Array
(
    [content] => 
    [params] => Array
        (
            [0] => /forum/index.php?threads/tsmc%E2%80%99s-n3p-to-enter-mass-production-in-the-second-half-of-next-year-rumored-inclusion-of-tesla-in-customer-list.19376/
        )

    [addOns] => Array
        (
            [DL6/MLTP] => 13
            [Hampel/TimeZoneDebug] => 1000070
            [SV/ChangePostDate] => 2010200
            [SemiWiki/Newsletter] => 1000010
            [SemiWiki/WPMenu] => 1000010
            [SemiWiki/XPressExtend] => 1000010
            [ThemeHouse/XLink] => 1000970
            [ThemeHouse/XPress] => 1010570
            [XF] => 2021370
            [XFI] => 1050270
        )

    [wordpress] => /var/www/html
)

TSMC’s N3P to Enter Mass Production in the Second Half of Next Year, Rumored Inclusion of Tesla in Customer List

Daniel Nenni

Admin
Staff member

tsmc3-624x286.jpg


TSMC, the world’s leading foundry in the 3nm technology, is reportedly experiencing a surge in the number of New Tape-Outs (NTOs) for the 3nm family in 2024, with Clients such as MediaTek, AMD, NVIDIA, Qualcomm, and Intel.

Among the 3nm family, the N3P process, set for mass production in the second half of 2024, is also making significant progress. Rumors suggest that Tesla has been added to the list of customers, with plans to utilize the N3P for the production of next-generation Full Self-Driving (FSD) chips after its launch.

Currently, Tesla has placed orders with TSMC for numerous chips related to electric vehicles. For instance, the supercomputer chip “D1” is utilizing TSMC’s 7nm technology along with advanced packaging processes.

Reportedly, according to industry sources, Tesla’s older FSD chips were initially produced using Samsung’s 14nm process, later upgraded to Samsung’s 7nm process. Subsequently, considering design upgrades, production quality, and scale, Tesla has shifted its HW 4.0 autonomous driving chip production to TSMC, utilizing the 5nm technology family.

The latest information per the report indicates that Tesla has recently initiated a NTO process with TSMC, planning to utilize the N3P for the production of the fifth generation of self-driving vehicle chips. Market expectations are high, with the influx of relevant orders suggesting that Tesla has the potential to become one of TSMC’s major clients.

According to TSMC’s previously disclosed process roadmap, the N3P process is an advanced version within the 3nm family, scheduled for production in 2024. Compared to the N3E, the N3P boasts a 5% improvement in performance, a 5% to 10% reduction in power consumption, and a 1.04 times increase in chip density.

 
Tesla has recently initiated a NTO process with TSMC, planning to utilize the N3P for the production of the fifth generation of self-driving vehicle chips. Market expectations are high, with the influx of relevant orders suggesting that Tesla has the potential to become one of TSMC’s major clients.
Number of iphones sold per year is roughly 100X the number of teslas sold per year. Are the number of cutting edge chips per car enough to make Tesla's volume more than 10% that of apple?
 
Number of iphones sold per year is roughly 100X the number of teslas sold per year. Are the number of cutting edge chips per car enough to make Tesla's volume more than 10% that of apple?

Agreed. Automotive looks like a bubble to me. We have not had so many car companies since the early 1900s and it will end just like it did back then with a handful of car companies eating the others, absolutely.
 
Number of iphones sold per year is roughly 100X the number of teslas sold per year. Are the number of cutting edge chips per car enough to make Tesla's volume more than 10% that of apple?
Just above 1% I think.

The FSD Computer v3 has 2 x 260mm 14nm chips today. Presumably next gen v4 will still have two relatively large dies (for redundancy). Tesla is already selling cars at a rate of 2 million per year, = 4 million “large” chips. Compares to ~ 320 million Apple products annually with Apple Silicon. In 2025 Tesla could hit 4 million cars - but that’s top end. Maybe 2-3% in terms of silicon area.

Separately I’m a little surprised that Tesla would going all the way to N3P for FSDv4 after using Samsung 14nm for FSDv3.
 
Number of iphones sold per year is roughly 100X the number of teslas sold per year. Are the number of cutting edge chips per car enough to make Tesla's volume more than 10% that of apple?
Great point. Do you think, Tesla cloud could add meaningful volume to TSMC - given Cloud server chips are quite large. Last year I came across Jefferies report that estimates by 2026 wafer consumption to produce server chips will be close to smartphone wafer consumption (though server volume is a fraction of smartphone volume).
 
Back
Top