Quadric rides the shift from cloud AI to on-device inference — and it’s paying off

In a significant move towards local AI processing, companies and governments are increasingly seeking tools that can operate AI directly on devices. Quadric, a chip-IP startup founded by former members of the pioneering bitcoin mining company 21E6, is at the forefront of this transition. The company is expanding its influence beyond the automotive sector into laptops and industrial devices with its innovative on-device inference technology. This strategic expansion is proving fruitful, as Quadric anticipates licensing revenues to soar between $15 million and $20 million in 2025, a remarkable increase from roughly $4 million in 2024, according to CEO Veerbhan Kheterpal. Based in San Francisco with an additional office in Pune, India, Quadric aims to reach up to $35 million in revenue this year, further solidifying its royalty-driven on-device AI business model. The company has seen a significant boost in valuation, now standing between $270 million and $300 million, up from approximately $100 million during its 2022 Series B funding round. Recently, Quadric completed a $30 million Series C funding round led by ACCELERATE Fund, managed by BEENEXT Capital Management, bringing its total funding to $72 million. This influx of capital reflects the growing interest from investors and chip manufacturers in migrating AI workloads from centralized cloud infrastructures to local devices and servers. Initially focused on the automotive industry, where on-device AI enhances real-time functionalities such as driver assistance, Quadric has identified a broader market potential. The rise of transformer-based AI models in 2023 has accelerated the push for running AI locally, shifting the business landscape dramatically over the past 18 months. Rather than manufacturing chips, Quadric licenses programmable AI processor IP, which serves as a blueprint for clients to integrate into their own silicon. This includes a comprehensive software stack and toolchain that facilitates on-device model execution for various applications, encompassing vision and voice technologies. The company's clientele spans diverse industries, including printers and automotive sectors, featuring prominent names like Kyocera and Japan's Denso, a supplier for Toyota. Looking ahead, Quadric is exploring opportunities beyond conventional commercial deployments by tapping into markets interested in 'sovereign AI' strategies, aimed at reducing dependence on U.S.-based technologies. The company is particularly focusing on potential partnerships in India and Malaysia, leveraging the expertise of strategic investors like Moglix CEO Rahul Garg to inform its India-centric initiatives. As the demand for localized AI solutions grows, driven by the rising costs of centralized infrastructures and the challenges of developing hyperscale data centers, Quadric is well-positioned. This shift towards 'distributed AI,' where inference tasks are processed on laptops or small on-premise servers, is gaining traction. Industry insights, such as those from the World Economic Forum, highlight this trend of moving AI inference closer to users, a sentiment echoed by reports from EY regarding the increasing emphasis on domestic AI capabilities. However, the fast-paced evolution of AI models presents a unique challenge for chip manufacturers. Kheterpal emphasizes the necessity for programmable processor IP that can adapt through software updates, avoiding the expensive redesigns that are often required as models transition from previous architectures to current transformer-based systems. Quadric aims to distinguish itself from competitors like Qualcomm, which typically integrates its AI technology into its processors, and traditional IP providers like Synopsys and Cadence, which offer neural processing engine blocks. The company's flexible, programmable approach enables clients to keep pace with evolving AI models through software updates instead of requiring extensive hardware redesigns, a crucial advantage in an industry where chip development timelines can span years, yet model architectures are continuously shifting. While Quadric is still in the early stages of scaling, the future looks promising as it works to convert existing licensing agreements into high-volume shipments and recurring revenue streams.