Login to Continue Learning
After securing a substantial contract with Samsung Foundry for the next-gen AI6 chips built on the 2nm process, Tesla now reportedly opted for Intel’s packaging services for its Dojo supercomputer. According to a report by ZDNet Korea, this decision means that Intel might also become a major supplier, providing breathing room for the company.
Interestingly, both Samsung and Intel offer independent chip and packaging services. Tesla’s choice of both companies suggests careful selection based on specific needs. It is claimed that the next-gen Dojo 3 supercomputer may separate “chip mass production from module packaging,” which could benefit both Samsung and Intel despite their competitive relationship in this segment.
Currently, Tesla cannot rely solely on TSMC because the custom D1 chips used in the Dojo supercomputer are large and require multiple chips arranged on a single wafer. Previously, TSMC’s SoW (System-in-Wafer) packaging provided benefits from redistribution layers, but given the high demand for CoWoS and its derivatives, it is not feasible to meet Tesla’s limited orders.
Intel, eager for breakthroughs, has shown readiness to offer packaging services for Dojo chips. Team Blue will likely use Intel’s EMIB (Embedded Multi-Die Interconnect Bridge) technology, which connects multiple dies using tiny bridges as an alternative to SoW and offers higher modularity. This flexibility in integration and fine-tuning may explain Tesla’s inclination towards including Intel in its supply chain.
📚 Reading Comprehension Quiz
Why did Tesla reportedly choose Intel’s packaging services for its Dojo supercomputer after securing a contract with Samsung Foundry?
Please login or register to take the quiz and earn points!
