Daisytuner has secured €6 million in seed funding to scale its technology that automates tuning of AI models to the chips they run on. The round was co-led by Elaia and LEA Ventures, with participation from Amadeus APEX Technology Fund and Seedcamp.
The Berlin-based deep tech startup pairs a self-learning, open-source compiler with a patented optimisation platform. The compiler converts code from frameworks such as PyTorch into native code for Nvidia, AMD and next-generation chips, while AI agents test optimisations on real hardware around the clock. Each verified improvement is stored and reused so subsequent workloads benefit from prior results.
Daisytuner’s technology and Daisyflow
Daisytuner’s production platform, Daisyflow, runs inside customers’ own infrastructure. Early results cited by the company show Qwen 2.5 running up to 18% faster than vLLM on AMD Instinct GPUs. The company describes a workflow in which compiler-generated native code and continuous hardware testing produce reusable optimisations that accumulate across workloads.
Founding, investors and next steps
CEO and co-founder Lukas Trümper said compute demand is outpacing any single chip generation, yet software keeps much available hardware sidelined. Trümper founded the company in 2024 with Adrian Schmitz, drawing on research from ETH Zurich, TU Munich and Lawrence Livermore National Laboratory.
Sébastien Lefebvre of Elaia said Daisytuner lets code run at full performance on any silicon without rewrites, while Alexi Malikotsinas of LEA Ventures highlighted compute efficiency and sovereignty. The funding will be used to expand Daisyflow, grow the team and develop the compiler technology further.

