Fixing AI's Most Expensive Problem โ Junchen Jiang, Tensormesh CEO
Following Tensormesh's $๐ฎ๐ฌ๐ funding announcement from AMD Ventures, @CoreWeave, NVentures (NVIDIA), Valley Capital Partners, and Laude Ventures, our CEO and Co-Founder Junchen Jiang sat down with the TechBeats podcast to talk about the journey from a research insight at the University of Chicago to building the infrastructure layer behind the "Big Data of AI.
Junchen shares how "KV cache," once a dismissed concept in academic research, became the epicenter of AI acceleration, and how Tensormesh built the first caching-accelerated inference platform for enterprises and the GPU ecosystem.
๐ง๐ต๐ฒ ๐ฐ๐ผ๐ป๐๐ฒ๐ฟ๐๐ฎ๐๐ถ๐ผ๐ป ๐ฐ๐ผ๐๐ฒ๐ฟ๐:
โ00:00 Introduction
โ01:08 ย The Origin Story and Insight Behind LMCache
โ06:27 ย "๐๐ฉ ๐๐ฎ๐ฐ๐ต๐ฒ" term origin in academia
โ11:25 ย Day 2 of AI: The Inference Bottleneck ("KV Cache")
โ14:59 ย The Secret to Sustaining a Thriving Open Source Community
โ16:42 ย Tensormesh Inference Platform V1 and $0 Cached Tokens
โ18:28 ย Designing for an Open, Agnostic Ecosystem (Serverless to On-Prem)
โ21:42 ย The Relationship with Hardware Vendors
โ24:01 ย KV Cache in the Agentic Era
โ27:38 ย What GPU and AI Cloud Backing Says About Tensormesh
โ29:49 ย Closing Thoughts: Where Tensormesh Goes Next
Useful links:
- LMCache (open source): https://github.com/LMCache/LMCache
- Press release: https://www.tensormesh.ai/blog-posts/...
