๐จ๐ฝโ๐ Welcome to Episode 10 of “Tech Beats Unplugged” with Junchen Jiang, CEO & Co-Founder of Tensormesh.
$20M to Fix AI’s Most Expensive Problem.
Fresh off Tensormesh’s $20M funding announcement backed by AMD Ventures, CoreWeave, NVentures (NVIDIA), Valley Capital Partners, and Laude Ventures, Junchen joins Kosseila HD to unpack the journey from a research insight at the University of Chicago to building one of the fastest-growing AI infrastructure companies.
In this episode, we dive into the story behind LMCache, why KV cache evolved from a largely misunderstood concept in academia into one of the most critical layers of modern AI inference, and how Tensormesh is building an open, caching-accelerated inference platform for the next generation of AI.
We’re excited to share one of our most technical conversations yet.
Topics discussed:
- (00:00) Introduction
- (01:08) The Origin Story and Insight Behind LMCache
- (06:27) The Origin of “KV Cache” in Academia
- (11:25) Day 2 of AI: The Inference Bottleneck
- (14:59) The Secret to Sustaining a Thriving Open Source Community
- (16:42) Tensormesh Inference Platform V1 and $0 Cached Tokens
- (18:28) Designing an Open, Agnostic Ecosystem (Serverless to On-Prem)
- (21:42) Working with Hardware Vendors
- (24:01) KV Cache in the Agentic Era
- (27:38) What GPU & AI Cloud Backing Says About Tensormesh
- (29:49) Closing Thoughts: Where Tensormesh Goes Next
Show Notes
- LMCache (Open Source): https://github.com/LMCache/LMCache
- Tensormesh: https://www.tensormesh.ai/
- Tensormesh Funding Announcement: https://www.tensormesh.ai/blog-posts/
- Follow Tensormesh on X: https://x.com/tensormesh
- Connect with Junchen Jiang: https://www.linkedin.com/in/junchenjiang/
๐ About Junchen Jiang
Junchen Jiang is the CEO & Co-Founder of Tensormesh and the co-creator of LMCache, an open-source project developed from research at the University of Chicago. His work focuses on large-scale AI inference infrastructure, KV cache systems, and building an open ecosystem that spans GPUs, inference engines, storage, and enterprise AI deployments.
Follow Tech Beats Unplugged for more conversations with founders, researchers, and engineers building the future of AI infrastructure, open source, cloud, and startups.