Diamond Foundry Unveils 200-mm-Einkristalldiamanten: A New Material That Could Transform AI Hardware Cooling
While diamonds are typically reserved for jewelry, Diamond Foundry is aiming to move them into the high-performance computing realm. On November 5, 2026, the company announced the presentation of its first continuous 200-mm-Einkristalldiamanten, reaching a scale comparable to standard semiconductor wafers. Given diamond's exceptional thermal conductivity, this material could play a crucial role in managing heat for next-generation AI accelerators, though it remains an unpolished crystal rather than a production-ready substrate.
The presentation of the 200-mm-Einkristalldiamanten took place during the CarbonSemi 2026 event in Ningbo, China. CEO Martin Roscheisen introduced the achievement, which is rooted in over 25 years of research by Stefan Gsell and Martin Fischer. Both researchers joined Diamond Foundry in 2022 following the acquisition of Augsburg Diamond Technology. This development marks a significant milestone in growing large-scale single-crystal diamonds, moving beyond the smaller dimensions that previously limited their industrial application.
The scale of the crystal is the key factor here. While synthetic diamonds have been used in technical applications for some time, the 200-mm diameter opens the door to manufacturing larger technical components from a single crystal. This size corresponds to the classic 8-inch wafer format widely used in analog circuits, sensors, and power electronics. However, a 200-mm crystal is not a direct replacement for silicon wafers; its vastly different material properties would require specialized processing steps for integration into existing manufacturing lines.
Addressing the Thermal Challenges of AI
Modern AI accelerators face a growing thermal challenge: high power density on extremely small surface areas leads to intense local temperature spikes, known as hotspots. Diamond could serve as a highly efficient heat spreader in these scenarios. It is important to clarify that diamond would not replace the cooling system itself; instead, it would distribute heat across a larger area, allowing existing heat sinks or liquid cooling systems to remove it more effectively. This could significantly improve the thermal conditions within a processor package.
Despite the potential, several hurdles remain before this material becomes a commercial reality. The presented crystal is unpolished and lacks data regarding production yield, defect density, or the final cost of a 200-mm substrate. Furthermore, large single crystals are prone to lattice defects that can compromise mechanical, thermal, or electrical properties. For practical use, the material requires precise cutting, grinding, and polishing to meet industrial quality standards.
The Path to Industrial Scalability
Diamond Foundry is already expanding its production capabilities, including a multi-billion dollar project in Spain linked to the European Chips Act. While the company suggests that 300-mm crystals might be achievable in the future, such claims remain technical projections for now. The ultimate success of this technology depends on whether Diamond Foundry can achieve economic competitiveness through high-quality, reliable, and cost-effective mass production.
The thermal management challenges posed by the rapid expansion of artificial intelligence are driving a search for materials that can handle extreme heat densities. Diamond Foundry has announced the development of 200-mm-Einkristalldiamanten, a significant leap in synthetic diamond manufacturing that aims to address the cooling bottlenecks in next-generation AI hardware.
Scaling Up Diamond Substrates
While diamond has long been recognized for its exceptional thermal conductivity—far surpassing traditional silicon—the practical application in large-scale semiconductor manufacturing has been limited by the difficulty of producing large, high-quality single crystals. The introduction of 200-mm-Einkristalldiamanten represents a major scaling milestone. Moving to a 200-mm diameter format brings diamond substrates closer to the standard wafer sizes used in conventional CMOS manufacturing, potentially easing the integration process into existing high-performance computing (HPC) workflows.
The primary advantage of using these large-format single-crystal diamonds lies in their ability to act as highly efficient heat spreaders. As AI accelerators and high-power chips operate at increasingly higher power densities, traditional cooling solutions often struggle to dissipate heat fast enough to prevent thermal throttling. Diamond's unique physical properties allow it to move heat away from the active silicon layers much more effectively than standard materials.
Impact on AI and High-Performance Computing
The potential impact of this technology is most significant in the data center and AI training sectors. As models grow in complexity, the hardware required to run them generates concentrated heat that can limit performance. By integrating 200-mm-Einkristalldiamanten into the packaging or substrate layers of AI chips, manufacturers could theoretically achieve higher clock speeds and more stable performance under heavy workloads. This development is not just about cooling a single component, but about enabling a new tier of high-density computing architectures.
Гика
