Daisytuner Raises €6 Million to Accelerate AI Model Execution Across Chips
TechFlow reports that on November 16, according to EU-Startups, German DeepTech startup Daisytuner, based in Darmstadt, has closed a €6 million seed round led by European VCs Elaia and LEA Ventures, with participation from Amadeus APEX Technology Fund and Seedcamp.
Founded in 2024 by CEO Lukas Trümper and Adrian Schmitz, the team brings research backgrounds from ETH Zurich, the Technical University of Munich, RWTH Aachen University, TU Darmstadt, and Lawrence Livermore National Laboratory. The company automates AI model performance engineering using a self-learning compiler paired with a proprietary optimization platform: its open-source compiler, DOCC, translates code from frameworks like PyTorch into native chip-specific code, supporting GPUs from NVIDIA and AMD as well as newer accelerator architectures. AI agents continuously test and optimize on real hardware, logging validated results into a performance database to enable reuse across similar code.
Coinciding with the funding round, the company also launched Daisyflow, a production-ready inference engine supporting language, vision, speech, and recommendation models, designed for deployment within customer-owned infrastructure. Daisytuner states that on AMD Instinct GPUs, Daisyflow runs Qwen 2.5 up to 18% faster than the widely used industry standard vLLM. The newly raised funds will be allocated to expand Daisyflow’s enterprise footprint, scale the engineering team, and extend compiler support for emerging chip architectures.





