Intel Investor Meeting Keynote

[hist78]

"Shareholders can sell if they don't like strategy, like I should have at $68 after Gelsinger was appointed. (What a dummy I was!)"

Intel is in fighting mode and still have a lot good strength. They have a great potential as long as they don't waste money in the process to regain their industry leadership.

If you are old enough you probably can remember Boston Chicken/Boston Market. I loved their food and business model so I bought their stocks in hoping McDonald's will acquire them. I thought they were a good fit for what McDonald's didn't have.

My logic and prediction were totally correct. Except McDonald's bought Boston Market at a bankruptcy court. I think I'm a true dummy!

 

[Portland]

"Shareholders can sell if they don't like strategy, like I should have at $68 after Gelsinger was appointed. (What a dummy I was!)"

Intel is in fighting mode and still have a lot good strength. They have a great potential as long as they don't waste money in the process to regain their industry leadership.

If you are old enough you probably can remember Boston Chicken/Boston Market. I loved their food and business model so I bought their stocks in hoping McDonald's will acquire them. I thought they were a good fit for what McDonald's didn't have.

My logic and prediction were totally correct. Except McDonald's bought Boston Market at a bankruptcy court. I think I'm a true dummy!

Boston market didn't make a graphics card for industrial machinery that will revolutionize the industry. This is the first of many.

 

[TX Fly Fisher]

The strategy is sound, but execution will tell the tale. I'm in.

 

[benb]

Nenni comment on Intel departures: Semiconductor production ramp is burnout inducing. And ramping new fabs is harder still. It's something to watch, and at Intel they have a 7 year, 4 month sabbatical leave that people tend to leave after taking. On the other hand, people are leaving other companies to go to Intel. And hiring is hard at all semiconductor companies right now.
Aside: Medicine is also in crisis from burnout: https://www.nytimes.com/2021/10/18/us/coronavirus-public-health.html
Hist question about top line (revenue): My impression is Intel is different than most other semiconductor companies. They can ship "garbage" low priced chips for Chromebooks or server chips, trade off revenue for margins. Since they are maintaining margins at 51-53, I think that's your answer--they are cutting down on the garbage, cutting down the top line, and selling higher value chips into a constrained market.

 

[hist78]

During the meeting, Intel revealed that under their Smart Capital strategy Intel will work with Brookfield Asset Management to "explore project finance options to help fund new Intel manufacturing sites and certain related renewable power opportunities."

I guess Intel is planning to use "Sell and Leaseback" strategy to improve cash and capital position. There are pros and cons on Sell and Leaseback. Hope Intel will use those cash in areas that can improve Intel's long term technology and manufacturing capabilities instead of wasting on stock buyback program and dividends.

 

[smeyer0028]

Intel (and I assume AMD) new low power CPUs are much
faster and have much longer battery life. My example is
Verilog simulation. M1s are very slow for compute intensive
applications. My example is a LG gram with an 11th generation
Intel laptop CPU plus new Linux 5 series kernels that are
much better for swapping. The 11th generation i5 does
not slow down from over heating the way my previous
8th gen ASUS did. Battery life really is 20 or more hours
if I turn down screen brightness some. Consumer prices
are down. Grams at only 1kg are around 800-900 USD.

 

[benb]

During the meeting, Intel revealed that under their Smart Capital strategy Intel will work with Brookfield Asset Management to "explore project finance options to help fund new Intel manufacturing sites and certain related renewable power opportunities."

I guess Intel is planning to use "Sell and Leaseback" strategy to improve cash and capital position. There are pros and cons on Sell and Leaseback. Hope Intel will use those cash in areas that can improve Intel's long term technology and manufacturing capabilities instead of wasting on stock buyback program and dividends.

Intel has a supply chain outsourcing agreement for fab parts with a third party. Fabs need spare parts to prevent downtime, and the parts are long lead time, specialized, sometimes intellectual property. By writing a check, Intel gets assured spare parts (that they own) while the cost is off their books. A smart move in today's world for sure.
With BAM, I think it's similar, write a check, get some assets off the books. If you read the press release, Intel has mostly committed to build fab shells, not operational fabs. So there may be empty shells in Ohio for a while. F42 was an empty shell for close to 10 years.

