Quadric is keeping pace with the shift from cloud AI to on-device inference, and it’s paying off

Companies and governments are looking for tools to run AI locally with the goal of reducing cloud infrastructure costs and building sovereign capabilities. Quadric, an IP chip startup founded by veterans of early Bitcoin mining company 21E6, is trying to fuel this shift, expanding beyond cars to laptops and industrial devices, with on-device inference technology.

This expansion is already paying off.

Quadric will record between $15 million and $20 million in licensing revenue in 2025, up from about $4 million in 2024, CEO Veerbhan Kheterpal (pictured above, center) told TechCrunch in an interview. The company, which is headquartered in San Francisco and has an office in Pune, India, is targeting up to $35 million this year as it builds its equity-based hardware AI business. That growth has boosted the company, which now has a post-cash valuation of between $270 million and $300 million, up from about $100 million in a 2022 Series B, Kheterpal said.

It also helped attract investors to the company. Last week, Quadric announced a $30 million Series C round led by ACCELERATE Fund, managed by BEENEXT Capital Management, bringing its total funding to $72 million. The increase comes as investors and chipmakers look for ways to push more AI workloads from centralized cloud infrastructure to on-premises devices and servers, Kheterpal told TechCrunch.

From cars to everything

Quadric started out in the automotive space, where on-device AI can power real-time functions like driver assistance. The proliferation of transformer-based models in 2023 has pushed inference into “everything,” Kheterpal said, creating a sharp turn in business over the past 18 months as more companies try to run AI on-premises rather than relying on the cloud.

“Nvidia is a powerful platform for AI in data centers,” Kheterpal said. “We were looking to build a similar CUDA-like or programmable AI infrastructure on the device.”

Unlike Nvidia, Quadric doesn’t manufacture the chips itself. Instead, it licenses IP for a programmable AI processor, which Kheterpal describes as a “blueprint” that customers can embed into their own silicon, along with a suite of software and a toolchain to run models, including vision and audio, on the device.

The startup’s customers include AI-enabled printers, cars and laptops, including Kyocera and Japanese auto supplier Denso, which makes chips for Toyota vehicles. The first products based on Quadric technology are expected to ship this year, starting with laptops, Kheterpal told TechCrunch.

However, Kheterpal said Quadrick is now looking beyond traditional commercial deployments and to markets to explore “sovereign AI” strategies to reduce reliance on existing US-based infrastructure. The startup is exploring customers in India and Malaysia, and considers Moglix CEO Rahul Garg as a strategic investor who will help shape its “sovereign” approach in India, he added. Quadric employs approximately 70 people worldwide, including about 40 in the United States and about 10 in India.

The trend is driven by the rising cost of centralized AI infrastructure and the difficulty many countries face in building large-scale data centers, which has led to increased interest in “distributed AI” setups where inference is run on laptops or small servers inside offices rather than relying on cloud services for each query, Kheterpal said.

The World Economic Forum noted this shift in a recent article, as AI reasoning moves closer to users and away from purely centralized architectures. Likewise, EY said in a November report that the sovereign AI approach has gained momentum as policymakers and industry groups seek to bolster domestic AI capabilities that include computing, modeling and data, rather than relying entirely on foreign infrastructure.

The challenge for chipmakers is that AI models evolve faster than hardware design cycles, Kheterpal said. Customers need programmable processor IP that can keep up with software updates rather than requiring costly redesigns every time infrastructures shift from previous vision-focused models to current switch-based systems, he said.

Quadric is positioning itself as an alternative to chip vendors like Qualcomm, which typically uses its own AI technology inside its own processors, as well as IP vendors like Synopsys and Cadence, which sell neural processing engine blocks. Qualcomm’s approach could lock customers into their own silicon, while traditional IP vendors offer engine blocks that many customers find difficult to program, Kheterpal said.

Quadric’s programmable approach allows customers to support new AI models with software updates rather than hardware redesigns, giving an advantage in an industry where chips can take years to develop, while model architectures change in a matter of months today.

However, Quadric is still in the early stages of its creation, with only a few signed customers so far, and much of the long-term upside depends on converting today’s licensing deals into high-volume shipments and recurring royalties.