hist78
Well-known member
Can you please explain what your "leverage"meant in this Intel-Apollo deal?
Apollo acquires 49% of Intel's Fab34
- Thread starter SemiFunctional
- Start date
Can you please explain what your "leverage"meant in this Intel-Apollo deal?
True foundry partner? How do you define that? You may argue whether the right strategy is to maximize short term or long-term profit, but these are purely business arrangements. If you really want to draw a partnership analogy, at best, these are open marriages.
I suspect you may have confused cordiality with true friendship. Right now, allocation is more important to Jensen than price. He wouldn't be where he is without being cutthroat competitive.
Intel has been a TSMC customer for ages. The only difference now is that they even fab their crown jewels there too; and they have announced their merchant foundry ambition. I maybe a fifteenth class customer at TSMC, but Intel is not third class there, intel is probably in the top tens. So, what should TSMC do, kill them before they regain strength? Qualcomm is also shopping at Samsung, choke them off too? Companies compete and collaborate, usually, if they don't like the competition part, the solution is to buy it out!
PG was definitely arrogant when he said Nvidia was just lucky. But he was also saying the truth when he said people wanted alternatives to Cuda. He just wasn't as polished and cordial as Jensen. I suspect after all the legal back and forth between the two companies over the years, Jensen must be saying (in private) that Intel finally is exactly where it deserves to be!
So, really, what should PG do to turn Intel around?
jiakun0618
New member
Hi hist78, I don't have experience in PE industry and I'm no expert on the issue. I meant to say that I thought the assumed yield of over 10% is unlikely in this deal. Since you brought up the question again, I looked at it again and found that the deal was structured in a way between equity and credit.
Here is the credit rating analysis from S&P Global after the deal:
04-Jun-2024 | 16:21 EDT
Intel Corp. ‘A-’ Ratings Affirmed On The Announcement Of Ireland Fab Joint Venture With Apollo; Outlook Remains Negative
Link: https://disclosure.spglobal.com/ratings/en/regulatory/article/-/view/type/HTML/id/3191329
As explained in the report, the nature of this JV deal was different than the Brookfield deal, which was considered a debt deal, and the expected return for Apollo is high single digit percentage. As disclosed by Intel, the yield should be between its cost of debt and cost of equity.
Regarding how to gain leverage and generate return for Apollo's investors and co-investors, I don't have that information. Reading from a FT article, I have the impression that there are co-investors in the deal.
https://www.ft.com/content/601491ea-6815-4ea6-a7a6-4ad614710439
Another FT article provided more background about Apollo. If the funding sources come from banks or insurance companies, they are using leverage by nature of their business.
https://www.ft.com/content/814b07ed-d557-4ae7-bef2-1bd2a0dc5763
hist78
Well-known member
"Intel has been a TSMC customer for ages. The only difference now is that they even fab their crown jewels there too; and they have announced their merchant foundry ambition. I maybe a fifteenth class customer at TSMC, but Intel is not third class there, intel is probably in the top tens. So, what should TSMC do, kill them before they regain strength? Qualcomm is also shopping at Samsung, choke them off too? Companies compete and collaborate, usually, if they don't like the competition part, the solution is to buy it out! "
IMO, Intel will be treated as a third class customer because Intel and Pat Gelsinger's behavior.
In my mind, there are three classes of TSMC customers:
First class: Large volume, good profit, long term and large scale collaborations with mutual respect. I call it the premier customer class.
Second class: various order volumes, reasonable profit, normal business relationship, may have a long term strategical importance even it is small currently. I call it the regular customer class.
Third class: small, medium, or large volume orders but with a problematic relationship. This kind of customers many go in and out all the time as if they are choosing a fast food restaurant. From time to time the third class customer may go out in public to accuse TSMC is not a reliable source. Even worst, from time to time when there's a full moon, the CEO of the third class customer may be waving a kitchen knife in public to show his intent to kill TSMC. I simply call them the third class customers.
Last edited:
Remember this Sun Tzu saying “keep your enemies closer!” Nothing like being the best service to your aspiring enemies division provide them with the best service ( schedule, customer attention, hit every PDK on time and exceed their expectation).
Seriously there is no better process offering than N3, hook them and deliver A16 and A14 on schedule and that enemiy may have no choice but become your big customer or slip to irrelevance.
Now TSMC delivering Compute, GPU and base die, if they continue to execut it will be very hard for the product teams to go back
hist78
Well-known member
"I thought the assumed yield of over 10% is unlikely in this deal."
Let's look into several indicators to see what's the typical of return a Private Equity (PE) is looking for.
1. "Private equity allocations by state pensions produced a 11.0% net-of-fee annualized return over the 23-year period ending June 30, 2023, exceeding by 4.8% the 6.2% annualized return that otherwise would have been earned by investing in public stocks."
