Intel Delays its Ohio fabs

[blueone]

Seems like an easy job for AI, especially for modern designs based on cell libraries. Once you factor out the low level clutter and the buffers the circuit net will be more obvious. It may be tough to reconstitute the high level logic that humans understand since early steps in EDA may convert clusters of logic to primitive gates like NAND which function correctly but are, in quantity, gibberish to humans. But there might be tools around to help, maybe not on the open market but developed by folks interested in sucking in outside IP.

Hmmm... thought provoking. I'm wondering what the training will look like.

 

[hist78]

When you are Intel, who have build multi-Bn projects in multiple companies, there is probably depth of managerial understanding of how to do greenfield gigafab in a new country.

When you have done it only in Taiwan, there are probably multiple, even overwhelming, surprises in Arizona. Compare with Samsung, who are not on their first rodeo, and seem to be making good progress.

It makes sense that TSMC is much happier in Japan with a bunch of partners in a consortium who know how to navigate the politics and practicalities.

There are two major differences between Taiwan/Japan and US.

1. In the US, the endless, meaningless, and sometimes ridiculous political infightings have dragged too long, especially among those people from the same party. For example, several Republican congressmen from Ohio actually voted against the Chips Act, even though the Act got overwhelming bipartisan support in the Senate. Additionally, the legislation to streamline the Federal environmental review process (can take 3 to 5 or more years to complete if no waiver granted) is going nowhere in the US House while Senate passed it unnimously. Don't expect TSMC or any companies to help politicians to sort out their own differences and grievance.

2. The labor unions (such as piping, HVAC, and sheet metal) in Arizona think they know everything about building an advanced fab. TSMC's knowledge and its track records to complete one to two fabs every year mean nothing to them. The unions believes they "deserved" and "entitled" this or that before they can even prove themselves.

When the unionized workers with the supposedly superior "union quality" reputation failed to install bolts properly or simply forgot to install enough bolts on the Boeing 737 Max 9, what will those international companies think? Expand their manufacturing in the US or avoid coming to US unless they have to? If they have choice, such as TSMC, do they put US at the #1 position in the list to build a new factory or just go big in other countries instead?

 

[hist78]

US companies do share buybacks to:

1. Reduce shareholder dilution from employee stock options and RSUs.
2. Bypass the double taxation in the US tax code of dividends. (Companies pay corporate income tax on their net income, and then stockholders pay income tax on the dividends.) Since buybacks reduce the number of shares outstanding they improve the Price/Earnings ratio of the company's stock because net income is divided by a smaller number of shares, which enriches shareholders and defers taxation until the stock is sold.
3. By buying hundreds of thousands or millions of their own shares, usually during price dips, they effectively raise the market price of their shares, making stockholders happier.

The local governments get no benefit. Progressive Democrats in Congress generally hate stock buybacks because they think the strategy only benefits corporate executives and the very rich. They are grossly mistaken. Warren Buffett said this about them: “When you are told that all repurchases are harmful to shareholders or to the country, or particularly beneficial to CEOs, you are listening to either an economic illiterate or a silver-tongued demagogue (characters that are not mutually exclusive)”

But the progressives in the current administration are against buybacks, so the Commerce Department's implementation of CHIPS Act places restrictions on them for companies taking the subsidies.

In Intel's case, unfortunately there is a bad effect (or intention) to artificially push the stock price higher in order to hide the problems Intel was and is facing.

When Intel finally ran out money to do the share buy back, all those problems hiding there for years became so huge and so difficult to solve.

 

[hist78]

Yes.

I get the political consequence part, and somewhat for cost (like water and power), but for Intel's Copy Exactly strategy, how are these other factors affected by geography?

Location will affect cost and efficiency. Intel's new fabs are all in high cost area while Pat Gelsinger keeps pointing out Intel's cost is too high to be competitive.

How can Intel improve its cost structure when it keeps doing the opposite ways?

Intel looks at all the costs and incentives and strings and makes the correct decision. I am pretty sure Intel knows the details of what the financials are better than the people blogging online. I have seen the contracts and I can state that they are more detailed than people on here understand..... but Intel actually knows all the details and the financial numbers. Where I think Intel is incorrect is in their ability to be successful in foundry and ramp .... although it is becoming clear that their strategies are more complex and nuanced than most people thought.

Although it's not limited to Intel, but Intel did repeatedly make wrong decisions based on the same information and number other companies saw.

The critical part is how Intel interpreted those data and how Intel's leadership, such as Intel's Chairman or CEO, to use those analysis to make the right decisions. Or, in the worst situation, it's about how to make wrong decisions less frequently.

