Today, 26.05.2026
Good morning, Human. The week opens with a Finnish startup raising capital to solve a problem most people don't know exists: the data bottleneck hiding inside hardware engineering. Meanwhile, Stanford researchers have published what may be the most comprehensive study yet of algorithmic hiring discrimination, and the EU's AI Act enforcement timeline continues its slow-motion recalibration. The through-line connecting these stories is infrastructure, both computational and institutional, and the question of who gets to build it.
In Brief
What: Tampere-based Quanscient has closed a €10 million Series A to scale its cloud-native multiphysics simulation platform, led by Copenhagen's 55 North, the world's largest dedicated quantum venture fund. Why it matters: The round signals that European quantum capital is now flowing into the enabling software layer, not just hardware, betting that AI-driven hardware design will require fundamentally different simulation infrastructure. What it means for Europe: With 55 North's first close at €134 million and Novo Holdings committing €100 million to quantum investments, the Nordics are positioning themselves as the capital formation hub for Europe's quantum stack, while the Quanscient deal suggests the investment thesis is expanding beyond qubits to the data pipelines that will feed them.
These questions about infrastructure, capital, and who shapes the rules are exactly what Human x AI Europe will tackle in Vienna on May 19, where founders, investors, and policymakers are gathering to work through the practical challenges of building in this environment.
The Lead: Quanscient and the Simulation Data Problem
The headline number is €10 million, but the more interesting figure is 89%. That's the percentage of engineers who, according to Quanscient's own research, routinely simplify their physics models to fit within runtime budgets. The implication is stark: the simulation data currently being generated across hardware engineering is not rich enough to train the physics-aware AI models that next-generation design tools will require.
Quanscient, founded in 2021 and headquartered in Tampere, has built what it describes as the first Computer Aided Engineering (CAE) platform designed from the ground up for AI-driven workflows. The platform is code-driven, cloud-scalable, and built to generate the volume of multiphysics data that machine learning models need to learn from. The company claims simulations up to 100 times faster than incumbent tools, which translates to runtimes cut by up to 99%.
The Series A was led by 55 North, the Copenhagen-based quantum fund that announced its first close at €134 million in October 2025, and Austrian industrial investor B&C Group. Existing backers Maki.vc, Crowberry Capital, QAI Ventures, and First Fellow Partners all re-participated. The funding is earmarked for international expansion and for what Quanscient describes as the market's first platform to unify simulation, quantum algorithms, and AI integration.
The lead investor is itself a signal. 55 North is led by Owen Lozman, formerly of M Ventures, alongside Helmut Katzgraber, a computational physicist previously at Amazon and Microsoft, and Kai Hudek, formerly of IonQ. The fund's thesis is that quantum technologies have become a strategic priority at national and European levels, with the G7 pledging coordinated action amid $40 billion in global public funding. Quanscient is one of the fund's first publicly disclosed investments, alongside Finnish quantum-hardware unicorn IQM and German cryogenics specialist Kiutra.
The strategic logic here is worth unpacking. Quantum computing's eventual advantage will likely emerge first in simulation-heavy domains: drug discovery, materials science, fusion energy, advanced semiconductors. But those advantages require training data that current simulation workflows cannot produce at scale. Quanscient is betting that the bottleneck isn't compute, it's the data pipeline feeding the compute. As CEO Juha Riippi put it: "AI will not transform hardware engineering unless simulation itself is rebuilt for it."
The Funding Picture: Nordic Quantum Capital Takes Shape
The Quanscient round lands in a broader context of Nordic quantum capital formation that has accelerated dramatically over the past year. PitchBook data shows funding for European quantum computing startups reached €1.5 billion in 2025, up approximately 170% from the year before.
Denmark has emerged as a particularly active hub. Beyond 55 North's fund, the country launched QuNorth in 2025, backed by €80 million from EIFO and the Novo Nordisk Foundation, to procure and operate Magne, set to become one of the world's most powerful commercial quantum computers through a partnership between Atom Computing and Microsoft. Novo Holdings itself has committed DKK 1.4 billion (approximately €188 million) to quantum investments, with its portfolio now reaching approximately €100 million across direct investments and fund commitments.