Source: https://caia.org/blog/2024/04/23/long-term-private-equity-performance-2000-2023
Note: This 11.0% annualized return is the return received by those 94 state pension systems from the Private Equity. The Private Equity management must demand much higher investment return on the PE's investment targets in order to cover their own fees, profit, and loss.
2. "According to Cambridge Associates, for the 20-year period ended in June 2020, PE had average annual returns of 14.65% compared with the S&P 500, which had average annual returns of 5.91% over the same period. However, these high averages are not the case every year. The 10-year average annual return also ended June 2020 was 13.99% for the S&P 500, compared with 13.77% for private equity."
Source: https://www.titan.com/articles/private-equity-returns
Note: IMO, PE's average annual return of 14.65% is not cheap at all.
3. The ROI of Apollo Global Management (the parent company of Apollo Asset Management):
Obvious, the return Apollo demanded should be high enough to weather the loss and down cycles. Otherwise no investor will bother to put money into Private Equity that Apollo managed. It's impossible that Apollo can afford to price their investment cheap at Intel.
Source: https://www.macrotrends.net/stocks/charts/APO/apollo-global-management/roi
Last edited:
hist78
Well-known member
TSMC is probably doing exact like that except Intel is not in the TSMC's premier customer club.
We also need to understand that with the signing of Brookfield and Apollo deals, Intel product/design division lost a lot of flexibility to choose the best foundry to win. The Intel-Brookfield and Intel-Apollo deals force Intel's 3 most advanced fabs (2 in Arizona and 1 in Ireland) to meet wafer volume and profit required by the contracts. Intel product Division must use those 3 fabs first otherwise Intel breached the contracts and commitments.
hist78
Well-known member
"As disclosed by Intel, the yield should be between its cost of debt and cost of equity."
Intel did stated that about the Intel-Brookfield deal. Can you please give me some pointers that Intel said it again for the Intel-Apollo deal?
Is it possible that Apollo or Brookfield asked a return rate less than a corporate bond but greater than the dividend Intel stock paid (1.6% most recent number)? IMO, It's very unlikely.
"On January 12, 2024 Issuer Micron Technology issued international bonds (US595112CD31) with the coupon rate of 5.3% in the amount of USD 1000 mln maturing in 2031. The issues were sold at the price of 99.93% at par. The bookrunners of the placement were Morgan Stanley, Credit Agricole CIB, Bank of Nova Scotia, CIBC, HSBC, Mitsubishi UFJ Financial Group, TD Securities, Truist Bank, Wells Fargo."
Source: https://cbonds.com/news/2663615/
I don't think under so much uncertainty, competitions, and risks, a Private Equity can be happy enough to accept a return from Intel less than the Micron's 5.3% corporate bond rate. Unless a Private Equity is OK a 1% ~ 5% return just like a average grocery store does.
MKWVentures
Moderator
Intel product division does not pay actual cost. They pay market price, which is currently defined as TSMC price. So if there margin pressure will be take by Intel Foundry like it is today. This was the entire purpose of the accounting change.... to clarify that the Product group is actually pretty successful and that the manufacturing group is not currently. Perhaps Intel will use this new accounting breakout in the next few years for some purpose LOL.
This is STILL a good deal for Intel. Allows intel to build and ramp new fabs without the negative FCF.
Side note: Intel will ALWAYS preferentially load Fab 34 with Intel3 and Intel4. No other Intel 3/4 fabs exist or are planned as of now.
tooLongInEDA
Moderator
So how long do we think the Intel Foundry "market price" for Intel product divisions will remain at the "TSMC price" ? Once the option to choose between IFS or TSMC is off the table (and I note the fact that TSMC requires some pre-booking to get any capacity, so you can't easily make a late switch), what's to stop the IFS pricing rising ?
It's even arguable that Intel product groups might want to pay a lower price for using IFS in the future if it turns out that the total value/quality/service package from TSMC is superior. And if Intel has to cut pricing to drum up external IFS demand (as seems likely), would the same apply to Intel products in IFS ?
In theory the separation of product and foundry businesses removes all this opacity. In practice, with all these complicated PE deals, there may be more opacity than ever ...
The way I understand it, for IDM2.0 to be viable:
1. IF must cough up performance-wise (i.e. PPA) competitive node, and
2. IF cost must be lower than TSMC price.
If both are satisfied, going TSMC will be leaving money on the table.
hist78
Well-known member
To me, a separate Profit and Loss financial report from Intel Foundry (IF) division and demanding a TSMC comparable pricing offered by the Intel Foundry to the Intel Product division are meaningless, if it's not a financial gimmick.
There is only one stock symbol, INTC, represents one company called Intel. At the end of each month and year, only Intel's revenue and profit (neither IF nor Intel Product Division alone) matter to Intel's stock performance and the amount of dividend Intel can give. The revenue and profit/loss for Intel Foundry is a floating number that Intel can easily move back and forth between Intel Foundry and Intel Product Division. The goal is to maximize Intel's (INTC) competitiveness, revenue, and profit and eventually to help INTC stock performance. This is what the IDM model is. It's what Intel has been telling the world that Intel IDM mode is superior than competitors'.