A CFO or Chief Marketing Officer must yield to CEO's and Chairman's wishes even the CEO or Chairman might be wrong.

 

[blueone]

There are two major differences between Taiwan/Japan and US.

1. In the US, the endless, meaningless, and sometimes ridiculous political infightings have dragged too long, especially among those people from the same party. For example, several Republican congressmen from Ohio actually voted against the Chips Act, even though the Act got overwhelming bipartisan support in the Senate. Additionally, the legislation to streamline the Federal environmental review process (can take 3 to 5 or more years to complete if no waiver granted) is going nowhere in the US House while Senate passed it unnimously. Don't expect TSMC or any companies to help politicians to sort out their own differences and grievance.

Agreed, though I think the CHIPS Act is so poorly conceived I wish it had failed. TSMC, Samsung, and Intel (among others) were building before the Act was even finished, so I'm not sure what it's accomplishing except for creating more useless bureaucracy and government spending. Agreed also about permitting, though as I've pointed out before, the underlying problem is really the NEPA legislation.

2. The labor unions (such as piping, HVAC, and sheet metal) in Arizona think they know everything about building an advanced fab. TSMC's knowledge and its track records to complete one to two fabs every year mean nothing to them. The unions believes they "deserved" and "entitled" this or that before they can even prove themselves.

When the unionized workers with the supposedly superior "union quality" reputation failed to install bolts properly or simply forgot to install enough bolts on the Boeing 737 Max 9, what will those international companies think? Expand their manufacturing in the US or avoid coming to US unless they have to? If they have choice, such as TSMC, do they put US at the #1 position in the list to build a new factory or just go big in other countries instead?

While I despise labor unions and I think they are destructive (I have been a union member, so I know first hand how bad they truly are), you seem to have no idea how US unions work. The unions have nothing to do with quality standards during construction or manufacturing. Quality processes and their verification are defined by the contractors and manufacturers, not by the unions. Boeing's problems are irrelevant in a discussion about TSMC, and while I have lots of (negative) opinions about Boeing it is not productive to bring up their issues (and Spirit Aerosystems' issues) in a semiconductor forum.

 

[blueone]

Location will affect cost and efficiency. Intel's new fabs are all in high cost area while Pat Gelsinger keeps pointing out Intel's cost is too high to be competitive.

How can Intel improve its cost structure when it keeps doing the opposite ways?

Where has Gelsinger said Intel's manufacturing costs are too high?

 

[blueone]

In Intel's case, unfortunately there is a bad effect (or intention) to artificially push the stock price higher in order to hide the problems Intel was and is facing.

When Intel finally ran out money to do the share buy back, all those problems hiding there for years became so huge and so difficult to solve.

This is your - in my opinion invalid - interpretation. Intel stock was underperforming for several years, because Intel CPU designs fell behind AMD, and because of Intel's infamous missteps in fab process development. All of Intel's problems are due to poor management, and whatever stock buybacks they did were grossly insufficient to cover up the problems.

One other advantage I forgot to list of stock buybacks... buyback levels can be altered almost transparently year to year. The alternative, using dividends to return excess cash to shareholders, is not transparent at all. Reducing the dividend is considered a very negative action, while reducing the buyback level is just a line sparsely covered in the press in a quarterly report.

 

[hist78]

This is your - in my opinion invalid - interpretation. Intel stock was underperforming for several years, because Intel CPU designs fell behind AMD, and because of Intel's infamous missteps in fab process development. All of Intel's problems are due to poor management, and whatever stock buybacks they did were grossly insufficient to cover up the problems.

One other advantage I forgot to list of stock buybacks... buyback levels can be altered almost transparently year to year. The alternative, using dividends to return excess cash to shareholders, is not transparent at all. Reducing the dividend is considered a very negative action, while reducing the buyback level is just a line sparsely covered in the press in a quarterly report.

We can list all the benefits of the stock buyback program that hundreds of companies have done before but are those benefits really achieved in the Intel's case? Are there any negative impacts?

Intel spent US$133.86 billion since 2000 or US$83.54 billion since 2010 to purchase Intel's shares. But Intel's nominal 2023 revenue has shrunk back to 2011 level. If we consider the inflation factor, Intel's $54.23 billion 2023 revenue is less than the its $59.46 (CPI adjusted) 2000 revenue.

In the mean time, Intel is struggling to find money to invest and to compete.

