As of 19 Nov 2026, Daisytuner has raised $6.6M across 1 verified round on record (amounts in other currencies converted to USD); latest: €6 million seed round led by Elaia and LEA Partners, announced 16 Nov 2026.
A DeepTech startup working on automating AI model performance engineering across any chip
daisytuner.comLinkedInDarmstadt, GermanyAIFounded 2024
20 sources · checked 4 times, last 11h ago
| Round | Amount | Announced | Lead | Buzz |
|---|---|---|---|---|
| Seed | $6.6M €6 million | 2026-11-16 | Elaia, LEA Partners | No recent discussion |
20 sources · checked 4 times, last 11h ago | ||||
Daisytuner builds a compiler, optimization platform and inference stack that automatically makes AI models and other software run faster on CPUs, GPUs and emerging chips for developers and enterprises.
Daisytuner targets a bottleneck the pages describe as manual, slow performance engineering that leaves diverse chips and existing machines underused and can force rewrites when models or hardware change. Its funding is a bet that a self-learning compiler, performance database and Daisyflow can move more AI inference onto customer-owned, heterogeneous hardware, with an early company-reported result of Qwen 2.5 running up to 18% faster than vLLM on AMD Instinct GPUs.
Checked against 10 sources 3d ago
Found on X with Firecrawl search
Also named as backers in company research: APEX Ventures, Angel Invest, Jens Lapinski, Christian Stiebner, Sara Schneider, Elias Schneider, Hans-Jürgen Schmidtke, Jessica Holzbach, Mario Götze, Inari VC, Auke van Urk, Thomas Andrae, Martin Geier, Jens Benatti, Nick Schleich, Ricardo Jimenez