Why on the earth Intel Foundry, who needs to shoulder so much responsibilities as a component of the IDM model, also needs to be treated as an external vendor like TSMC?
MKWVentures
Moderator
I think you are missing how all this came about. I wrote about it in a couple articles but to recap:
1) Intel had 2x the cost of TSMC since 2012. Intel Fab cost was a major issue for past 10+ years.
2) Intel 10nm was a disaster. it caused Intel to redo the roadmap every quarter and lose leadership
3) Bob Swan and the product divisions decided to outsource leading edge products. Leading processes, lower price (this is being seen now).
4) Pat came in and said to NOT outsource.... grow foundry instead. To get product groups to agree (they wanted to outsource), He said they would only be charged market price, not the much higher Intel cost. He said Intel would be competitive by 2025. this will probably be about 3 years behind schedule.
So Intel foundry needs to be subsidized or no one would choose them. That may change someday.... but it hasnt happened yet.
the future reason for why you split out the two companies is obvious. The options and trigger points are in place.
Lets see how foundry finances and those highly publicized fab ramps (52, 62, Ohio, etc) go.
jiakun0618
New member
Thanks for the additional information.
I meant to say that it's unlikely that Intel has to bear an annual payment yield of over 10% on this deal. As you posted originally, you assumed 15% yield on the debt obligation. It's highly unlikely in this particular deal.
You used private equity industry's average return and Apollo's return to assume the yield required on the deal. If this deal is an equity investment, the return may come mostly from a future exit.
jiakun0618
New member
Cost of debt is related to Intel's credit rating. But it should fall into the range of 5-6%. Cost of equity depends on the stock's volatility and other risk factors, but it should falls to anywhere between 10-12% or a range wider than that. So the yield on Brookfield deal should be between these two numbers.
You are right. I probably shouldn't have mixed the Apollo one with the Brookfield deal. That one was a debt deal, and this one is more like an equity deal. The S&P report gives a lot of details and states clearly that Apollo will receive dividend payments.
hist78
Well-known member
Intel Corporation subsides Intel manufacturing arm (Intel Foundry) since day one because its IDM business model and will keep doing so (again because it is staying with the IDM model) until one day Intel Product division and Intel Foundry split into two independent companies.
To have a separate Profit and Loss (P&L) public reporting from the Intel Foundry is totally different from to have two independent companies. If Pat Gelsinger really wants to hold Intel Foundry accountable, he can do so using internal and confidential financial data to achieve the same goal.
Intel has only one accounting book that really matters as long as it stays as an IDM. Intel Foundry's loss will be covered by its brothers and sisters withing the Intel Corporation. There is no way Intel Foundry can go down without dragging the Intel and Intel Product division with it.
Can Intel Foundry tell Intel Product division to wait for another 6 months because they would like to sign contracts to manufacture chips for Qualcomm, Nvidia, and AMD first? Because those Intel competitors' products are more promising than Intel's or those companies are willing to pay more than Intel Product division? If the answer is no, then Intel Product division should not demand TSMC like pricing from Intel Foundry.
Intel and Pat Gelsinger can't keep telling people IDM is live and well but publicly shaming Intel Foundry (an integral part of Intel IDM) for its loss. Let me repeat my earlier comments:
"Why on the earth the Intel Foundry, who needs to shoulder so much responsibilities as a component of the IDM model, also needs to be treated as an external vendor like TSMC?"
Unless Intel is actually planning the eventual breakup that will create two independent companies, one for product (fabless) and one for the Intel Foundry.
Last edited:
MKWVentures
Moderator
YupWill it be a clean breakup or will Intel turn into a holding company with Altera, Mobileye, Intel Foundry and real Intel (data center and client products)? If there is a clean breakup, would you expect that another entity would need to acquire Intel Foundry to keep it afloat (like a three way merger between GlobalFoundries, Samsung and Intel Foundry)?
I saw Zinsner cut 6 or 8 B of annual costs off Micron in three, four years.
You need to identify the costs before you can cut them. No one is shaming anyone. PG also said there were rampant mis-used of super-hot lots from Products groups because there weren't enough book-keeping/accounting consequences.
Will IF cost be as low as TSMC's, unlikely even after major cost cutting, especially given TSMC is way ahead in depreciation while Intel is just beefing up EUV capacity; but are there significant cost cutting potentials? I listened to PG and DZ and didn't feel like they were just hallucinating.
MKWVentures
Moderator
thats where it depends on how successful they are. Intel will have all options at their disposal. one choice if foundry fails, One if it succeeds. Lots of legal and corporate organization options exist