US-based Atomica Launches AI Thermal Microstructures Platform

en.Wedoany.com Reported - Atomica, a US-based micromachining foundry, has announced the launch of its AI Thermal Microstructures Platform. This platform brings cooling capabilities closer to the heat sources of AI hardware, targeting high-density AI computing, advanced packaging, photonics systems, and other hardware applications. It marks the latest product line following the AI Optical Connectivity, AI Optical Source, and AI Photonic Integration platforms.

As the density of AI computing systems continues to rise, more heat is generated within smaller spaces, posing greater challenges for thermal management inside dense packaging. Traditional solutions such as air cooling, heat sinks, cold plates, and data center cooling remain important, but localized heat and thermal gradients may require package-level thermal management solutions.

The platform integrates technical approaches including microfluidic cooling, thermal vias, interposers, sensing structures, bonding, and wafer-level packaging, enabling coolant to be directed near high-heat regions for heat distribution, pressure and flow monitoring, isolation of temperature-sensitive components, and thermal control within advanced packaging.

Target applications for the platform include liquid-cooled AI accelerator modules, GPU and chiplet packaging, high-bandwidth memory integration, photonic interposers, optical engines, co-packaged and near-packaged optics, external laser source systems, edge AI systems, industrial electronics, defense applications, and high-reliability sensor modules.

Thermal management is also a critical concern in optical integration. As lasers and optical engines move closer to processors and switches, temperature gradients and thermal drift can affect performance and stability. Micromachined thermal structures can be used to manage these effects in compact optical systems.

Customers can start with feasibility structures or thermal test vehicles and progress through engineering samples, functional prototypes, qualification lots, and production phases.