Breakfast Bytes Blogs

Yesterday, in my post Intel IDM 2.0, I gave a summary of Pat Gelsinger's presentation last week on... well, everything he is doing as CEO of Intel after not much more than a month in the job—somehow I think that he's been thinking about what he would do if he was CEO of Intel for some time. Some of what he presented was the state of Intel's process roadmap, their processor roadmap for the next couple of years, and some financial outlook. But far and away the most interesting aspect of his presentation was that Intel was going to seriously get back into the foundry business. You see can my post from yesterday for a bit more detail, or better still watch the video (which is embedded in yesterday's post or you can watch it here). The most significant points, I think, are:

• A separate business unit reporting directly to the CEO, led by Randhir Thakur, who had been heading up Intel’s supply-chain operations
• An investment of $20B to build two new fabs in Arizona, some of the capacity to be dedicated to foundry
• Further announcements of new capacity in the US, Europe, and "other global locations" later this year
• Availability of the most advanced Intel processes (so 10nm now and 7nm in 2023 presumably)
• Availability of a portfolio of IP including Arm and RISC-V....but most significantly x86 cores and other previously proprietary IP

Just a bit of background. Intel has attempted to build chips for other companies before. They started up an ASIC business in the 1980s. I was at VLSI Technology competing with them, but the received wisdom at VLSI was that Intel would not last long once it discovered that it could not make the same revenue per ASIC wafer as it could with microprocessors. And indeed Intel shut it down fairly quickly. Then, a few years ago, Intel started up a foundry business called Intel Custom Foundry or ICF. I don't recall them ever announcing any marquee customers apart from Altera (who it acquired). I don't think ICF was ever officially discontinued, it just faded away.

In Pat's presentation, he said that there was foundry interest from Amazon, Cisco, Qualcomm, Google, Ericsson, IBM, imec, and Microsoft. Since all those companies will have had to approve their name being used, I have to assume the interest is genuine. Of course, it's a big step from expressing interest in a keynote to placing production orders for thousands of wafers per week. One trend has been for system companies, especially hyperscale companies with huge data centers, to build their own silicon. I think part of the motivation for creating an Intel foundry business and making the x86 IP available to it, is just this. If Intel customers can't build their own silicon around x86 processors then they will just use other processors. Building their own silicon is more important than the instruction set.

After his presentation, which was presumably pre-recorded, Pat took live questions. The Q&A starts exactly at the 30-minute mark in the video if you want to skip to it. Pat was joined by Anne Kelleher who heads up Technology Development (meaning semiconductor process development), and CFO George Davis. Not joining him, but namechecked anyway, were Keyvan Esfarjani, who heads up Global Manufacturing and Operations, and Randhir Thakur the newly appointed head of Intel Foundry Services. The reason that they could not be on the call is that they were in Arizona for the ribbon cutting for the two new fabs to be built there. You can see that ceremony, which also involved Arizona Governor Doug Ducey and US Secretary of Commerce Gina Raimondo:

If I'd been invited to ask the first question, I'd have asked something similar to Mike Rogoway of The Oregonian newspaper, who actually asked the first question:

This isn’t Intel’s first shot at foundry. There was a big push in that direction about seven years ago. What didn’t work back then and why is it going to work this time?

One thing that you get as a new CEO is that you were not responsible for what came before, so you can be less defensive about it than when it happened on your watch. But it is refreshing anyway to hear Pat admit openly to problems with Intel's process roadmap, or in this case, its previous attempt at entering the foundry business. Pat said:

Thanks very much, Mike. First I’d say the market is really different today. As you’ve seen there is extraordinary demand for semiconductors, strong interest on the part of U.S. and European institutions and governments. I’d also say that our first efforts were somewhat weak. We learned a lot through them but we didn’t really throw ourselves behind them. As you heard today with our major capacity investment, a separate organization, a separate P&L, reporting directly to me, we’re going after this much more aggressively.

Also, we’re bringing together the leading process and packaging technology, the best that we have to offer is going to be made available to our foundry customers, and a world-class IP portfolio of industry IP, but bringing together the Intel IP: graphics, AI, interconnect, and x86 cores. We’re putting it all on the table. Imagine if you were a major cloud service provider saying, “Boy, I have tens of millions of cores that are running, and now I can optimize them for my business, and add some of my stuff and maybe take out some things that I don’t utilize?” This is a powerful strategy and we believe that now is the time for Intel Foundry Services and I’m commited to make this a huge success for Intel. It’s a key piece of our IDM 2.0 strategy.

