German Startup Raises $6.5M to Make AI Run on Any Chip

·
Listen to this article~5 min

Daisytuner, a German startup, raises $6.5M to make AI run efficiently on any chip, challenging NVIDIA's dominance and launching Daisyflow, a universal inference engine.

You know how everyone's talking about AI compute being the new oil? Well, a German startup just raised $6.5 million to make sure that oil doesn't get burned up in one engine. Daisytuner, based in Darmstadt, is on a mission to make AI models run efficiently on any chip—not just NVIDIA's. And they've just closed a Seed round to do it. The round was co-led by European VCs Elaia and LEA Ventures, with Amadeus APEX Technology Fund and Seedcamp joining in. But the money is only part of the story. Daisytuner also launched Daisyflow, a high-performance inference engine that runs AI models on any hardware. It's like giving your AI a universal adapter. ### Why this matters Here's the thing: most AI compute today is locked into a single vendor's software stack. That's great if you're that vendor, but not so great for everyone else. As Sébastien Lefebvre, Partner at Elaia, puts it: "Compute is the scarcest resource in AI, and most of it currently sits behind a single vendor's software stack. Daisytuner lets the same code run at full performance on the silicon you choose, without a rewrite." In other words, they're removing a major bottleneck. ### How it works Founded in 2024 by Lukas Trümper and Adrian Schmitz, Daisytuner is built on a team with over a decade of research experience from places like ETH Zurich, TU Munich, RWTH Aachen, TU Darmstadt, and Lawrence Livermore National Laboratory. They've developed an open-source compiler called DOCC that translates code from frameworks like PyTorch into native code for each target chip—whether it's NVIDIA or AMD GPUs or newer architectures. The magic is in the automation: AI agents continuously test optimizations on real hardware, and every verified result is stored in a performance database. That means each new workload benefits from past optimizations. ### The bigger picture This funding comes at a time when AI companies and hyperscalers are shifting from adding more capacity to optimizing existing infrastructure. Performance engineering has traditionally been manual and slow, done by a handful of frontier labs. But with hardware becoming more diverse, each new chip or GPU generation risks breaking earlier optimizations. Daisytuner automates this process, replacing manual tuning with a self-learning system. "Demand for AI compute is growing faster than any one chip generation can serve," said Trümper. "The hardware to meet it is already out there, from established GPUs to a wave of new accelerators, but software keeps much of it on the sidelines. We're working with chip vendors, European hyperscalers and enterprises to change that, so every chip runs at full speed and new ones go into production the moment they arrive." ### Daisyflow: the product Alongside the funding, Daisytuner launched Daisyflow, a production-ready AI inference engine. Built on open frameworks, it serves language, vision, speech, and recommendation models using APIs that engineering teams already use. It can be deployed inside a customer's own infrastructure, giving companies full control over their data and costs without needing to build large in-house teams. In a statement, Daisytuner says the platform showed Qwen 2.5 running up to 18% faster than the industry-standard vLLM on AMD Instinct GPUs. The platform is now available via the company's Cloud Console, which brings together the DOCC compiler, continuous benchmarking on real hardware, a web IDE for designing custom optimizations, and a live performance database queried automatically during compilation. ### What's next The startup plans to use the fresh capital to scale Daisyflow's reach across enterprises, grow its engineering team, and broaden its compiler's support for emerging chip architectures. With AI compute demand showing no signs of slowing, Daisytuner's approach could be a game-changer for companies looking to get the most out of their hardware—without being locked into one vendor. So next time you hear about AI compute shortages, remember: the hardware might already be there. It's the software that needs to catch up. And Daisytuner is working on exactly that.