Array
(
    [content] => 
    [params] => Array
        (
            [0] => /forum/threads/baird-notes-supply-chain-talk-of-intel-fab-spinoff-with-tsmc-involved.22079/page-2
        )

    [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
)

Baird notes supply chain talk of Intel fab spinoff with TSMC involved

The above scheme can be somewhat "played" by TSMC and its customers. But I can think of a lot of ways that benefit Intel and Micron: 1) create R&D credit, which rewards U.S. based R&D, not just manufacturing. 2) these credits are sellable, exactly like EV credits.
anyone on the forum who has connections to intc government affair team, could you please fwd this idea to them?

IMHO, putting just tariffs won't really help IFS nor U.S. based R&D/manufacturing, as semi cycle is too long. By creating R&D credits and prepayment credit, $$ immediatelly flows to IFS (and micron, globalfoundries, and TSMC AZ plant, for that matter).
 
So now you think that Intel Foundry is hopeless to be kept in-house?

Do you understand a TSMC fab is very different from a Intel fab. Even if TSMC needs more capacity in US, why can't TSMC build it by itself instead of salvaging the money losing IFS?

I think Intel shareholders should not hope TSMC will come to bail them out. Finding a good CEO and a good business strategy are much more feasible and practical for the Intel.
I never say Intel Foundry was hopeless. I wrote those after the August earning.

A deal is an economic arrangement. Intel management has inside information and can make a decision that bests the interests of all stakeholders. Likewise, it is the same situation for TSMC.
 
It should be based on the asset value at a minimum, i.e., ideally the amount of time-valued money TSMC would need to construct the fabs on its own. Then, it would be jointly owned. Intel's book value is around $100 billion or maybe more. TSMC would compensate Intel at least $50 billion, which would be sufficient for Intel to pay off its debt. This would allow Intel to focus on its design business without selling off Altera and Mobileye.
It won't be at those numbers.... no one should pay that much for IFS. Look to IBM foundry sell off for a template for how this could work out. Also expect partial sell offs and maybe a third team involved . Going all in on Intel manufacturing and then figuring out it was a mistake is difficult to deal with..... and now you have Brookfield and Apollo contracts in the mix.
 
anyone on the forum who has connections to intc government affair team, could you please fwd this idea to them?

IMHO, putting just tariffs won't really help IFS nor U.S. based R&D/manufacturing, as semi cycle is too long. By creating R&D credits and prepayment credit, $$ immediatelly flows to IFS (and micron, globalfoundries, and TSMC AZ plant, for that matter).
There are plenty of credits today. And all companies know how to work the game to get "RnD" credits. The government did its part. IFS still has nightmare financial issues going forward after 4 years.
 
There are plenty of credits today. And all companies know how to work the game to get "RnD" credits. The government did its part. IFS still has nightmare financial issues going forward after 4 years.
Not really. The Biden admin is quite stingy to hand out R&D credits because there is no "income" to support generous payouts. What CHIPs act did was to support build, build and build, not advanced R&D. Now comes the era of tariffs. Basically if you do R&D and manufacturing outside of U.S., you are going to sponsor whoever does it inside U.S.

Prepayment into future US foundry manufacturing is also important, as there is not much advanced chips manufacturing capacity in the US today, need prepayment to make this incentive work.

This "tariff and credit" scheme is like stick and carrots. And it can be set expire some time in the future (e.g. 10 years). This scheme works like EV credits, which definitely helped Tesla become what it is today.
 
Not really. The Biden admin is quite stingy to hand out R&D credits because there is no "income" to support generous payouts. What CHIPs act did was to support build, build and build, not advanced R&D. Now comes the era of tariffs. Basically if you do R&D and manufacturing outside of U.S., you are going to sponsor whoever does it inside U.S.

Prepayment into future US foundry manufacturing is also important, as there is not much advanced chips manufacturing capacity in the US today, need prepayment to make this incentive work.

This "tariff and credit" scheme is like stick and carrots. And it can be set expire some time in the future (e.g. 10 years). This scheme works like EV credits, which definitely helped Tesla become what it is today.
Tariffs are always bad. it forces people to do stupid things to avoid tariffs and customers pay. (I can list items from the supply chain "bag of tricks")
Credits are good, ITCs are perfect (You pay 3B, you get 1B back). if want something, give a credit.
NEVER PREPAY GRANTS. Companies need to have skin in game first. Government Grants end up with a "where did they money go?" and a answer of "we will be looking into that".

