted, Tisza secured 138 seats in the 199-seat parliament on 53.6% of the vote. Orbán’s Fidesz took 55 seats on 37.8%. The supermajority gives Magyar the constitutional authority to reverse significant elements of Orbán’s structural changes to Hungary’s institutions, judiciary, and media. 3.3 million Hungarians voted for Magyar — the highest vote total any Hungarian party has ever received. Turnout was nearly 80%. Orbán called Magyar to concede on election night, calling the result “clear and painful.” Magyar told supporters: “Together we replaced the Orbán regime, together we liberated Hungary. Tonight, truth prevailed over lies.”

The European reaction was immediate. Von der Leyen: “Hungary has chosen Europe. Europe has always chosen Hungary. Together we are stronger.” EP President Metsola: “Hungary’s place is at the heart of Europe.” Macron, Merz, Starmer, Tusk, and Zelensky all congratulated Magyar within hours. JD Vance, who had flown to Budapest to rally with Orbán, did not comment publicly.

Ireland’s Department of Enterprise, Tourism and Employment published the announcement Sunday as part of its EU Council Presidency programme. The International AI Summit will officially launch European AI Innovation Month in partnership with the European Commission, at the RDS Dublin on 14 October 2026, running through to Brussels on 17 November. The summit theme is “Harnessing AI to Revolutionise Europe’s Competitiveness” and will focus on Applied AI and sectoral value creation — emphasising Europe’s shift from research to deployment. Sessions will cover AI infrastructure, computing power, sustainable energy, talent, and the full spectrum from generative to agentic AI to quantum convergence. Participants will include EU Commissioners and Ministers, C-suite leaders, and globally recognised AI experts. Headline speakers and an Innovation Spotlight exhibition will be announced in coming weeks.

PwC published its 2026 AI Performance study on Sunday. The global survey of 1,217 senior executives at large, primarily publicly-listed companies across 25 sectors found that 74% of AI’s economic value is being captured by just 20% of organisations — a stark divide between a small group of AI leaders and the majority of businesses still in pilot mode. The research measured revenue and efficiency gains attributable to AI against industry medians, and analysed the impact of 60 AI management and investment practices. The single strongest factor separating AI leaders from everyone else: pursuing new revenue opportunities as industries converge, using AI to grow beyond their traditional sector boundaries. This outranked efficiency gains as a performance driver.

AI leaders were two to three times more likely to use AI to identify and pursue growth opportunities, and 2.6 times more likely to report that AI had improved their ability to reinvent their business model. They deploy AI inside the enterprise fundamentally differently: nearly twice as likely to use AI in advanced ways across multi-task automated workflows, and increasing autonomous decision-making at almost three times the rate of others. Responsible AI is also a differentiator — not a compliance burden. AI leaders are 1.7 times more likely to have a responsible AI framework and 1.5 times more likely to have cross-functional AI governance boards. Their employees are twice as likely to trust AI outputs.

GLBNXT (Global Next) launched its sovereign AI platform on Sunday, founded by Jan Saan — former CTO of CM.com — and Richard van Anholt, who built the full production platform in under twelve months with a team of four. The core architectural claim is specific: the platform runs on European cloud infrastructure only, eliminating exposure to the US CLOUD Act — legislation that can compel American-owned cloud providers to hand over data to US government authorities regardless of where the servers physically sit. European organisations using AWS, Azure, or Google Cloud operate within CLOUD Act jurisdiction by default. GLBNXT removes that exposure at the foundation level.

The platform integrates open-source tools into an enterprise-grade stack with built-in compliance for the EU AI Act, GDPR, NIS2, and the EU Data Act. It includes full design, build, deploy, and scale capabilities; on-premise options; audit logs; and a model hub for comparing LLMs without vendor lock-in. ISO 27001 certified. Strategic partnerships with Dell Technologies and NVIDIA for enterprise hardware and AI acceleration. The company recently closed a funding round and is already winning customers in the public sector, consultancies, and ICT organisations. Deployment to production in 30 days versus the 12–18 months typical for hyperscaler enterprise AI projects.

The Dutch vehicle authority RDW approved Tesla’s Full Self-Driving Supervised on April 10, 2026 under UN Regulation 171 — the EU standard for Driver Control Assistance Systems — after 18 months of testing, 1.6 million kilometres on EU roads, 13,000 customer ride-alongs, and compliance documentation covering over 400 requirements. This is the first regulatory green light for any commercially-active hands-free driver assistance system on EU public roads under the shared EU framework. Tesla pushed the software update via OTA the following day. By Sunday, the system was live and rolling out to compatible Hardware 4 vehicles in the Netherlands at €99 per month or €7,500 to purchase.

The RDW’s own statement makes the regulatory distinction clear: FSD Supervised is a driver assistance system, not an autonomous or self-driving system. The driver remains legally responsible at all times and must be ready to intervene immediately. Eye-tracking cameras monitor attentiveness; a sequence of visual, audio, and haptic alerts triggers if the driver is not attentive. The RDW also explicitly notes that European FSD software differs substantially from the US version — the two are not comparable. Europe requires prior type approval; the US uses self-certification with retrospective review.

The week that began with the EU AI Continent’s one-year milestones and the Stargate UK pause ended with Hungary voting out Orbán in the most AI-disinformation-saturated election in EU history — and democracy working regardless. From Monday to Sunday: UK government threatening prison for tech executives over AI deepfakes; OpenAI facing DSA designation at 120 million EU users; ESC NextGen advancing AI cardiology; Stargate UK’s IPO factor revealed; Ireland claiming its Presidency moment with the October AI Summit; PwC quantifying the AI performance divide; GLBNXT building sovereign AI beyond US legal reach; Tesla going live in the Netherlands as Europe’s first approved hands-free driving system.

The PwC 20/80 finding and the Hungary election result read together across the week’s coverage. PwC says the AI performance gap is widening because most organisations are not converting access to AI into revenue growth. The Hungarian voters had access to documented AI disinformation at scale — and voted at 80% turnout for the candidate it was designed to defeat. Both cases tell the same story in different registers: the outcomes from AI are not determined by the technology. They are determined by what human institutions, organisations, and individuals choose to do with it. That is still true. For now.