Global spending on AI infrastructure has grown rapidly as companies build large GPU clusters to train and run AI models. However, much of that hardware does not operate at its full potential.
Optimising GPU performance usually requires deep technical knowledge of chip architecture, compilers, and low-level software systems. Most performance-critical code is still written and tuned manually, which can slow development and limit efficiency.
Standard Kernel is attempting to automate this process. Palo Alto–based Standard Kernel has raised $20 million in a seed funding round.
The round was led by Jump Capital, with participation from General Catalyst, Felicis, Cowboy Ventures, Link Ventures, Essence VC, and several angel investors.
Strategic backers also include CoreWeave Ventures and Ericsson Ventures, as well as industry figures such as Jeff Dean and Jonathan Frankle.
The Palo Alto–based company is developing technology that uses AI to automatically generate GPU kernels, the core software components that control how efficiently AI models run on hardware.
Automatically generate ultra-optimised GPU software
Led by Anne Ouyang, the company’s system uses AI to create highly specialised GPU kernels tailored to specific workloads and hardware configurations. By generating code directly at the chip’s instruction level, the platform aims to replace standard libraries with software tailored to each specific AI task.
According to the company, partner testing showed performance improvements ranging from 80% to 4x for certain AI workloads running on NVIDIA H100 GPUs. In some cases, the system outperformed NVIDIA’s cuDNN library, which is widely used to optimise AI workloads.
“Standard Kernel is tackling one of the most consequential challenges in modern compute, driving optimisation deep within the systems stack where performance is won or lost,” said Brian Venturo, Co-founder and Chief Strategy Officer, CoreWeave.
Venturo continues, “As AI adoption continues to scale, breakthroughs in the layers beneath today’s models will define the next generation of capabilities. That depth of technical ambition and the calibre of the team are precisely why CoreWeave Ventures is proud to invest in Standard Kernel as they shape the future of AI systems.”
Funding deployment
With the new funding, the company plans to continue developing its autonomous kernel generation platform and expand deployments with enterprise and AI-focused companies.
The startup’s team includes engineers and researchers from institutions such as MIT, Stanford, the University of Illinois Urbana-Champaign, and Shanghai Jiao Tong University.
Standard Kernel has also contributed to open-source projects and benchmarks related to AI performance testing, including KernelBench and Kernel Tree Search.