None of this has anything to do with Intel IFS. IFS has no high confidence plan to be profitable. Intel 3 is losing money, 18A will lose money. Intel does not have meaningful foundry revenue for at least 2 years. This is an Intel problem, not a US problem.
 
Tariffs are always bad. it forces people to do stupid things to avoid tariffs and customers pay. (I can list items from the supply chain "bag of tricks")
Credits are good, ITCs are perfect (You pay 3B, you get 1B back). if want something, give a credit.
NEVER PREPAY GRANTS. Companies need to have skin in game first. Government Grants end up with a "where did they money go?" and a answer of "we will be looking into that".

None of this has anything to do with Intel IFS. IFS has no high confidence plan to be profitable. Intel 3 is losing money, 18A will lose money. Intel does not have meaningful foundry revenue for at least 2 years. This is an Intel problem, not a US problem.
Let me explain prepay credit a bit. Imagine you were NVDA, Trump admin added a tariff of $5B annually. Now you can pay the tariff directly, or you can get credit to offset this tariff. Two sources of credit: 1) if NVDA did some chip manufacturing in the U.S. for $1B, it gets $300M credit; 2) if NVDA prepay TSMC or IFS $1B for some manufacturing work in the next 3 years, it gets $300M credit, 3) IFS and Micron does advanced R&D in the U.S., they are going to have some credit which can be sold to NVDA.

Guess what NVDA would do in the above example? It would prepay fab commitment or buy R&D credits, instead of paying tariffs out of pocket.

Government does not need to use tax payer money or prepay anyone in this process. It is fab customers who prepay.
 
Let me explain prepay credit a bit. Imagine you were NVDA, Trump admin added a tariff of $5B annually. Now you can pay the tariff directly, or you can get credit to offset this tariff. Two sources of credit: 1) if NVDA did some chip manufacturing in the U.S. for $1B, it gets $300M credit; 2) if NVDA prepay TSMC or IFS $1B for some manufacturing work in the next 3 years, it gets $300M credit, 3) IFS and Micron does advanced R&D in the U.S., they are going to have some credit which can be sold to NVDA.

Guess what NVDA would do in the above example? It would prepay fab commitment or buy R&D credits, instead of paying tariffs out of pocket.

Government does not need to use tax payer money or prepay anyone in this process. It is fab customers who prepay.
Btw, this is not to benefit any particular company, though I do expect IFS and Micron would be the largest beneficiaries. TSMC can do advanced R&D in the U.S. as well, though unlikely. It can also set up new manufacturing and assembly plants to fulfill customer-prepaid committments, which is actually likely in this scheme.
 
It won't be at those numbers.... no one should pay that much for IFS. Look to IBM foundry sell off for a template for how this could work out. Also expect partial sell offs and maybe a third team involved . Going all in on Intel manufacturing and then figuring out it was a mistake is difficult to deal with..... and now you have Brookfield and Apollo contracts in the mix.
Intel is not IBM. I think the number makes sense as for TSMC to build those fabs themselves they would pay more. iPhone does not worth 1000$ but the number makes sense as it provides equivalent or higher utility for a buyer.
 
IFS fabs would likely require significant retooling to be usable for TSMC. Just as you cannot take an IFS tapeout and move it to TSMC, the way the fabs are set up are materially different. Starting from specific machine customizations before they even leave the toolmakers. On top of that, TSMC would have to retrain the Intel staff to bring them up to TSMC standards. Which is a multi-year process. In the meantime, the expensive fabs and tooling are just sitting around depreciating.
 
I am simply saying that a deal needs to satisfy the expectations of both sides simultaneously. Obviously, without tariffs, this is less likely to happen. I am sure we have all experienced situations where we felt that a service or an item was not worth its price under fair market conditions, but there was definitely a reason for setting that price.
 
I wrote about this approach on this forum after the August earnings call. At that time, I thought this approach was the cheapest for both sides. Probably no one believed it, and most viewed TSMC with a winner-takes-all perspective.

