Array
(
    [content] => 
    [params] => Array
        (
            [0] => /forum/threads/kaiser-intel-is-too-big-to-fail.23495/
        )

    [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] => 2021770
            [XFI] => 1050270
        )

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

Kaiser: Intel is Too Big to Fail

Daniel Nenni

Admin
Staff member
The conversation revolves around the U.S. government's potential investment in Intel and MP Materials, with Winston Maas suggesting it resembles a sovereign wealth fund, though the speaker, a venture capitalist, views it more as a utility play aimed at enhancing domestic refining capacity for national security. The discussion highlights geopolitical pressures, particularly related to Taiwan and TSMC, emphasizing the need for U.S. self-sufficiency in semiconductor manufacturing. The speaker supports the government's stake in Intel, citing its potential to become a TSMC substitute or backup, leveraging its technology and encouraging commitments from Nvidia and AMD for fab and packaging. However, transforming Intel into a third-party refining model and building a domestic supply chain is seen as a decade-long project, not a quick fix.

The speaker notes Intel’s recent struggles—missed earnings, layoffs, and a "mess"—but argues its size makes it "too big to fail," justifying government support to shift it toward a TSMC-like model. Investment viability is left to public market experts, though the national security angle is deemed critical. On Nvidia, despite strong earnings driven by high demand for Blackwell chips (sold out into next year due to TSMC capacity constraints), the speaker sees supply chain bottlenecks, not demand issues, as the challenge. This aligns with strong growth in related firms like Core and Lambda Labs, suggesting the AI sector is in its early stages.

Regarding China, the speaker acknowledges the technical difficulty of replicating Nvidia’s Blackwell chip domestically, especially with future advancements like the Rubin chipset and TSMC’s 2nm production. While short-term imitation is challenging, the gap with Chinese alternatives is widening, reinforcing the strategic importance of U.S. investments in Intel to mitigate geopolitical risks tied to Taiwan’s semiconductor dominance.

 
The signal to noise ratio regarding the semiconductor industry in the financial press is horrendous. :rolleyes: Perhaps it always has been, but the AI explosion and Intel's problems has made it more obvious.

I was at the Global Foundries Summit today and was asked about politics and semiconductors. It really is entertaining. I'm wondering if GF will get political attention soon? Chuck Schumer is the political force behind GF and Chuck is not popular with the current administration. I suggested to GF that they get a meeting with Howard Lutnick and get the party started. :ROFLMAO:

GF CEO Tim Breen has an interesting background: https://www.linkedin.com/in/tim-breen-79a24b/

14 years with Mubadala, who holds the majority 82% shares of GlobalFoundries. The current administration should not have a problem with that.
 
Everybody is a semiconductor expert now. :ROFLMAO:

By the way, while people are talking about the future of Intel, TSMC, GlobalFoundries, and Samsung in US based semiconductor manufacturing, why isn’t the Trump administration doing anything about Wolfspeed, a true American IDM currently under Chapter 11 bankruptcy protection?

Wolfspeed is an important company in the semiconductor supply chain and is in desperate need of support.

Does the Trump administration really have any interest in understanding the semiconductor industry?
 
By the way, while people are talking about the future of Intel, TSMC, GlobalFoundries, and Samsung in US based semiconductor manufacturing, why isn’t the Trump administration doing anything about Wolfspeed, a true American IDM currently under Chapter 11 bankruptcy protection?

Wolfspeed is an important company in the semiconductor supply chain and is in desperate need of support.

Does the Trump administration really have any interest in understanding the semiconductor industry?
Is the goal to find the least successful companies and then back them?
 
The partial nationalization of Intel might have presented a timely chance to divide the company. Maintaining its design and manufacturing arms under one roof is proving unsustainable. ARM and TSMC have thrived precisely because they don’t challenge their customers. Intel, on the other hand, risks being trapped in a precarious situation: competing simultaneously with the very companies it supplies and the foundries it relies or may rely on. I’d genuinely welcome any counterarguments, because from where I stand, it feels like we’re witnessing the birth of something dangerously unwieldy.
 
