The Next Frontier AI Challenge, revealed at EurIPS in December, directs applicants not to catch up with OpenAI but to aim for the next architectural S-curve, offering up to €1 billion in follow-on funding for the three winning labs.
SPRIND, Germany’s federal agency for groundbreaking innovation, has begun accepting applications for its Next Frontier AI Challenge. This €125 million, 24-month structured competition aims to create up to three European frontier AI labs from scratch.
The application timeline is open until 1 June 2026, with jury presentations set for 24–25 June and the first ten funded teams commencing projects in July.
Announced at EurIPS in Copenhagen on 3 December 2025, a European conference endorsed by NeurIPS, the leading AI research conference worldwide, the challenge is clear about Europe’s current standing.
“Europe’s AI innovation competitiveness lags far behind the USA and China,” the challenge brief states. “Without creating its own models, Europe risks increasing its strategic dependence on these technologies.”
The 💜 of EU tech
The latest updates from the EU tech scene, a story from our experienced founder Boris, and some debatable AI art. It’s free, weekly, in your inbox. Sign up now!
The goal is not to bridge this gap on the existing path but to bypass it by aiming at what SPRIND defines as the next S-curve—the upcoming architectural and paradigmatic advancement beyond current transformer-based systems.
How does the competition work?
The €125 million is allocated across three stages with progressive down-selection. In Stage 1, up to ten teams each get up to €3 million over seven months, with their main task being initial technological proof points for their frontier hypothesis, a technical report, a preprint, experimental artifacts, or evidence of a new scaling dimension or emergent phenomena.
Up to six teams move on to Stage 2, receiving up to €8 million each over eight months, where the focus shifts to production-ready engineering processes, validated scaling dimensions, and the first discovery of what SPRIND terms ‘technical secrets,’ proprietary insights offering significant performance gains.
Up to three winners then advance to Stage 3, securing up to €15.5 million each over nine months, aiming for a functional frontier system prototype, user-facing applications under testing, and an investment-grade data room prepared for subsequent capital raising.
The total non-dilutive funding per team across all three stages is €26.5 million, a substantial seed for a serious AI lab, though less than what frontier labs like Anthropic or Mistral have amassed.
The true prize is what follows: SPRIND explicitly designs the program backward from a target €1 billion scale-up round for each winning lab at the end of the 24-month competition, positioning that capital as akin to a US mega Series A that transforms a ‘serious seed lab’ into a ‘genuine frontier player.’
The €1 billion is not within the challenge budget and would require the labs to raise it from external investors; SPRIND’s Financing Workstream is structured to help teams build investment-grade data rooms to substantiate that raise.
The architectural wager
The challenge is explicitly technology-agnostic in its submission criteria but equally clear about what it does not seek.
SPRIND’s disallowed categories are informative: incremental transformer optimization without fundamentally new capability horizons; reproduction or derivatives of established models like rebuilding OpenAI, Llama, or Qwen; incremental efficiency improvements such as better quantization or leaner MoE routing; traditional agent architectures lacking systemic innovation; domain-specific fine-tuning without foundational innovation; and brute-force scaling as the primary innovation approach.
What