US-based Diamond Foundry announced on November 5, 2026 that it had produced a single-crystal synthetic diamond 200 millimetres, or 20 centimetres, across. The company describes it as the largest single-crystal diamond produced so far; that record claim is the company’s own statement.
Diamond is usually associated with rings and jewellery. Its importance, however, extends beyond its appearance: it could play a role in computing, particularly in removing heat from processors.
Diamond disc image supplied with the article.
What does diamond have to do with computers?
Natural and laboratory-grown diamond have the same carbon crystal structure. One important property of high-purity diamond is its high thermal conductivity. Under suitable material conditions, it can conduct heat roughly five times better than copper, although the value depends on purity, structure and measurement conditions.
High-performance AI processors generate substantial heat. Removing it is important for efficient operation and managing cooling requirements. Diamond-based materials have been studied and used for cooling electronic components for many years.
Why does 200 millimetres matter?
The distinguishing features of Diamond Foundry’s announcement are the size and continuity of the crystal. The company says it achieved a single, continuous crystal 200 millimetres across rather than a structure assembled from smaller crystals. This scale is significant for compatibility with semiconductor manufacturing processes.
The announcement does not mean that AI chips made entirely from diamond have entered mass production. Processing the crystal, adapting it to manufacturing and assessing commercial applications are separate steps.
300 millimetres is not yet a product announcement
The company’s CEO describes 300 millimetres as a foreseeable next step. It should not be presented as an already-produced 300-millimetre product or a firm delivery timetable.
Once associated primarily with jewellery, diamond could become an important material for more efficient cooling of AI processors.
Praesentium eum architecto. Qui cumque quo autem enim et neque sed. Repudiandae dolorem dolor aliquam itaque placeat quia sunt odit ut. Perferendis libero optio reiciendis facilis ut minima. Quia voluptas molestiae. Vero voluptates et sunt qui animi dignissimos dolorem et quis.