The investment thesis is evolving. Early quantum funding concentrated on hardware, the qubits themselves. The Quanscient deal suggests capital is now flowing into the enabling software layer: the simulation tools, algorithms, and data infrastructure that will make quantum-classical hybrid systems useful. This is a maturation signal. The question is no longer just "can we build quantum computers?" but "what will we do with them, and what data will we need?"
Think Tank Watch: Stanford HAI on Algorithmic Hiring
Stanford's Human-Centered Artificial Intelligence institute has published what it describes as the first large-scale study of hiring algorithms in the wild, and the findings are sobering. The research, released today, follows 3.4 million job applications through AI screening systems used by major employers.
The context is important: 90% of U.S. employers now use AI screening tools to sort and rank job seekers, with most relying on the same few third-party vendors. When one algorithm influences many employers, the impact on job seekers compounds. The study found concerning patterns in how systems reject candidates, with evidence of racial bias and what the researchers term "systemic rejection," where qualified candidates are filtered out before human review.
This research arrives as the Mobley v. Workday lawsuit continues to work through federal court. In May 2025, Judge Rita Lin granted preliminary certification for the case to proceed as a nationwide collective action under the Age Discrimination in Employment Act (ADEA). The collective could potentially include hundreds of millions of job applicants who were screened through Workday's AI-powered tools since September 2020. Workday has represented that 1.1 billion applications were rejected using its software tools during the relevant period.
The legal question at the heart of the case is whether AI screening tool providers can be held liable for disparate-impact discrimination, even when the employer technically controls the final hiring decision. The court found Workday was sufficiently involved in the hiring process to be held potentially liable as an "agent" of the employers. For European policymakers watching the EU AI Act's employment provisions take shape, this U.S. litigation offers a preview of the accountability questions that will eventually arrive on this side of the Atlantic.
The Regulatory Calendar: AI Act Timeline Shifts Again
On May 7, negotiators from the Council of the European Union, the European Parliament, and the European Commission reached a provisional agreement on the Digital Omnibus on AI, marking the first set of amendments to the EU AI Act since its adoption in June 2024.
The most significant change is a staggered deferral of compliance deadlines. For high-risk AI systems under Annex III (use-based systems, including those used in employment decisions), obligations are postponed from August 2, 2026 to December 2, 2027, a 16-month delay. For Annex I systems (product-regulated, including medical devices and machinery), obligations move from August 2027 to August 2028.
The delay reflects the reality that the regulatory infrastructure is not yet ready. National competent authorities remain partially designated, and accredited bodies capable of conducting conformity assessments are still in short supply. The technical standards that companies need to demonstrate compliance are running approximately two years behind schedule. CEN-CENELEC, the standards-setting bodies, were meant to have standards ready by April 2025; the current estimate is late 2026.
The agreement also includes a new prohibition targeting "nudifier" applications that generate intimate content without consent, effective December 2026, and clarifies that organizations may process special category data (race, health, sexual orientation) where strictly necessary for bias detection and correction. The Machinery Regulation has been moved from Annex I Section A to Section B, shifting AI-enabled machinery from a dual-compliance model to one where sector-specific laws are paramount.
For organizations doing business in the EU, the message is mixed. The extra runway is real, but the core obligations remain unchanged. Companies that treat this as permission to pause compliance planning are assuming substantial operational risk, particularly as assessor availability is expected to tighten by late 2026.
The Infrastructure Play: Europe's Compute Gap
The AI race is increasingly a race to power, and Europe faces structural challenges. According to the International Energy Agency, data centre capacity increased by 20% last year, mostly in the U.S. and China. By 2030, electricity consumption from data centres is projected to more than double.
Europe's constraint is speed to power. Electricity prices are higher than the U.S., land is more scarce, and grid connections can take up to a decade in some European markets. The consequences are visible: OpenAI stepped back from a major UK data centre project, and some facilities are circumventing grid queues by hooking up directly to gas-fired power plants, clearly in tension with the region's net-zero ambitions.