If Intel didn't spend that amount of money mentioned above in stock buybacks, they can allocate $20 or $30 billion in R&D and in building new fabs, easily!

Many people pointed out that Intel often canceled projects before it can achieved the results or was late/lacking in investing in new critical products and technologies. I think part of reasons is that Intel share repurchase program siphoned out too much cash from Intel.

As long as Intel's stock price looks good, who cares?

Until they can't.

 

[blueone]

We can list all the benefits of the stock buyback program that hundreds of companies have done before but are those benefits really achieved in the Intel's case? Are there any negative impacts?

Yes, Intel received benefits from the buybacks. The biggest benefit in the long term is lack of dilution from employee compensation in RSUs and stock options. Are you aware that all Intel employees at all levels of the company (including, for example, administrative assistants) are eligible to receive stock benefits? Additionally, Intel has an employee stock purchase program, in which Intel employees can buy stock at a 15% discount. Since Intel has about 130,000 employees, they need significant buybacks simply to reduce dilution.

Intel spent US$133.86 billion since 2000 or US$83.54 billion since 2010 to purchase Intel's shares. But Intel's nominal 2023 revenue has shrunk back to 2011 level. If we consider the inflation factor, Intel's $54.23 billion 2023 revenue is less than the its $59.46 (CPI adjusted) 2000 revenue.

No question, Intel has floundered. In 2000 (I was working there at the time), Intel dominated the CPU market. The loss of leadership is tragic, but as I mentioned in a previous post, Intel's problems are due to mismanagement, not underspending on R&D.

In the mean time, Intel is struggling to find money to invest and to compete.

If Intel didn't spend that amount of money mentioned above in stock buybacks, they can allocate $20 or $30 billion in R&D and in building new fabs, easily!

Intel has been spending a lot on fab construction and upgrades regardless of the past buyback spending.

Many people pointed out that Intel often canceled projects before it can achieved the results or was late/lacking in investing in new critical products and technologies. I think part of reasons is that Intel share repurchase program siphoned out too much cash from Intel.

Not correct in my direct experience. Intel often has poor-conceived strategies which do not succeed, and it's not for lack of spending. Or, a new CEO is appointed and pushes a different set of strategies than his predecessor(s), and cancels projects before they come to fruition. This was especially true in 2000-2010.

 

[tooLongInEDA]

Where has Gelsinger said Intel's manufacturing costs are too high?

Can't track it down right now, but I recall him explicitly saying this in an interview within the last few months.

 

[count]

Can't track it down right now, but I recall him explicitly saying this in an interview within the last few months.

Doesn't sound like something Pat would say, at least not in those words. If he was making comments on costs, he'd probably do it in a roundabout way, saying something to the effect of "Margins will be great, but not has high a TSMC"

 

[blueone]

Can't track it down right now, but I recall him explicitly saying this in an interview within the last few months.

There have been articles about Intel's foundry construction costs in The Register, if that's what you're talking about. For example:

Intel turns to private equity to help pay for new factories

That $52b CHIPS Act ain't enough – we need another $15b

www.theregister.com

and

Inflation makes the cost of Samsung's Texas fab site ruinous

South Korean titan budgeted $17 billion. Could now need over $25 billion to bring the facility online

www.theregister.com

Even Gelsinger wouldn't comment about production costs, IMO. That information is very confidential.

 

[MKWVentures]

There have been articles about Intel's foundry construction costs in The Register, if that's what you're talking about. For example:

Intel turns to private equity to help pay for new factories

That $52b CHIPS Act ain't enough – we need another $15b

www.theregister.com

and

Inflation makes the cost of Samsung's Texas fab site ruinous

South Korean titan budgeted $17 billion. Could now need over $25 billion to bring the facility online

www.theregister.com

Even Gelsinger wouldn't comment about production costs, IMO. That information is very confidential.

Even without my comments (based on a model of Intel and TSMC costs). The CFO has stated that Intel needs to improve costs to match competitors (this was in the whole integrated margin presentation). Intel has also stated that when they break out manufacturing/IFS, the manufacturing margins will be negative. Intel has made it very public that costs need to improve in order to compete.... hence the breakout. @blueone @tooLongInEDA

 

[blueone]

Even without my comments (based on a model of Intel and TSMC costs). The CFO has stated that Intel needs to improve costs to match competitors (this was in the whole integrated margin presentation). Intel has also stated that when they break out manufacturing/IFS, the manufacturing margins will be negative. Intel has made it very public that costs need to improve in order to compete.... hence the breakout. @blueone @tooLongInEDA

Good point. Margins must be low, since Intel is outsourcing some of their highest value fab work to TSMC. I've seen the comments about manufacturing margins expected to be negative, which was/is clearly an underutilization issue. As for IFS, I'd make a small bet that their IFS revenue does not even cover their current R&D investments for IFS staff, PDK development, and funding for ports of IFS-critical IP. (And considering all of this, I really don't understand the recent run-up in INTC share price.)

