The European Union is financing 29 million euros for laboratories capable of designing and executing their own experiments without human intervention between each cycle. These closed-loop systems could reduce the cost of a scientific discovery. But the Commission’s SRIP 2026 report frames the problem clearly: Europe already produces world-leading science that it struggles to convert into technological leadership. If autonomous laboratories concentrate in the same clusters that already dominate, this gap between scientific excellence and industrial capacity could widen.
The Essentials
- Opened on October 13, 2026, the RAISE call totals 30.5 million euros, of which 29 million funds automated scientific discovery, including demonstrators, financed by Horizon Europe.
- The European Commission’s SRIP 2026 report places the Union in second place globally for scientific output, but documents insufficient conversion of this strength into productivity gains and patents.
- A RAISE autonomous laboratory can chain certain planning, execution, and analysis steps in a closed loop, with human oversight provided for safety, accuracy, and ethics.
- The structural risk: if access to these infrastructures remains reserved for the best-resourced universities and companies, automation risks deepening the geographic and institutional gaps already visible in Europe.
- RAISE demonstrators may reduce experimental costs for their users; governance and funding for the future Data Lab for Science still need to be worked out.
A Laboratory That Programs Itself
To understand what is at stake with RAISE, we first need to understand what a closed-loop laboratory actually does. In a classical laboratory, a researcher formulates a hypothesis, plans an experiment, waits for results, interprets them, then decides on the next step. This cycle can take weeks. An autonomous system executes the same cycle in hours, sometimes minutes, by integrating machine learning at each iteration.
The most cited model is the Robot Scientist developed by Ross King at Aberystwyth University: capable of formulating hypotheses, designing experiments to test them, executing them, and interpreting results without human intervention. This system notably identified potential inhibitors in yeast metabolic pathways, a demonstration published in Science in 2009 that laid the foundations for the field. Since then, autonomous laboratories have appeared in chemistry, synthetic biology, and materials science. RAISE funds the next generation: demonstrators capable of operating on more complex problems, with finer feedback loops.
The economic stakes are direct. Pharmaceutical research costs more than a billion dollars on average per approved drug, with much of that cost coming from experimental exploration. In materials chemistry, candidates to evaluate for a next-generation battery number in the millions. A system that reduces time per iteration by a factor of ten radically changes this calculation. The terms of access to these systems will affect how the benefits they produce are distributed.
The SRIP 2026 Diagnosis
The report on the scientific and innovation performance of the Union (SRIP 2026) draws a precise picture. In 2020, China clearly surpassed the EU for publications in the top 10 percent most cited. In the global share of publications in the top 1 percent most cited, the United States remains first and China surpasses the European Union. Its universities train world-class researchers. But between this knowledge production and productivity gains in the real economy, there is a gap that the report has been documenting for several editions.
This gap takes several forms. High-technology-value patents remain concentrated in a small number of regions: Île-de-France, Bavaria, Randstad, a few Swedish and Finnish corridors. The venture capital funding research valorization is concentrated even more. Researchers who move from university to business do so predominantly toward a handful of clusters. South Korea and Finland have demonstrated that one can build valorization ecosystems over a generation.
But the path requires accessible infrastructures, not just prestigious publications.
The decision to finance autonomous laboratories through Horizon Europe directly raises the question of concentration. If these systems remain confined to major research clusters, they will amplify existing concentration. If they are designed as shared infrastructures, they could become a lever for rebalancing. RAISE encourages cooperation and sharing in some of its components, without guaranteeing that the demonstrators will be shared infrastructures.
29 Million Euros for What, Exactly
The RAISE call stands for “Resource for AI Science in Europe.” It finances projects aimed at integrating laboratory automation, data infrastructures, and AI-driven decision-making to produce systems capable of exploring a problem space autonomously. The 29 million euros finance up to three research and innovation projects on automated scientific discovery. The call requests interdisciplinary collaboration, with mechanisms facilitating the sharing of experimental data and collaboration between laboratories; dissemination of results also falls under the general rules of Horizon Europe.
This requirement for sharing is the central political point. In the current race for artificial intelligence for science, several companies, notably North American actors like Insilico Medicine, Recursion Pharmaceuticals, or divisions of Microsoft Research, have developed proprietary systems that keep their architectures and results closed. Competitive advantage rests on the accumulation of proprietary experimental data. An autonomous laboratory that generates thousands of experiments per day produces training data that no one else possesses.
The Commission is making a different bet. It finances infrastructures whose results must, at least partially, remain accessible to the European scientific community. This is consistent with the logic of Horizon Europe, which requires open access to publications. But open access to publications is not the same as open access to raw experimental data, nor to trained models. The exact conditions for sharing set out in RAISE will determine whether this bet holds over time.
The calls for comments on these conditions remain open until the selection of the first consortiums.
One practical point deserves attention: 29 million euros for closed-loop demonstrators constitutes a seed-phase budget. A complete automation line for a materials chemistry laboratory can cost several million on its own. RAISE finances the proof of concept and reference architectures. Deployment at scale will come later, with other instruments and other debates about the access model.