 

[triceratops24]

I still don't understand why Intel wants to get into foundry. They're fighting multiple wars - AI, CPU, GPU, DPU and foundry which will be a business with low margins and they are spreading their resources across all these units. Just look at Samsung. They were the technology leader in DRAM and NAND and while plowing resources into foundry. Micron has caught up to them in DRAM and Micron and SK Hynix are now ahead of them in NAND. In foundry, they're 1-2 years behind TSMC. Their resources are spread thin across multiple fronts.

 

[Daniel Nenni]

I think the financial justification is to extend the life of their fabs. Half of TSMC's business is mature nodes so Intel wants some of that. If Intel takes the process lead they may get some of the leading edge fab business. I do know that QCOM is working with Intel on the 20A process.

I still don't understand why Intel wants to get into foundry. They're fighting multiple wars - AI, CPU, GPU, DPU and foundry which will be a business with low margins and they are spreading their resources across all these units. Just look at Samsung. They were the technology leader in DRAM and NAND and while plowing resources into foundry. Micron has caught up to them in DRAM and Micron and SK Hynix are now ahead of them in NAND. In foundry, they're 1-2 years behind TSMC. Their resources are spread thin across multiple fronts.

 

[blueone]

I still don't understand why Intel wants to get into foundry. They're fighting multiple wars - AI, CPU, GPU, DPU and foundry which will be a business with low margins and they are spreading their resources across all these units. Just look at Samsung. They were the technology leader in DRAM and NAND and while plowing resources into foundry. Micron has caught up to them in DRAM and Micron and SK Hynix are now ahead of them in NAND. In foundry, they're 1-2 years behind TSMC. Their resources are spread thin across multiple fronts.

My opinion:

- Intel is one of the few companies on the planet capable of manufacturing advanced nodes and with the know-how to operate such fabs efficiently. Capitalizing on that capability is a smart move, done well. Yeah, they're behind TSMC, but there was a time when TSMC was behind Intel.

- When Intel was a mostly single product company, CPUs, they had the advantage of tuning their fab processing for the particular circuitry CPUs needed, and that alone gave them an advantage. Before memory controllers were integrated onto the CPU dies, they also had the financial advantage of a captive chipset business which used their N-1 process, which made client and server chipsets very profitable. That paradigm has been gone for over 15 years. As you pointed out, they now have a broad line of purpose-built product lines which have differing process requirements (you forgot Intel's Ethernet switches [Barefoot] and FPGAs), and lack of fabrication flexibility (and ability to efficiently accommodate more modest fabrication volume requirements) makes Intel a major customer of TSMC for its own chips. Being a competent foundry would allow Intel to bring these chips back in-house and increase the margins on their merchant products by eliminating TSMC's 50%+ margins.

- In my opinion, there's an existential threat to the business models of big one-size-fits-all merchant chip vendors (like Intel, Broadcom, and Qualcomm) that is quickly emerging, with modern design tools, IP, and even open source HW (e.g. RISC-V, and I think more coming) making it easier for chip start-ups (e.g. Ampere) and especially major cloud datacenter companies, and even automakers to produce their own chip designs, only looking for someone to fab it for them. I'm also wondering, will this trend to in-house user chip design lift foundry margins simply due to higher demand? I think it might, so in the long run Intel's foundry business may be more strategic and a bigger contributor to gross margins than is currently seen as the case.

 

[Daniel Nenni]

- In my opinion, there's an existential threat to the business models of big one-size-fits-all merchant chip vendors (like Intel, Broadcom, and Qualcomm) that is quickly emerging, with modern design tools, IP, and even open source HW (e.g. RISC-V, and I think more coming) making it easier for chip start-ups (e.g. Ampere) and especially major cloud datacenter companies, and even automakers to produce their own chip designs, only looking for someone to fab it for them. I'm also wondering, will this trend to in-house user chip design lift foundry margins simply due to higher demand? I think it might, so in the long run Intel's foundry business may be more strategic and a bigger contributor to gross margins than is currently seen as the case.

I agree. Long term domain specific chips from systems companies will dominate the market. Between domain specific chips and ARM/RISC-V based chips I think x86 based chips are in danger of extinction.

Remember, back in the 1970-80s the system's companies made their own chips. Mostly computer and instrumentation companies. Then came the IDMs (Intel and Motorola) followed by the fabless transformation.

Today the majority of revenue for EDA, IP, ASIC, and Foundries are from systems companies and that trend will continue, my opinion.