The EU's response is the AI Factories initiative, which leverages EuroHPC supercomputing capacity to develop trustworthy generative AI models. Currently, 19 AI Factories and 13 Antennas are operational, with at least 9 new AI-optimised supercomputers to be procured and deployed across the EU. The InvestAI Facility will comprise a new European fund of €20 billion to create up to 5 AI Gigafactories, large-scale facilities dedicated to training next-generation AI models containing trillions of parameters.
The question is whether this public infrastructure can move fast enough to matter. U.S. hyperscalers control nearly 70% of the European cloud market. The five largest U.S. cloud and AI infrastructure providers have collectively committed to spending between $660 billion and $690 billion on capital expenditure in 2026, nearly doubling 2025 levels. Europe's €20 billion commitment, while substantial, operates on a different scale and timeline.
The Numbers That Matter
€10M — Quanscient's Series A, led by 55 North and B&C Group, for quantum-native simulation infrastructure.
89% — Share of engineers who routinely simplify physics models to fit runtime budgets, per Quanscient research.
€1.5B — European quantum computing startup funding in 2025, up 170% year-over-year.
90% — Share of U.S. employers using AI screening tools to sort job applicants.
1.1B — Applications rejected through Workday's software tools during the period covered by the Mobley lawsuit.
16 months — Delay for EU AI Act high-risk system obligations, from August 2026 to December 2027.
€20B — EU InvestAI Facility commitment for AI Gigafactories.
The Week Ahead
The provisional agreement on the Digital Omnibus on AI still requires formal adoption by the European Parliament and Council. Watch for the final text, which will confirm the exact timeline shifts and any last-minute adjustments to the bias detection provisions.
The Mobley v. Workday case continues to develop, with plaintiffs having filed an amended complaint on March 30 in response to Judge Lin's partial dismissal earlier that month. The case remains one of the most closely watched tests of AI accountability in employment decisions.
The European Commission's AI Office continues to provide guidance on AI Act implementation through its Service Desk. Organizations should monitor for updated guidelines on high-risk system classification, particularly given the timeline extensions.
The Thought That Lingers
There's a quiet irony in today's news. A Finnish startup raises capital to solve the data bottleneck in hardware simulation, while Stanford researchers document how algorithmic systems are creating new bottlenecks in human opportunity. Both stories are about infrastructure, the computational kind and the institutional kind, and both reveal the same truth: the systems we build to accelerate decisions also encode the assumptions we bring to them. The question isn't whether AI will reshape engineering and hiring and governance. It's whether we're building the feedback loops to catch what we're getting wrong.
Frequently Asked Questions
What is Quanscient and why is their funding significant?
Quanscient is a Finnish startup that has built the first Computer Aided Engineering platform designed specifically for AI-driven workflows. Their €10 million Series A funding is significant because it signals that European quantum capital is now flowing into enabling software infrastructure, not just hardware, betting that AI-driven hardware design will require fundamentally different simulation capabilities.
How does the Stanford study on algorithmic hiring impact European companies?
The Stanford study documents systematic bias in AI hiring tools used by 90% of U.S. employers. For European companies, this research provides crucial insights as the EU AI Act's employment provisions take shape, offering a preview of the accountability questions that will arise when similar regulations are enforced in Europe.
What are the key changes in the EU AI Act timeline?
The most significant change is a 16-month delay for high-risk AI systems, moving compliance deadlines from August 2026 to December 2027. Product-regulated systems see their deadlines shift from August 2027 to August 2028. The delays reflect the reality that regulatory infrastructure and technical standards are not yet ready.
Why is Europe facing challenges in AI infrastructure development?
Europe faces structural challenges including higher electricity prices than the U.S., land scarcity, and grid connections that can take up to a decade in some markets. While the EU has committed €20 billion through the InvestAI Facility, U.S. providers collectively plan to spend $660-690 billion on capital expenditure in 2026 alone.
Human×AI Daily Brief is compiled from The Next Web, Tech.eu, Stanford HAI, PitchBook, The Quantum Insider, Inside Privacy, Fisher Phillips, European Commission, and OutSolve. This is meant to be useful, not comprehensive.