Another aspect that is different today is the perspective of the US government. Until 12/14nm, there was US-based manufacturing at GLOBALFOUNDRIES Fab 8 in Malta, NY with a roadmap to 7nm and beyond. But GF pivoted and dropped out of that race, so all the advanced-node foundry capacity is in Asia. This is something that Pat discussed explicitly in his presentation.

The US government is acutely aware of the lack of advanced-node capacity in the US, since it (mostly) only buys from "trusted" foundries. The word "trusted" has a precise meaning in this case, meaning on US soil and staffed with US citizens.

But it is not just the government. The pandemic has disrupted supply chains in a couple of ways. First, because of limitations on normal work. Secondly, somewhat ironically, because work-from-home has increased demands for technology in general and semiconductors in particular. As a result, fabs are full. You've probably heard of shortages and line-down situations in the automotive industry. These have been made worse by the need to temporarily shut down all the fabs in Texas due to the lack of electricity during the unprecedented cold weather in February, and then a fire at one of Renasas' fabs in Japan that makes automotive microcontrollers among its products. On top of that, there is a shortage of shipping containers, less room in the cargo holds of passenger flights, a backup of ships outside Long Beach...and now the Suez Canal is blocked.

All of this has led companies to want to have more flexibility in their supply chains. For some, it may be important to have the capability for advanced nodes on US soil. For others, it is just the need for more overall capacity worldwide, and preferably from a different supplier in a different location so they are not going to be disrupted by earthquakes, hurricanes/typhoons, geopolitics, or even very cold weather. I bet in the scenario planning at Volkswagen in Germany, a freak winter storm in Texas was not a bullet point on their Powerpoint.

Dean Takahashi of VentureBeat asked about the two fabs in Arizona, what nodes, and so on. Pat ducked that question a bit:

We're not being explicit on the nodes but we are saying they are EUV-capabable, so 7nm and beyond capable. We'll get  more specific as we go forward in time. They will be for committed capacity for our foundry customers so they can show up and say "I want 5K wafer per week" and we'll be able to say "Yes, we're making that commitment to you".

Stephen Nellis of Reuters asked the other obvious question about foundry:

How are you going to balance external and internal demand? And how are you going to manage that when tough decisions come down? We've got a limited supply chain right now. If you can't, say, get substrate to meet your own demand, how are you going to balance your commitment to foundry customers?

Pat started out with the fact that he is going to run this as a completely separate business, and then channeled his mentor Andy Grove from when he was much younger, and that he'd talked about in his prepared presentation:

Part of it is setting this up as a separate business unit. Separate P&L, separate accountability for it. And we're going to make commtiments to our customers in those areas, and we're going to meet our commitments.
...
Part of the announcement today of the two new fabs is to create more capacity. This is exactly what we're leaning in to. And we expect that we'll announce our next factory locations in the US, as well as Europe, within the next year, because we see this as a strong market demand overall.
...
We're going to work hard to always be on the front foot of the technology, of the capacity, and overachieving on every aspect of our business as we look forward. I’ll say one of the themes to my young CEO tenure: execute, execute, execute. We’re bringing back the execution discipline of Intel. I call it the Grovian culture that we do what we say we will do. That we have that confidence in our execution. That our teams are fired up. That we said we’re going to do ‘x’, we’re going to 1.1x, every time that we make a commitment. That’s the Intel culture that we are bringing back. The enthusiasm we’re seeing inside of the company, the passion to execute. The commitment to leadership. That competitive zeal. And to collectively, as that emerges, we have confidence that we’re going to meet not just the existing customers, we’re going to be leaders in the market, and we’re going to satisfy the new foundry customers, because the world needs more semiconductors, and we’re going to step into that gap in a powerful and meaningful way.

When Chinese premier Zhou Enlai was asked whether he thought the French Revolution was a success, he said: "Too soon to tell". According to Henry Kissinger, who was there, he actually thought the question was about the 1968 student uprisings in Paris a few years earlier, but as Kissinger said, it was a mistake "too delicious to correct".

Will Intel Foundry Services be successful? Too soon to tell.

Sign up for Sunday Brunch, the weekly Breakfast Bytes email.