 

[count]

Intel is very tight-lipped about Intel 7 yields, but I think it would be a good guess that they're lower than the tuned 14nm processes that Intel fiddled with for years. The most information I got out of an employee for Intel 7 was that "Yields are good". Yeah, right. What is good? 85%? 90+%? At Intel's volumes every percent in yield improvement has a huge bottom line impact. I remember when the numbers were openly discussed, but that was a long time ago.

In several decades of working for big companies, I have found that when a big company really wants to succeed in a market, from the BoD on down to the engineers, and does what it takes to succeed no matter what it takes, for the most part they generally do mostly succeed. Most big companies I've been in or close to never achieve that sense of purpose. It does seem that Intel really is willing to do whatever it takes to compete with TSMC in fab process (we'll see about being a premier foundry), as measured by investments, acquisitions, and risks taken. With the possible exception of Tesla, can we think of a company more willing to bet it all on winning than Intel currently is?

The problem is Intel is betting it all on an IDM business model that doesn't make sense anymore. You can be a successful fabless company or you can be a successful foundry. IDM no longer works.

 

[hist78]

My opinion:

- Intel is one of the few companies on the planet capable of manufacturing advanced nodes and with the know-how to operate such fabs efficiently. Capitalizing on that capability is a smart move, done well. Yeah, they're behind TSMC, but there was a time when TSMC was behind Intel.

- When Intel was a mostly single product company, CPUs, they had the advantage of tuning their fab processing for the particular circuitry CPUs needed, and that alone gave them an advantage. Before memory controllers were integrated onto the CPU dies, they also had the financial advantage of a captive chipset business which used their N-1 process, which made client and server chipsets very profitable. That paradigm has been gone for over 15 years. As you pointed out, they now have a broad line of purpose-built product lines which have differing process requirements (you forgot Intel's Ethernet switches [Barefoot] and FPGAs), and lack of fabrication flexibility (and ability to efficiently accommodate more modest fabrication volume requirements) makes Intel a major customer of TSMC for its own chips. Being a competent foundry would allow Intel to bring these chips back in-house and increase the margins on their merchant products by eliminating TSMC's 50%+ margins.

- In my opinion, there's an existential threat to the business models of big one-size-fits-all merchant chip vendors (like Intel, Broadcom, and Qualcomm) that is quickly emerging, with modern design tools, IP, and even open source HW (e.g. RISC-V, and I think more coming) making it easier for chip start-ups (e.g. Ampere) and especially major cloud datacenter companies, and even automakers to produce their own chip designs, only looking for someone to fab it for them. I'm also wondering, will this trend to in-house user chip design lift foundry margins simply due to higher demand? I think it might, so in the long run Intel's foundry business may be more strategic and a bigger contributor to gross margins than is currently seen as the case.

There are many benefits for Intel to get into foundry business especially Intel's internal order volume is relatively smaller than what foundries can do. The scale of economy will be getting worst in the long run if Intel doesn't have a way to increase the volume, for both Intel and foundry clients.

But there is an issue people has not discussed too much yet. TSMC builds fabs and capacity based on customers' demand and commitment. TSMC knows clearly who will be the customers and their order commitments before starting a new fab project. On the other hand, Intel gets into the foundry business late so they are building fabs (4 of them) with limited customers. How will that work out? Is that very risky?

Fabs are very expensive to build and to operate. The spirit of "If you build it, they will come" is just too risky.

 

[kh812000]

I'm a bit surprised that ppl on this board are so skeptical of intel's strategy. Execution is another story.

 

[blueone]

But there is an issue people has not discussed too much yet. TSMC builds fabs and capacity based on customers' demand and commitment. TSMC knows clearly who will be the customers and their order commitments before starting a new fab project. On the other hand, Intel gets into the foundry business late so they are building fabs (4 of them) with limited customers. How will that work out? Is that very risky?

Fabs are very expensive to build and to operate. The spirit of "If you build it, they will come" is just too risky.

I strongly suspect Intel already has something akin to letters of intent from big chip designers that essentially say you get the fabs up and running and working well and we'll write contracts with you. I would guess a lot of companies are nervous about being so dependent on TSMC, when the Chinese have their sights on Taiwan. (Even Samsung, considering South Korea's vulnerabilities.) A second source with big volume capability would probably make some C-suite people sleep better. Intel has made some very suspect product and technology decisions over the years, but Intel's finance people always struck me as professional and methodical.

