Array
(
    [content] => 
    [params] => Array
        (
            [0] => /forum/index.php?threads/amd-nvidia-and-tsm-to-lead-intel.14479/
        )

    [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
)

AMD, Nvidia and TSM to lead Intel?

Arthur Hanson

Well-known member
With AI/ML coming on strong across almost all applications, the combination of AMD, Nvidia, and TSM is already set to be in a leadership position. All three are moving fast with all having excellent management and technical teams. Does Intel even stand a chance or continue to lose ground. Any thoughts about what companies will be the future of AI/ML which is set to dominate almost everything we touch or at least manage it. Is Gelsinger going to be able to compete against this formidable trio in AI/ML and almost everything else?
 
My view is that Intel has fallen behind and will not be able to get back in the race. Intel is now behind both TSMC and NVidia in market cap, and will probably fall behind Broadcomm soon. I anticipate that by the time 2025 rolls around (when Intel claims it will be at foundry parity), they won't even be in the top 5 companies by market cap, and they won't be in the top 3 in revenue. I think Intel stock is almost a short, but it's valuation relative to it's peers already reflects it's weaker position.
 
Nvidia is the most powerful company in AI/ML. What is what Nvidia has done but not by Intel? Nvidia provides total solutions. From all I can see, AI training was done on Nvidia’s workstations.

In Nvidia’s picture, GPU-CPU is just like big.LITTLE core: GPU handles all heavy computation, and CPU is reduced to a supporting role. If this is the case, ARM CPU is good enough.

It is hard for me to believe Intel would agree with Nvidia’s picture. In that sense, Intel should come up with its own version of the total solution to stay relevant in AI/ML.

Nvidia has many formidable competitors, one of them being Cerebras WSE2 (TSMC 7nm). I am amazed by the determination of this visionary CEO and by the daunting task ahead. With TSMC's technical supports, we might be listening to the next Jensen if works out.

 
The current situation could be best explained when comparing Intel's upcoming products with competition. Intel is basically trying to just max out it's performance on 8 cores while they will be working with inferior nodes compared to what TSMC has to offer. With server products it is even worse. Nvidia is making it's own ARM platform and AMD making 128core. Tom from Moore's law is dead even said that his sources say that AMD considers ARM products bigger threat than Intel is that's why AMD is heavily investing in new solutions for server products.
 
And another thing I wanted to add is that all the companies are fighting different types of battles. AMD is fighting for time, they need it to grow to competetive size. Intel on the other hand is fighting to repair processes and update internal culture so they can be more effective at meeting their goals.
 
With AI/ML coming on strong across almost all applications, the combination of AMD, Nvidia, and TSM is already set to be in a leadership position. All three are moving fast with all having excellent management and technical teams. Does Intel even stand a chance or continue to lose ground. Any thoughts about what companies will be the future of AI/ML which is set to dominate almost everything we touch or at least manage it. Is Gelsinger going to be able to compete against this formidable trio in AI/ML and almost everything else?

Had Pat Gelsinger properly partnered with TSMC I would have given Intel a much better chance. Now that Intel is directly competing with TSMC I see little chance. The only thing Pat's IDM 2.0 narrative has accomplished is to motivate TSMC even more. Do not poke the friendly dragon unless you want to get burned.
 
Back
Top