Every once in a while, we meet a team whose journey mirrors the very evolution of the industry they are reshaping. That is the case with yasp. The company was born out of the lived frustrations of engineers who spent years at the frontlines of AI development. Building computer vision systems and optimizing for automotive applications, they ran into the same barriers time and again: rising compute costs, painfully slow iteration cycles, and the grip of vendor lock-in.
yasp was created to break this cycle. At its heart is the Agentic AI Compiler, a tool that takes the complexity out of AI infrastructure. Instead of forcing developers to rewrite models for every new chip or cloud setup, yasp automatically generates optimized kernels for the hardware at hand. The vision is bold yet simple: write once, deploy anywhere. Early results show just how powerful this approach is. Faster training, smoother deployment, and significant efficiency gains across both cloud and edge environments.
Tremendous growth potential
The timing for this innovation could not be better. The global AI infrastructure market is projected to exceed USD 400 billion by 2030, with software and services expected to account for as much as USD 200 billion of that spend. Within this layer, performance optimization tools, like compilers, kernel generation, inference engines, and runtime optimizations, are estimated to capture up to 30% of the market.
Training and deploying AI models today remains an expensive and cumbersome process. Most enterprises rent diverse GPU resources in the cloud, only to spend weeks re-engineering those models for smaller, constrained devices at the edge. The outcome is wasted money, lost time, and delayed innovation. By automating this process, yasp doesn’t just make things faster, it makes cutting-edge AI accessible to more companies, regardless of the hardware they use.
Why we invested in yasp
Behind this vision is a team that knows the pain points firsthand. CEO Stefan Krassin, CTO Reza Rahimi, and the broader yasp team bring decades of experience from Algolux, Torc Robotics, and semiconductor engineering. They understand the daily struggles of AI developers because they’ve lived them. Their decision to build yasp was about more than technology; it was about giving developers back control over their time, budgets, and hardware choices.
For us at Capnamic, conviction came quickly. We didn’t just see a product. We saw the foundation of a platform that could change how enterprises build, train, and deploy AI models. By turning advanced hardware optimization into something as simple as a single API call, yasp lowers barriers while setting a new standard for efficiency and flexibility.
We are proud to lead their USD 5m seed round with our colleagues from MBG and to support Stefan, Reza, and the entire yasp team as they redefine what’s possible in AI development. This investment is not just capital; it’s a commitment to backing the people and ideas that will shape the next decade of AI infrastructure.
.png)
.avif)
.avif)
.avif)
.avif)
.avif)
