TSMC has begun producing semiconductors for Tesla’s supercomputer Dojo.
On the 5th (local time), the Liberty Times cited sources that TSMC announced this production commencement at a recent technology conference in California, USA.
TSMC has begun manufacturing semiconductors for Tesla’s supercomputer Dojo.
Dojo is an artificial intelligence supercomputer that trains autonomous driving software using data and video footage from Tesla’s vehicles.
Tesla plans to build a 10MW data center at the Gigafactory in Austin, Texas, to train autonomous driving software.
Sources say that TSMC provides a system with computational performance more than 40 times higher than the current one through an advanced process called Chip-on-Wafer-on-Substrate (CoWoS).
Another source revealed that TSMC plans to create a wafer system by integrating the Chip-on-Wafer-on-Substrate process technology and the System on Integrated Chips (SoIC) process by 2027.
The wafer system provides a 40-fold increase in computational ability, more than 40 photomasks, and space for over 60 high-bandwidth memory (HBM) chips.
TSMC announced at the technology conference that it plans to start producing semiconductors with a 1.6-nanometer process in the second half of 2026. This is interpreted as an intermediate step between the 2-nanometer process to be applied in 2025 and the 1.6-nanometer process to be applied in 2027.
Most Commented