Quadric Secures $30M in Series C Funding for On-Device AI Technology

Quadric just closed a $30M Series C, oversubscribed, disciplined, and timed like someone who actually understands silicon cycles instead of chasing headlines. Founded in 2016, Quadric did not begin with a manifesto. Veerbhan Kheterpal, Nigel Drego, and Daniel Firu were building an autonomous agricultural robot and ran straight into physics. Existing processors could not deliver real-time performance, power efficiency, and flexibility at the edge. Instead of negotiating with broken tradeoffs, they built Chimera, a fully programmable general-purpose neural processing unit designed to survive model evolution, not fear it.

This round was led by the ACCELERATE Fund at BEENEXT Capital Management, with Hero Choudhary calling out Quadric's traction in Asia and the kind of execution that hints at a generational outcome. Uncork Capital and Pear VC returned again, with Jeff Clavier leaning in the way investors do when customers are not experimenting but shipping. New institutional backing from Volta, Gentree, Wanxiang America, Pivotal, and Silicon Catalyst Ventures rounds out a syndicate that knows hardware timelines, not just software hype.

What separates Quadric is not the $30M. It is the evidence. Product revenue more than tripled in 2025. Design wins are stacking across edge LLM inference, automotive, and enterprise computer vision. Kyocera is live. DENSO is deep. MegaChips is active. TIER IV is onboard. A major edge server LLM silicon provider in Asia is signed and quiet, which is usually where the real momentum hides. This is inference doing work on device, under power, on schedule.

Chimera scales from 1 TOPS to 864 TOPS, runs CNNs, Vision Transformers, and LLMs up to 30B parameters, supports 4 bit weights, wide AXI interfaces, and does it all in C++. One software stream across neural nets, DSP, and control. No juggling acts. Quadric put more than 70% of R&D into software from day one, which is why customers can move from engagement to production ready silicon in under six months.

Leadership matters. Veerbhan Kheterpal brings rare software to silicon fluency. Nigel Drego brings architectural depth that shows up in compilers as much as cores. Daniel Firu makes sure the product survives contact with manufacturing and customers. Steve Roddy knows how IP platforms actually scale. Add Joachim Kunkel, who helped build Synopsys IP into a $1.9B business, and the picture sharpens.