Tesla is reportedly hiring TSMC to produce 3nm chips for autonomous driving
One of billionaire Elon Musk’s companies, Tesla, is reportedly hiring TSMC, a Taiwanese factory for the production of chips for the needs of a new 3 nm processor intended for autonomous driving. The new chip will likely be manufactured on an improved N3P manufacturing process, just in time for one of the upcoming fully autonomous driving (FSD) hardware, if the latest rumors are anything to go by.
This is surprising, because in this way Tesla plans to use a production process that is not intended for the automotive industry for its FSD SoC, writes the China Times portal.
The TSMC N3P process is the latest generation manufacturing technology that offers high performance, high transistor density and relatively low power consumption. For demanding processors like the Tesla FSD hardware, the N3P process could be just what the company needs. Apparently, several companies have already decided to use this TSMC production process, so Tesla will not be alone in its intention. However, there is one catch regarding N3P and FSD chips: when exactly does Tesla plan to use this technology?
Previous media reports indicated that Tesla plans to use TSMC N4 technology to build its FSD hardware platform 4 in Arizona during 2024. With TSMC having to delay chip production at its Fab 21 plant until 2025, Tesla decided to either transfer its FSD 4 design to N3P and manufacture it in Taiwan, or slow down production of FSD 4 hardware in the US for six or more months. Perhaps the company is also preparing an FSD 4.5 or even an FSD 5.0 platform that will use N3P, but this is just speculation.
Since information about Tesla’s use of TSMC N4 or N3P technology comes from unofficial sources, it should be taken with a grain of salt. It should also be noted that neither the N4 nor the N3P processes are intended for automotive applications. Instead, TSMC offers N5A and N3A technologies designed for cars.
TSMC N5A manufacturing technology is qualified to the AEC-Q100 Grade 1 standard and is known for its rigorous criteria of chip quality and reliability. Chips that comply with the AEC-Q100 standard must function in extreme temperatures from -40 to +150 degrees Celsius and pass numerous reliability and durability tests to ensure long-term performance in harsh conditions, writes Tom’s Hardware. In addition, the N5A complies with ISO 26262 requirements for functional safety as well as IATF16949 for quality management, defect prevention and variation reduction. TSMC N3A technology should follow N5A in terms of quality, reliability and durability.
To advance its N3 and N5 manufacturing processes for automotive applications, TSMC had to modify both technologies, so it is not possible to use N3 or N5 IP for N3A and N5A technologies, which would make it strange for Tesla to design an FSD SoC or SiP on the processes which are not intended for the automotive industry if it plans to adopt appropriate production technologies later.