Leuven: Belgian research administrators are entering the busiest ten weeks of their year, because the largest doctoral training call in Europe closes on 24 November and consortium building takes longer than anyone budgets for.
The Marie Sklodowska-Curie Actions Doctoral Networks call for 2026 opened in late May with an indicative budget of 593 million euros. The Commission expects to fund more than 130 projects and to support the training of around 1,950 doctoral candidates across every scientific field, according to the call announcement. Bottom up remains the organising principle. Applicants choose the subject, and evaluators judge the training programme rather than its fit with a policy priority.
The novelty this round is a pilot called RAISE, which backs doctoral networks applying artificial intelligence to scientific research. That framing is narrower than it first appears. RAISE targets AI as a research method inside chemistry, biology, materials science and comparable disciplines, not computer science departments training the next generation of model builders. The distinction reflects a genuine European weakness. Laboratories across the Union generate data at industrial scale and then analyse it with methods that have not kept pace, while the researchers who could close that gap gravitate toward employers who pay far better.
Doctoral Networks work differently from most instruments in the programme, and the difference is the point. A consortium of universities, research institutes and companies from several countries builds a joint training programme, and each recruited candidate spends time in more than one participating organisation. Industrial secondments are standard. The scheme pays competitive salaries under a single European employment framework, which for candidates from lower income member states can mean a doctoral position that pays more than a lectureship at home.
That last feature carries an uncomfortable consequence. The networks concentrate in countries with dense research infrastructure, and their graduates tend to stay near it. Widening measures exist, but the participation gap between the strongest and weakest research systems has proved stubborn across the whole programme, and this scheme is no exception.
Success rates supply the other reason for caution. Doctoral Networks routinely attract several times the number of proposals the budget can support, which means most consortia assembling now will spend months on applications that fail despite scoring above threshold. Research managers have complained for years that the cost of this churn, measured in unpaid academic time, is a real charge on the system that no accounting captures.
The call closes at 17.00 Brussels time on 24 November. Beyond that date sits the harder argument, over what replaces the current framework programme after 2027 and whether doctoral training keeps a dedicated line or dissolves into a larger competitiveness fund. Researchers who value this scheme precisely because it funds people rather than deliverables are watching that negotiation closely, and with reason.