The best way for USG to save Intel is to mandate Apple and nVidia working with Intel and set a target of on-shoring 20% of advanced foundry manufacturing volume (N2 and beyond) to Intel fabs by 2028 in exchange for zero foreign chip tariff for the time being. Intel seems incapable of gaining key customer commitment. USG should focus on this issue and make sure the US fabless companies are helping drive Intel fabs like what they have done with TSMC. Taking equity share in Intel makes no sense and does noting for Intel.
 
The partial nationalization of Intel might have presented a timely chance to divide the company. Maintaining its design and manufacturing arms under one roof is proving unsustainable. ARM and TSMC have thrived precisely because they don’t challenge their customers. Intel, on the other hand, risks being trapped in a precarious situation: competing simultaneously with the very companies it supplies and the foundries it relies or may rely on. I’d genuinely welcome any counterarguments, because from where I stand, it feels like we’re witnessing the birth of something dangerously unwieldy.
[This probably needs a whole new thread]

There are some indications that ARM is looking at producing complete chiplets - i.e. that it may move into actually creating designs in silicon. So the clear separation they've maintained up to now may not last. Indeed, some of their customers might be putting them under pressure to do so. It's not difficult to imagine that while their main customers want ARM to stay away from ever producing silicon, there may be a subset who prefer this. It will be interesting to see how this develops.

I agree in principle that it's better to keep clear boundaries to avoid conflicts of interest. But reality might become a little messier in some cases.
 
The partial nationalization of Intel might have presented a timely chance to divide the company. Maintaining its design and manufacturing arms under one roof is proving unsustainable. ARM and TSMC have thrived precisely because they don’t challenge their customers. Intel, on the other hand, risks being trapped in a precarious situation: competing simultaneously with the very companies it supplies and the foundries it relies or may rely on. I’d genuinely welcome any counterarguments, because from where I stand, it feels like we’re witnessing the birth of something dangerously unwieldy.
So how would you react if ARM chose to design and sell chips directly rather than selling IP?
 
So how would you react if ARM chose to design and sell chips directly rather than selling IP?
It depends on the business model. If ARM sells chiplets to companies like NVIDIA, that’s straightforward. But ARM selling Cortex M4 microcontrollers, for example, would obviously create complications. Similarly, it’s hard to imagine Intel selling laptop processors while also making Snapdragon chips for Qualcomm or Mx for Apple, there’s a clear conflict of interest.
The two businesses of Intel should be totally free to act on each other's market, without one activity influencing or limiting the other one.
 
[This probably needs a whole new thread]

There are some indications that ARM is looking at producing complete chiplets - i.e. that it may move into actually creating designs in silicon. So the clear separation they've maintained up to now may not last. Indeed, some of their customers might be putting them under pressure to do so. It's not difficult to imagine that while their main customers want ARM to stay away from ever producing silicon, there may be a subset who prefer this. It will be interesting to see how this develops.

I agree in principle that it's better to keep clear boundaries to avoid conflicts of interest. But reality might become a little messier in some cases.
This is an interesting strategy shift which I was unaware of. If ARM customers see it as a way to reduce licence costs or time to market and not direct competition, it may work. The case of Intel is somewhat different.
 
The best way for USG to save Intel is to mandate Apple and nVidia working with Intel and set a target of on-shoring 20% of advanced foundry manufacturing volume (N2 and beyond) to Intel fabs by 2028 in exchange for zero foreign chip tariff for the time being. Intel seems incapable of gaining key customer commitment. USG should focus on this issue and make sure the US fabless companies are helping drive Intel fabs like what they have done with TSMC. Taking equity share in Intel makes no sense and does noting for Intel
 
The best way for USG to save Intel is to mandate Apple and nVidia working with Intel and set a target of on-shoring 20% of advanced foundry manufacturing volume (N2 and beyond) to Intel fabs by 2028 in exchange for zero foreign chip tariff for the time being. Intel seems incapable of gaining key customer commitment. USG should focus on this issue and make sure the US fabless companies are helping drive Intel fabs like what they have done with TSMC. Taking equity share in Intel makes no sense and does noting for Intel.
Samsung beating Intel for Tesla’s business shows how hard it would be for Apple or Nvidia to switch to Intel. If Intel couldn’t win Tesla, which should’ve been much easier it’s unlikely they’ll succeed with Apple or Nvidia.

As for the U.S. government owning a stake in Intel, I agree it doesn’t make much sense, except maybe to keep LBT as CEO and prevent Intel’s stock from crashing.
 
Back
Top