 

[Daniel Nenni]

I strongly suspect Intel already has something akin to letters of intent from big chip designers that essentially say you get the fabs up and running and working well and we'll write contracts with you. I would guess a lot of companies are nervous about being so dependent on TSMC, when the Chinese have their sights on Taiwan. (Even Samsung, considering South Korea's vulnerabilities.) A second source with big volume capability would probably make some C-suite people sleep better. Intel has made some very suspect product and technology decisions over the years, but Intel's finance people always struck me as professional and methodical.

I'm sure conversations are ongoing but QCOM is the only big name I have heard thus far and they are doing a PDK for 20A. They won't be able to keep customers a secret since they will need Intel specific IP from the TSMC ecosystem. The only IP porting that I have seen thus far is for Mil/Aero which makes complete sense.

I can assure you the chip industry wants IFS to succeed. TSMC needs competition and Samsung is not enough. We need a three horse race. I remember when GF started they had a big fan base, myself included. We were all very excited to get some of that oil money and have a US based horse in the foundry race. Intel acquiring Tower send a really big message that Intel is in it for the long haul. Exciting times for sure.

 

[hist78]

Intel has made some very suspect product and technology decisions over the years, but Intel's finance people always struck me as professional and methodical.

I must apologize first if my following question upset you. How do you tell Intel's finance management is operated with professional integrity and long term vision?

 

[blueone]

I must apologize first if my following question upset you. How do you tell Intel's finance management is operated with professional integrity and long term vision?

No upset at all. I didn't mention long-term vision, I said professional and methodical. I know because I used to work for Intel, and I interacted with many finance people on multiple projects over the years. I also respected Intel's legal group. And I've worked for other companies, so I had the opportunity to compare and contrast the experiences I had.

 

[hist78]

No upset at all. I didn't mention long-term vision, I said professional and methodical. I know because I used to work for Intel, and I interacted with many finance people on multiple projects over the years. I also respected Intel's legal group. And I've worked for other companies, so I had the opportunity to compare and contrast the experiences I had.

I stopped trusting Intel's words and numbers since Intel's "Contra Revenue" scheme around 2013 through 2015 or 2016. In 2014 alone Intel lost at least $4 billion in the Mobile Computing division and the major cause was this "Contra Revenue" practice. In certain quarters Intel Mobile division even reported negative revenue (note: not negative profit)! I don't know how many Fortune 100 companies can have guts to do it. It's not surprising that once "Contra Revenue" practice stopped, so did Intel customers' mobile chips order.

Another example is Intel Stock Buyback Program.

From: https://www.intc.com/stock-info/dividends-and-buybacks

"We have an ongoing authorization (originally approved by our Board of Directors in 2005 and subsequently amended) to repurchase shares of our common stock in open market or negotiated transactions. As of December 25, 2021, we were authorized to repurchase up to $110.0 billion, of which $7.2 billion remained available."

From 2012 through 2021, Intel spent $67.9 billion to buy stocks and reduced outstanding shares for about 1 billion shares. What has Intel achieved? Now Intel is forecasted to have negative cash flow in 2022 and needs to use Sell and Leaseback to improve liquidity.

Intel's market capitalization is at $181.93 billion today. Compare this $181.93 billion market cap to $67.9 billion (stock buyback spending since 2012) or $102.8 billion (stock buyback spending since 2005), isn't it ridiculous?

And Intel tells us that now is the time taxpayers should give them money so they can save US semiconductor industry!

Intel's Board always has several directors who have strong finance background and experience. How can they knowingly let it happen? Intel's finance department must know the situation very well. But under the guidance of Intel Chairman, CEO, and Board of Directors, they are just a group of loyal soldiers. I hope one of those finance staffs at one point did tell Intel CEO that $67.9 billion or $102.8 billion spent on the stock buyback could have built many advanced fabs.

 

[Jumper]

Wall Street doesn't like the high capex required for Intel to expand manufacturing capacity because it's a short to mid-term headwind. Long-term it could pay off, but investors who get in now will need to be patient while it plays out. I suspect those who get in at these prices could see 20%+ annualized returns in five years, but that requires Intel to execute on their strategy. If they perform to plan, they could very well outperform the current semiconductor darlings over that period. We'll see.

I think that won't happen. AMD, Nvidia and most importantly TSMC will continue growing during these years and TSMC will most likely surpass Intel in revenue in next 5-7 years.