The Geography of Research Determines That of Employment
The question of concentration is not abstract. Research infrastructures largely influence the creation of scientific and technological jobs, patent filing, and the capture of industrial spillovers from a technological wave. The history of the internet, biotechnologies, or semiconductors has shown this repeatedly: the clusters that have the most advanced experimentation infrastructures absorb a significant share of the value created.
This dynamic is particularly clear in materials science and pharmacology, two fields where autonomous laboratories are most advanced. Liverpool University built a chemist robot capable of working 24 hours a day to explore photocatalyst formulations. University of Toronto developed a similar system for solar materials. These projects are concentrated in already well-resourced universities, with established industrial partnerships. A researcher in Slovakia, Bulgaria, or in a French regional university works in a fundamentally different experimental environment.
The reference architectures financed by RAISE can become reproducible models provided that precise technical choices are made in that direction. Modular systems, built on commercially available equipment, are easier to deploy in less-resourced laboratories than custom installations. Standardized interfaces allow different groups to contribute to common experiments. Data access policies for training influence the use, by teams of varying sizes, of models developed by a consortium.
On this point, RAISE is part of a broader movement around the sharing of research infrastructure in Europe. The Commission’s strategy for research and technology infrastructures identifies shared platforms as a priority for reducing inequalities between member states. But strategic intention and actual access model are two different things. The Union has already shown this tension in other technological fields: regulations are ambitious, but their application depends on governance mechanisms.
The Trade-offs of the 2026-2035 Decade
Over a ten-year horizon, two trajectories are plausible. In the first, autonomous laboratories financed by RAISE and their successors remain consortium infrastructures, accessible to project partners, but practically closed to institutions that lack the resources to join these consortiums. Accumulated experimental data creates proprietary advantages in a few clusters. European scientific productivity increases, but its geographic distribution narrows. The 2030 SRIP report will document progress in volume and increased concentration.
In the second trajectory, reference architectures developed by RAISE become open standards. Modular and affordable equipment allows regional universities to deploy autonomous experimentation capacities at a fraction of the cost of the first demonstrators. Shared data platforms allow dispersed teams to contribute to common training bases, creating collective goods that no one could build alone. Training programs accompany deployment so that access to the tool comes with the capacity to use it.
The difference between these two trajectories comes down to governance choices made between 2026 and 2030: licensing conditions, data access, and eligibility are framed by Horizon Europe rules, call conditions, and future beneficiary agreements. These choices are currently under deliberation, and selected consortiums can contribute to RAISE implementation and research agendas, under coordination of its central governance.
A signal to watch: how will the Commission define sharing obligations in grant agreements. The distinction between “open access to publications” and “open access to data and models” is technical-legal but decisive. A second variable: whether complementary infrastructure programs, such as those from the European Investment Bank for research equipment, align with RAISE architectures or develop their own standards. Fragmentation of standards is the fastest path to de facto concentration.
The automation of physical processes poses similar distribution questions in other sectors: technology reduces a cost, but who captures this reduction depends on access infrastructures. For research, the mechanism is the same. An autonomous laboratory that costs ten times less to operate than an equivalent human team theoretically expands Europe’s scientific capacity. But this expansion remains conditional on access models that allow institutions currently excluded from the experimental frontier to enter it.
Europe’s Comparative Advantage Is to Be Built, Not Protected
The Union has a real asset: world-leading scientific output, institutional diversity, and a regulatory framework that can make open access binding rather than optional. It has a real deficit: the valorization of its research into productivity gains, technological jobs, leadership in emerging markets.
Autonomous laboratories are a technology whose marginal cost of experimentation drops rapidly. This is exactly the kind of infrastructure where an open access policy can create a lasting collective advantage. A Slovenian university or a small Breton biotech that can access the same exploration tools as a major London or Munich consortium produces science that adds to Europe’s overall potential. A broader scientific community generates more hypothesis diversity, which is valuable in fields where the problem space is vast.
The 29 million euros constitutes the indicative budget for one RAISE topic on automated scientific discovery; the RAISE pilot mobilizes a total of 107 million euros in Horizon Europe 2026-2027. It is a beginning, not a complete policy. Its real value will be measured by what selected consortiums do with the architectures they develop: will they keep them as proprietary assets or publish them as common infrastructures. Grant and consortium agreements will frame certain access rights and intellectual property management, but they do not allow one to assert in advance how the distribution of European scientific capacities will be affected.
Sources
- European Commission, RAISE call, Automated Scientific Discovery (Horizon Europe, opened September 22, 2026)
- European Commission, Science, Research and Innovation Performance of the EU (SRIP) 2026, report available on the Directorate-General for Research and Innovation portal of the European Commission
- European Commission, strategy for research and technology infrastructures, European Research Area portal
- Ross King et al., “The Automation of Science,” Science, vol. 324, 2009, foundational study on the Robot Scientist and autonomous discovery in biology
- University of Liverpool, autonomous chemist robot project (Adam Floating), institutional communications from the chemistry department, 2022-2024
- University of Toronto, autonomous system for solar materials, publications from the Alán Aspuru-Guzik laboratory, 2021-2024
