Skip to content
AIPressRoom

FundingFoundation Models & Infrastructure

Daisytuner Raises €6M to Optimize AI Performance

Daisytuner raised €6M Seed funding co-led by Elaia and LEA Ventures to automate AI model performance engineering across diverse hardware.

Lukas Trümper, Co-founder and CEO of Daisytuner.

The big picture: Darmstadt-based DeepTech startup Daisytuner has raised €6 million in Seed funding to automate AI model performance engineering across any chip architecture. The company also launched Daisyflow, a high-performance inference engine designed to run AI models efficiently on diverse hardware.

Why it matters:

  • Compute scarcity: AI compute is a critical and scarce resource, with much of it currently tied to single vendors’ software stacks, limiting the efficient use of available hardware.
  • Manual bottleneck: Historically, optimizing AI performance at the chip level has been a manual, slow process, often breaking with new hardware generations and leaving most chips working below their potential.
  • Industry shift: The AI industry is increasingly prioritizing the optimization of existing AI infrastructure over simply adding more capacity, making efficient hardware utilization a key competitive advantage.

How it works:

  • Self-learning compiler: Daisytuner employs a self-learning compiler, DOCC, paired with a patented optimization platform to translate code from frameworks like PyTorch into native code for various target chips, including NVIDIA, AMD, and emerging architectures.
  • AI agents and database: AI agents continuously test optimizations on real hardware, storing verified results in a performance database. These results are then reused to improve each new workload, creating a continuous feedback loop.
  • Daisyflow engine: The new Daisyflow inference engine is production-ready, serving language, vision, speech, and recommendation models via standard APIs. It can be deployed within a customer’s own infrastructure, offering control over data and costs.

The catch: The AI optimization landscape is becoming increasingly competitive, with major chip vendors and hyperscalers also investing heavily in in-house solutions. Daisytuner must consistently demonstrate significant and measurable performance gains across a rapidly evolving hardware ecosystem to maintain its edge against these large players and other specialized startups.

Key Facts

  • Company: Daisytuner
  • Amount: €6M
  • Round: Seed
  • Investors: Elaia and LEA Ventures (co-lead), Amadeus APEX Technology Fund, Seedcamp
  • Founders: Lukas Trümper, Adrian Schmitz
  • Announced: 2026-11-09
  • Sector: AI Model Performance Engineering
  • Headquarters: Darmstadt, Germany

AIPressRoom Weekly

Every round we reported this week, the people behind the companies, and the events coming up. One email, every Tuesday.

More in Foundation Models & Infrastructure

All