 

[nghanayem]

Good point. Margins must be low, since Intel is outsourcing some of their highest value fab work to TSMC. I've seen the comments about manufacturing margins expected to be negative, which was/is clearly an underutilization issue. As for IFS, I'd make a small bet that their IFS revenue does not even cover their current R&D investments for IFS staff, PDK development, and funding for ports of IFS-critical IP.

Not much of a bet when IFS has a negative op income even with the IMS business.

(And considering all of this, I really don't understand the recent run-up in INTC share price.)

Agreed. My only guess is that the street didn't expect the bottom to be Q1'23 and that folks were expecting an imminent MSS swap with AMD. Combined with investors being shocked at not seeing products or process getting delayed. Now if that is more so a reflection of just how bad things were, or unreasonably low investor expectations I couldn't say for sure. My guess is a little of column A and a little of column B.

 

[Barnsley]

There are two major differences between Taiwan/Japan and US.

1. In the US, the endless, meaningless, and sometimes ridiculous political infightings have dragged too long, especially among those people from the same party. For example, several Republican congressmen from Ohio actually voted against the Chips Act, even though the Act got overwhelming bipartisan support in the Senate. Additionally, the legislation to streamline the Federal environmental review process (can take 3 to 5 or more years to complete if no waiver granted) is going nowhere in the US House while Senate passed it unnimously. Don't expect TSMC or any companies to help politicians to sort out their own differences and grievance.

2. The labor unions (such as piping, HVAC, and sheet metal) in Arizona think they know everything about building an advanced fab. TSMC's knowledge and its track records to complete one to two fabs every year mean nothing to them. The unions believes they "deserved" and "entitled" this or that before they can even prove themselves.

When the unionized workers with the supposedly superior "union quality" reputation failed to install bolts properly or simply forgot to install enough bolts on the Boeing 737 Max 9, what will those international companies think? Expand their manufacturing in the US or avoid coming to US unless they have to? If they have choice, such as TSMC, do they put US at the #1 position in the list to build a new factory or just go big in other countries instead?

Didnt Boeings current string of issues start when they decided to cut costs and outsource software development to folk being paid $5/hr ....

The current issue is just a continuation of very poor upper management decisions which seemed based solely on financials amd zero to with the planes , which one would think be the most important aspect.

Am sure it helped the costs and ultimately the bottom line , which what seems the most important thing on here.

 

[count]

Run up in price is across entire semi industry and AI hype, where any company that can make a claim they could capture any kind of AI revenue is getting bought.

 

[blueone]

Didn't Boeing's current string of issues start when they decided to cut costs and outsource software development to folk being paid $5/hr ....

Boeing denies that they outsourced the MCAS software to companies in India. Boeing's MCAS subcontractor was Collins Aerospace.

https://www.nytimes.com/2019/11/07/business/boeing-737-max-collins.html

The current issue is just a continuation of very poor upper management decisions which seemed based solely on financials and zero to with the planes , which one would think be the most important aspect.

Am sure it helped the costs and ultimately the bottom line , which what seems the most important thing on here.

IMO, you're exaggerating. ("zero to the planes") However, I think there's ample evidence that Boeing has a history for the past several years of lack of appropriate production oversight and quality assurance procedures.

I also don't think Boeing and Intel are at all comparable.

 

[Barnsley]

Boeing denies that they outsourced the MCAS software to companies in India. Boeing's MCAS subcontractor was Collins Aerospace.

https://www.nytimes.com/2019/11/07/business/boeing-737-max-collins.html

IMO, you're exaggerating. ("zero to the planes") However, I think there's ample evidence that Boeing has a history for the past several years of lack of appropriate production oversight and quality assurance procedures.

I also don't think Boeing and Intel are at all comparable.

To be honest me neither.

Its just that someone in the thread brought them up as a bastion of virtue at management level and it was the Unions that were bringing them down.

 

[blueone]

To be honest me neither.

Its just that someone in the thread brought them up as a bastion of virtue at management level and it was the Unions that were bringing them down.

Yeah, I remember, and it's incorrect. The one thing that both Intel and Boeing do have in common is that senior management incompetence brought them down.