Tensormesh

About Tensormesh

Tensormesh is an AI infrastructure optimization company that provides software to accelerate AI inference while reducing latency and GPU costs by up to 10x. The company emerged from stealth on October 23, 2025, with $4.5 million in seed funding led by Laude Ventures. Tensormesh was founded by faculty and PhD researchers from the University of Chicago, UC Berkeley, and Carnegie Mellon. The CEO and founder, Junchen Jiang, is a University of Chicago faculty member and a co-creator of LMCache, an open-source KV caching project. The founding team also includes Yihua Cheng (CTO) and Kuntai Du (Chief Scientist), both holding PhDs from the University of Chicago with expertise in AI systems and LLM inference.

The company's core technology commercializes the open-source LMCache project, which captures and reuses intermediate computation data that is typically discarded in AI inference processes. This KV-caching technique eliminates redundant computations, which is particularly effective for improving performance in chat interfaces and agentic systems. By productizing this caching technology, Tensormesh offers an enterprise-grade platform with enhanced usability, security, and manageability. This allows organizations to run AI on their own infrastructure—whether in the public cloud, private data centers, or hybrid environments—maintaining full control over their data and security. The software is available as both a SaaS solution and as standalone software.

Tensormesh's business model targets enterprises struggling with the high costs and complexity of AI inference. Instead of building costly in-house optimization systems, which can take significant time and engineering resources, clients can use Tensormesh's out-of-the-box solution. The platform is designed to be cloud-agnostic and integrates with leading inference engines like vLLM. The underlying LMCache technology has already been adopted and integrated by organizations such as Bloomberg, Red Hat, Redis, Tencent, Google, and Nvidia.