Chipadd fabricates ultra‑efficient copper micro‑cooling structures directly onto semiconductor dies using additive manufacturing, enabling integrated single‑phase or two‑phase liquid cooling at the chip junction. This reduces thermal throttling, lowers operating temperatures, and extends chip lifespan, allowing AI data‑center operators and hardware manufacturers to achieve higher GPU utilization, greater compute density, and lower energy consumption without redesigning existing packages.
Funding
Funding not disclosed
Founders
Product
Problem
AI data centers face thermal constraints that limit GPU utilization, cause performance throttling, and reduce chip lifespan, especially as compute power density increases. Traditional cooling solutions like cold plates add bulk and complexity, hindering efficient heat removal at the chip level.
Solution
Chipadd addresses these challenges by fabricating ultra‑efficient copper cooling structures directly onto semiconductor dies using additive manufacturing. The integrated cooling channels enable both single‑phase and two‑phase liquid cooling at the chip junction, lowering operating temperatures and minimizing thermal throttling. By reducing the thermal resistance at the source, the technology allows GPUs to sustain higher utilization rates, extends chip lifetime, and shortens data‑center commissioning time due to lower overall energy requirements. Chipadd’s approach is compatible with existing GPU and CPU packages, offering a drop‑in upgrade path without redesigning the entire system.
Target Audience
Primary customers are AI data‑center operators and hardware manufacturers seeking to increase compute density and efficiency while maintaining chip reliability.
Features
- Additive‑manufactured copper micro‑cooling structures deposited directly on the die for optimal heat transfer
- Supports both single‑phase and two‑phase liquid cooling regimes to match workload demands
- Enables higher sustained GPU utilization (30K–50K GPUs) by reducing thermal throttling
- Extends chip lifespan 2–3× through lower operating temperatures
- Reduces data‑center energy consumption and accelerates commissioning timelines
- Designed for seamless integration with current GPU and CPU form factors, requiring no major system redesign