Standard Kernel, a startup focused on automating low-level software optimization for artificial intelligence systems, announced that it has raised $20 million in seed funding. The round was led by Jump Capital with participation from General Catalyst, Felicis, Cowboy Ventures, Link Ventures, Essence VC, and a group of angel and strategic investors, including David M. Siegel, Jeff Dean, Jonathan Frankle, Michael Carbin, Sachin Katti, Walden Yan, CoreWeave, and Ericsson Ventures.
The Palo Alto–based company is building AI systems that autonomously generate highly optimized GPU kernels, the core computational units that determine how efficiently AI models run on hardware. The goal is to significantly improve AI workload performance without requiring changes to the models or the underlying hardware.
As investment in AI infrastructure continues to accelerate globally, companies are deploying massive GPU clusters. However, a significant portion of this hardware often fails to operate at peak theoretical performance. Extracting maximum efficiency from modern accelerators typically requires deep expertise in hardware architecture, compiler behavior, and low-level systems engineering. Much of the performance-critical code used today is still manually written and tuned, making it difficult to keep pace with rapidly evolving chips and increasingly complex workloads.
The Standard Kernel is addressing this problem by using AI to generate specialized GPU kernels tailored to specific workloads and hardware configurations. By operating deep in the computing stack and optimizing down to native chip instructions, the company replaces static libraries with code generated for the precise environment in which it runs.
In partner testing, the company reports performance improvements ranging from 80 percent to 4x on end-to-end workloads running on NVIDIA H100 GPUs. In some cases, the generated kernels outperform NVIDIA’s cuDNN library, which is widely regarded as one of the most highly optimized GPU libraries available.
Kernel generation has recently emerged as a benchmark task for large language models, but many existing approaches focus on higher-level abstractions or relatively simple workloads. The Standard Kernel targets instruction-level, hardware-specific kernel generation capable of matching or surpassing human-engineered implementations.
The company aims to automate this process so that AI workloads can achieve peak performance on new hardware platforms immediately, rather than waiting for the lengthy manual optimization cycles typically required when new chips are introduced.
With the new funding, Standard Kernel plans to accelerate development of its autonomous kernel generation platform, expand deployments with AI-native companies and enterprise partners, and advance toward adaptive systems software that continuously improves as new models and hardware platforms emerge.
Standard Kernel’s team includes researchers and engineers with backgrounds spanning machine learning, computer systems, and hardware-level optimization, with alumni from MIT, Stanford, the University of Illinois Urbana-Champaign, and Shanghai Jiao Tong University. The team has also contributed open-source research and benchmarks, including KernelBench and Kernel Tree Search.
KEY QUOTES:
“What excites us about Standard Kernel is that they are applying AI to one of the most manual and technically demanding layers of the stack. Hardware innovation is accelerating, but the software that extracts peak performance from it has lagged behind. Automating instruction-level optimization has the potential to meaningfully change how AI infrastructure scales.”
Saaya Pal, Partner at Jump Capital
“Standard Kernel is tackling one of the most consequential challenges in modern compute, driving optimization deep within the systems stack where performance is won or lost. 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 caliber of the team are precisely why CoreWeave Ventures is proud to invest in Standard Kernel as they shape the future of AI systems.”
Brian Venturo, Co-founder And Chief Strategy Officer, CoreWeave
“Kernel generation is key for improving performance and efficiency of AI hardware. As fleet sizes for users of AI hardware get larger, and more hardware diversity is introduced, Standard Kernel becomes key to deployment.”
Dylan Patel, Founder of SemiAnalysis
