Standard Kernel Raises $20M Seed Round to Let AI Rewrite the Software That Runs AI

The startup uses AI to generate highly optimized GPU kernels, improving AI workload performance without changing models or hardware.

PALO ALTO, Calif., March 11, 2026 /PRNewswire/ -- Standard Kernel, a startup building AI systems that automatically generate ultra-optimized GPU software, today announced a $20 million seed round led by Jump Capital, with participation from General Catalyst, Felicis, Cowboy Ventures, Link Ventures, Essence VC, and an incredible group of angels and strategic partners including David M. Siegel, Jeff Dean, Jonathan Frankle, Michael Carbin, Sachin Katti, Walden Yan, CoreWeave, and Ericsson Ventures.

As global investment in AI infrastructure accelerates, companies are deploying hundreds of billions of dollars into GPU clusters. Yet much of that hardware does not run at peak theoretical performance. Extracting maximum efficiency from modern accelerators requires deep expertise in hardware architecture, compiler behavior, and low-level systems optimization. Today, most performance-critical code is still written and tuned manually, making it difficult to keep pace with rapidly evolving chips and increasingly complex AI workloads.

Standard Kernel is taking a different approach. The company uses AI to autonomously generate highly specialized GPU kernels, the foundational units of computation that determine how efficiently models run. By operating deep in the stack and optimizing down to native chip instructions, Standard Kernel replaces static, one-size-fits-all libraries with code precisely tailored to specific workloads and hardware configurations.

In partner testing, Standard Kernel has demonstrated performance improvements ranging from 80 percent to 4x on end-to-end workloads running on NVIDIA H100 GPUs, outperforming NVIDIA's highly optimized cuDNN library in certain scenarios.

Kernel generation has recently become a popular benchmark task for large language models, but most existing approaches focus on higher-level abstractions or simpler workloads. Generating instruction-level, hardware-specialized kernels that match or exceed the best human-engineered implementations remains an open challenge. Standard Kernel's goal is to automate that frontier, enabling day-one peak performance on new hardware platforms without waiting for lengthy manual tuning cycles.

"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," said Saaya Pal, Partner at Jump Capital. "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."

"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," said Brian Venturo, Co-founder and Chief Strategy Officer, CoreWeave. "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."

"Kernel generation is key for improving performance and efficiency of AI hardware," added Dylan Patel, founder of SemiAnalysis. "As fleet sizes for users of AI hardware get larger, and more hardware diversity is introduced, Standard Kernel becomes key to deployment."

With the new funding, Standard Kernel will accelerate development of its autonomous kernel generation platform, expand deployments with AI-native and enterprise partners, and continue advancing toward adaptive systems software that improves alongside new models and new hardware.

About Standard Kernel

Standard Kernel brings together expertise across machine learning, computer systems, and hardware-level optimization. The team includes alumni from MIT, Stanford, UIUC, and SJTU, and has contributed widely used open-source research and benchmarks, including KernelBench and Kernel Tree Search.

Standard Kernel is hiring engineers and researchers interested in building AI systems that optimize AI itself. Learn more at https://standardkernel.com/ or contact [email protected].

SOURCE Standard Kernel