TSMC’s N5P 5nm Node Will Be Used By Japanese Firm For Automotive Chips

This is not investment advice. The author has no position in any of the stocks mentioned. WCCF TECH INC has a disclosure and ethics policy.

As the ongoing shortage of automobile chips continues to plague the market and the resolution of this shortage appears to be unlikely for at least a couple of months, Japanese firm Socionext has announced its decision to build chips for vehicles through the Taiwan Semiconductor Manufacturing Company's (TSMC) advanced 5nm semiconductor manufacturing process node.

Socionext sells chips that are designed exclusively to be used in electric vehicles and its products support advanced driver assistance systems (ADAS) and other autonomous driving applications. First sample products will be available next year revealed the company's management through a press release in morning time Japan today, where it also praised TSMC's prowess with leading-edge chip fabrication.

TSMC, Samsung Rumored To Hike Chip Prices By As Much As 20%

First Chips To Be Available For Sampling In Q2 2022 Reveals Socionext

The Japanese firm's decision to use TSMC's 5nm process for its processors highlights the increasing demand for these products in the market. This demand has grown steadily over the course of the past few years as automakers transition towards autonomous driving systems and look to integrate data from multiple sensors in their vehicles.

This in turn has strained global foundries and auto chip designers - an event that grew in its magnitude in the post-pandemic economic environment that firms have been facing for more than a year. Initially, as demand for automobiles slowed down, manufacturers reduced their orders for the processors only to increase them later on as China started to recover and demand for vehicles in the country picked up. However, by then, foundries' capacity for various process nodes had already been booked, and the manufacturers were forced to halt their production as chips for vehicles ran out.

To solve this, governments stepped in and requested Taiwan to help their automakers overcome the ongoing crisis. At this front, the Taiwanese government and TSMC agreed on a 'Super Hot' production run which requires additional capital investment and reduces the yield rate of their machines. This run should beef up chip supply in a couple of months, and the source of funds to enable it is currently unknown.

For its latest chips, Socionext will use TSMC's N5P process node. The N5P is the company's second iteration of its 5nm chip process and it delivers several improvements over its predecessors, including those for power consumption, transistor density and performance. TSMC's statements indicate that via middle-of-line and front-end-of-line optimizations, the second generation process node will allow for either a 7% performance or a 15% power efficiency gain. Furthermore, in statements made at the start of last year, the Taiwan-based foundry also indicated that it expects all of its major customers to shift to N5P this year – an important claim when we keep in mind Socionext's timeline of first sample availability.

Apple’s M1 Ultra Uses TSMC’s ‘InFO_LI’ Packaging Method, Enabling Lower Cost in Mass Producing the Custom SoC

In its decision to use the N5P, Socionext's management has cited the process's gains over the older 7nm process node. N5P will increase transistor density by 80% over the first-generation 7nm and provide a 20% performance gain or a 40% efficiency gain.

Judging by the fact that the first samples of the automotive chips will be available to Socionext's customers in the second quarter of next year, looks as if TSMC's existing customers that are relying on the N5P will continue to get the first chips that roll off the production line.

The Japanese firm has branches all over the world, including those in Santa Clara, California and Detroit, Michigan. Its 5nm automotive chip will use interconnect technology from Californian firm Arteris IP and Synopsis' HBM2E IP for memory throughput. The HBM2E is capable of operating at 3.6 Gbps heterogenous integration through short 2.5D interposer package routes.