Quadric Extends Series C to $46 Million as Edge AI Demand Accelerates

Quadric has extended its Series C financing to $46 million through a second close led by the International Finance Corporation, the private-sector investment arm of the World Bank Group. The transaction raises Quadric’s total capital secured to approximately $90 million and strengthens its ability to commercialize programmable artificial-intelligence processor intellectual property for on-device inference.

The financing follows Quadric’s $30 million first close in January 2026. Existing investors Pear VC, Uncork Capital, and BEENEXT increased their participation, while Offline Ventures joined as a new investor. The company plans to use the additional capital to expand engineering, customer support, and go-to-market capacity across automotive, AI PCs, enterprise systems, humanoid robotics, wearables, and networking equipment.

Quadric’s core technology is the Chimera general-purpose neural processing unit, or GPNPU. Unlike fixed-function accelerators designed around a narrow set of operators, Chimera combines scalar, vector, and matrix-processing capabilities in a unified architecture. This programmability allows the processor to execute conventional neural-network layers, transformer operations, control code, and nonstandard workloads without repeatedly transferring data between separate CPU, DSP, and NPU blocks.

That architectural model addresses a growing problem in edge AI: model architectures evolve faster than custom silicon design cycles. A specialized accelerator may perform well on the networks available when the chip is taped out, but efficiency can decline when new attention mechanisms, activation functions, quantization formats, or multimodal pipelines emerge. Quadric’s approach shifts more adaptability into the instruction set, compiler, and software toolchain, enabling customers to update workloads after silicon deployment.

The company licenses processor IP rather than selling finished chips. Semiconductor vendors can integrate Chimera cores into system-on-chip designs and configure compute capacity for different thermal, power, and performance envelopes. Quadric says the architecture can scale to 864 trillion operations per second and is available with automotive-grade implementation options. Its software stack includes graph compilation, model conversion, scheduling, optimization, simulation, and runtime tools intended to reduce the time required to move from model development to production silicon.

The extended funding round also reflects the market’s transition from cloud-only AI toward distributed inference. Running models locally can reduce network latency, bandwidth consumption, and cloud-compute costs while improving privacy and operational resilience. These benefits are particularly important in vehicles, industrial systems, robotics, and consumer devices, where connectivity may be intermittent and response times must remain deterministic.

IFC’s participation is strategically notable because it is the institution’s first investment in AI chip technology. The investor has positioned the deal as a way to broaden access to AI infrastructure beyond hyperscale data centers, especially in emerging markets. Programmable edge processors could support locally deployed language, vision, and automation systems in regions where cloud access is expensive, constrained, or subject to data-sovereignty requirements.

Quadric enters the next phase with improving commercial indicators. The company reported that product revenue more than tripled in 2025 compared with 2024 and that it reached profitability before announcing the second close. The new capital is therefore aimed less at proving the architecture and more at scaling deployments, supporting design wins, and addressing additional device categories.

For semiconductor customers, the central value proposition is lifecycle flexibility. By combining programmable compute with an integrated compiler and development environment, Quadric is betting that chipmakers will prioritize architectures capable of absorbing future AI models rather than maximizing benchmarks for a fixed workload. The Series C extension gives the company additional resources to test that thesis across a rapidly expanding edge-inference market.

Bottom line: Execution will ultimately depend on silicon partners converting evaluations into high-volume programs, maintaining software compatibility across product generations, and demonstrating competitive performance per watt as transformer, multimodal, and robotics workloads continue to change across global edge markets worldwide.

Also Read:

Disaggregating AI Compute to Break the Tokens Barrier

Podcast EP336: How Quadric is Enabling Dramatic Improvements in Edge AI with Veer Kheterpal

Quadric’s Recent Momentum & Funding Success