Anne Ouyang announced on social media that her team at Standard Kernel has successfully raised a funding round from prominent venture capital firms General Catalyst and Felicis, alongside a group of angel investors. The announcement highlights the company's focus on developing highly optimized AI kernels, which are critical for enhancing the performance of artificial intelligence models.
"Excited to share what friends and I have been working on at @Standard_Kernel," Ouyang stated in her tweet. "We've raised from General Catalyst (@generalcatalyst), Felicis (@felicis), and a group of exceptional angels."
Standard Kernel is developing specialized H100 BF16 kernels in pure CUDA+PTX, demonstrating significant performance improvements over existing solutions. The company claims its Matmul kernels achieve 102%-105% the performance of cuBLAS, while its Attention kernels surpass FlashAttention3 by 104%. Furthermore, their fused Llama3 FFN kernels reportedly deliver 120% the performance of PyTorch (gpt-fast).
The investment from General Catalyst and Felicis underscores a continued trend of venture capital interest in foundational AI infrastructure. Both firms have a history of backing innovative technology companies, particularly in the AI and deep tech sectors. This funding round is expected to accelerate Standard Kernel's development efforts and market penetration in the competitive AI hardware and software optimization landscape.
The company is actively recruiting talent, with Ouyang inviting individuals interested in working on AI kernel generation to reach out. Standard Kernel's focus on low-level optimization for NVIDIA H100 GPUs positions it to address the growing demand for efficient and high-performance computing in the rapidly evolving artificial intelligence industry.