"In addition, Intel’s fab could be spun off into a new entity jointly owned by TSMC and Intel, and run by TSMC, with the new entity receiving U.S. Chip Act funding, the firm adds."

Unless the US concede to China in some key aspects such as lifting the embargo of EUV, it's unlikely for China to approve any deal between Intel and tsmc.
and I guess it's not wise for Intel or tsmc to lose China market in the short to mid term.
 
No, you are ignoring the geo-risk aspect. $50 billion is a fair value to ask, as the Arizona fabs alone cost TSMC upwards of $50 billion.

On the other hand, maybe Intel believes it is a bad deal and thinks it can maximize value further. If IFS is profitable, its valuation should be 1-2 times the book value. TSMC can manage the geo-risk themselves, but at higher costs.
TSM has said multiple times, they don't want Intel's fab. Geo risk may be important, but TSM also build and building 3 fabs in Arizona, with additional capacity in Japan. There is definitely less worrying about TSM as I believe. The last very proactive comment by Pat G was 'you see the enthuaisam for geographically resilient supply chain is fading'. Trump is definitely not in any interest as far as I can tell. So many deep pocket billionaire who have affilate with TSM and benefit from that relationship doesn't want the government to support Intel right away; otherwise, it wouldn't be steel but semiconductor to be announced first. Trump has in chief for some time, and his action in semiconductor policy is just slow, and he doesn't talk as much.
 
Unless the US concede to China in some key aspects such as lifting the embargo of EUV, it's unlikely for China to approve any deal between Intel and tsmc.
and I guess it's not wise for Intel or tsmc to lose China market in the short to mid term.
It is a joint-venture not acquisition.
 
TSM has said multiple times, they don't want Intel's fab. Geo risk may be important, but TSM also build and building 3 fabs in Arizona, with additional capacity in Japan. There is definitely less worrying about TSM as I believe. The last very proactive comment by Pat G was 'you see the enthuaisam for geographically resilient supply chain is fading'. Trump is definitely not in any interest as far as I can tell. So many deep pocket billionaire who have affilate with TSM and benefit from that relationship doesn't want the government to support Intel right away; otherwise, it wouldn't be steel but semiconductor to be announced first. Trump has in chief for some time, and his action in semiconductor policy is just slow, and he doesn't talk as much.
JD spoke quite loud about it during an international AI conference yesterday.
 
Intel is not IBM. I think the number makes sense as for TSMC to build those fabs themselves they would pay more. iPhone does not worth 1000$ but the number makes sense as it provides equivalent or higher utility for a buyer.
Lets see what the valuation is. People do not like to pay a lot for other peoples fabs as assets depreciate and buyer has to take the hit. There will be some interesting accounting in these deal options (Sell is one, JV is another). You would definitely not pay the cost to build your own fab. that has literally never happened. Micron and GF were paid to take fabs (great deals by both teams).

The book value of each fab (needed for due diligence) would be interesting.... I wonder if Intel will keep accelerating writing off the assets?

I would expect some very complex deal and management structure like with Apollo and the deal Intel did with Hynix (which still isnt done)
 
Lets see what the valuation is. People do not like to pay a lot for other peoples fabs as assets depreciate and buyer has to take the hit. There will be some interesting accounting in these deal options (Sell is one, JV is another). You would definitely not pay the cost to build your own fab. that has literally never happened. Micron and GF were paid to take fabs (great deals by both teams).

The book value of each fab (needed for due diligence) would be interesting.... I wonder if Intel will keep accelerating writing off the assets?

I would expect some very complex deal and management structure like with Apollo and the deal Intel did with Hynix (which still isnt done)
But we need to remember the key point that TSMC would not do it if there were no tariffs. When I wrote that statement, Intel likely wanted to see it happen. However, at this moment, it might be the intent of both sides. Therefore, the calculations should reflect that.
 
Fabs could be given away for free and this would be a great deal for Intel.

I have been calling for an Intel foundry split for a very long time, looks like it could finally happen.
 
Fabs could be given away for free and this would be a great deal for Intel.

I have been calling for an Intel foundry split for a very long time, looks like it could finally happen.
Doesn't make sense. You would not give your house away for free to save continuing expenses. You would sell it at the maximum price that a particular buyer can afford, i.e. at his utility value. It is more so if the buyer is actively pursuing it.
 
Back
Top