AI news story
Future AI chips could be built on glass
Human-made glass is thousands of years old. But it’s now poised to find its way into the AI chips used in the world’…
Editor's take
Absolics is developing AI chips utilizing glass substrates, moving away from traditional silicon. This innovation aims to improve chip performance and efficiency, particularly for the demanding workloads of large-scale AI computation in data centers.
The significance lies in addressing the physical limitations of silicon, which is approaching its scaling limits. By employing glass, Absolics could enable denser chip architectures and better thermal management, crucial for the power-hungry AI models like those powering services from Microsoft Azure or Google Cloud. This could offer a competitive edge in the ongoing AI hardware arms race.
Future developments to monitor include the actual manufacturing yields and cost-effectiveness of these glass-based chips compared to established silicon foundries. The industry will also be watching if other major chip manufacturers, such as TSMC or Intel, adopt similar glass substrate technologies, or if Absolics carves out a niche for specialized AI accelerators.