Quadric is riding 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 by reducing the cost of cloud infrastructure and building sovereign capabilities. Quadricchip-IP startup founded by veterans of the early bitcoin mining company 21E6, is trying to master these changes, extending from automotive to laptops and industrial devices, with inference technology on devices.

That expansion has paid off.

Quadric will deliver $15 million to $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 based in San Francisco and has offices in Pune, India, is targeting up to $35 million this year as it builds its AI business in royalty-driven devices. That growth puts the company, which currently has a post-money valuation of between $270 million and $300 million, from about $100 million in Series B 2022, Kheterpal said.

It also helps attract investors to the company. Quadric declare Last week, the Series C round of $30 million was led by ACCELERATE Fund, managed by BEENEXT Capital Management, bringing the 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 local devices and servers, Kheterpal told TechCrunch.

From automotive to everything

Quadric starting in automotivewhere on-device AI can power real-time functions like driver assistance. Kheterpal said that the deployment of transformer-based models in 2023 will drive inference to “everything”, creating a clear business inflection over the past 18 months as more companies try to run AI locally rather than relying on the cloud.

“Nvidia is a powerful platform for data center AI,” said Kheterpal. “We are looking to build an infrastructure similar to CUDA or programmable for AI on devices.”

Unlike Nvidia, Quadric doesn’t make its own chips. Instead, it licenses its programmable AI processor IP, which Kheterpal describes as a “blueprint” that customers can embed into their own silicon, along with a software stack and toolchain to run models, including vision and voice, on devices.

Initial customers include AI printers, cars and laptops, including Kyocera and Japanese car 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.

Instead, Quadric is now looking at traditional commercial deployments and into markets that are exploring “sovereign AI” strategies to reduce reliance on US-based infrastructure, Kheterpal said. It initially explored customers in India and Malaysia, he added, and considered Moglix CEO Rahul Garg a strategic investor who helped shape India’s “sovereign” approach. Quadric employs nearly 70 people worldwide, including about 40 in the US and about 10 in India.

The push is driven by the rising cost of centralized AI infrastructure and the difficulty many countries have in building hyperscale data centers, Kheterpal said, leading to more interest in “distributed AI” setups where inference runs on laptops or small servers on premise in the office rather than relying on cloud-based services for every query.

World Economic Forum pointed for this change in the new article, because AI inference is closer to the user and away from a purely centralized architecture. Likewise, EY said in a November report that the sovereign AI approach has gained traction as policymakers and industry groups push for domestic AI capabilities that encompass computation, models, and data, rather than relying entirely on foreign infrastructure.

For chipmakers, the challenge is that AI models evolve faster than hardware design cycles, Kheterpal said. He says customers need programmable processor IP that can keep software updated rather than requiring costly redesigns every time the architecture moves from the previous vision-focused model to today’s transformer-based systems.

Quadric has positioned itself as an alternative to chip vendors such as Qualcomm, which typically use AI technology in its own processors, as well as IP suppliers like Synopsys and Cadence, which sell neural processing engine blocks. Kheterpal said Qualcomm’s approach can lock customers into their own silicon, while traditional IP suppliers offer machine blocks that many customers find difficult to program.

Quadric’s programmable approach allows customers to support new AI models through software updates rather than redesigning hardware, providing an advantage in an industry where chip development can take years, while architecture model shift in a few months the present day.

Still, Quadric remains early in its development, with few customers signed so far and much of its long-term value dependent on turning current licensing deals into high-volume shipments and recurring royalties.