Intel Warns That Its Foundry Strategy Is "Highly Risky" & Its Success Is "Highly Uncertain"
they also warn that they have limited experience in the third party foundry business, just FYI
Yes, those are the rather stark warnings about its foundry plans included in the “Risks” section of Intel’s 2023 Annual Report. Of course, companies put all kinds of shocking stuff in their 10K Risks sections, from wars to natural disasters and everything in between. It’s kind of what we expect and we tend to ignore what it says, just like the 40 pages of Terms & Conditions you need to accept after downloading a new App for example.
So, why pay any attention then to what Intel is saying about its foundry business? The reason is simple. It’s not so much about what the company says, it’s as much about when it says it. You see, there’s no mention of anything about risks associated with Intel’s foundry business in the 2022 Annual Report, even though their foundry plan had been flagged way back in March of 2021, details here. So why is it showing up now?
There’s another new risk item detailed in the 2023 Annual Report under the following heading:
“Our Smart Capital approach to capital spending, alternative financing arrangements and pursuit of government grants involves risks and may not be successful.”
Yet, the Smart Capital program was launched in August 2022, details here:
Again, why is this just now showing up for the first time in the 2023 Annual Report?
Before we look further into these two new Risk categories outlined by Intel, I want to share a real example from Intel’s past annual reports, around the time the company started running into problems with their 10nm process roadmap.
This was what Intel noted in the 2019 Annual Report in response to the 10nm roadmap delay. For the record, this is the first time Intel mentioned larger die size in an Annual Report. Note also how they simply refer to “increased product costs” without explaining why:
2019 Annual Report
Risks inherent in the development of next-generation process technologies include production timing delays, lower-than-anticipated manufacturing yields, and product defects and errata. For example, we announced earlier in 2018 that volume production on our 10nm process technology was being delayed from the second half of 2018 into 2019. We have made progress on improving 10nm yields during 2018, and we expect 10nm-based volume client systems on retail shelves for the 2019 holiday season. However, the delays in our transition to next-generation process technology may allow competitors to take advantage of potential improvements in performance, energy efficiency, cost and/or other features that may be offered by new process technologies developed by third-party foundries. Furthermore, without the benefit of next-generation process nodes, including additional competitive features on our products may result in larger die size products, manufacturing supply constraints and increased product costs.
The following year, die size is mentioned again, this time in two different contexts. You see, Intel had made a big deal about investing heavily in new manufacturing capacity- back then, they simply could not keep up with demand, believe it or not!
2020 Annual Report
However, despite increasing 14nm wafer capacity, we did not see a commensurate increase in client CPU unit volume as wafer capacity was largely consumed by increases in modem and chipset volumes, and unit die sizes. Our focus on capacity expansion and meeting customer expectations is critical as we move into 2020.
But Intel’s client PC processor die continued to increase in size in order to add more cores to keep with with the competition from AMD. Those larger dies consumed more wafer area, and therefore they failed to increase their chip production capacity.
There was a further problem. Not only were those larger dies consuming more wafer area, they were subject to greater numbers of defects, and for the first time, Intel connects the dots for us and admits that they are having lower yields as a result:
Furthermore, when the introduction of next-generation process nodes is delayed, including additional competitive features in our products can result in larger die size products, manufacturing supply constraints, and increased product costs. Lower manufacturing yields and longer manufacturing throughput times, compared to previous process nodes, can increase our product costs and adversely affect our gross margins, and can contribute to manufacturing supply constraints. In addition, as the die size of our products has increased and our manufacturing process nodes have shrunk, our products and manufacturing processes have grown increasingly complex and more susceptible to product defects and errata, which can also contribute to production timing delays and lower yields.
As a reminder, this information was in the 2020 Annual report, published over four years ago. If you know anything about Intel’s woes over the past five years or so, you may well recognise that the explanation in the above paragraph describes exactly what Intel’s problem was back then. However, this level of detail pretty much never surfaced on an earnings call or during a Q&A session.
I use this example simply to illustrate that there can be extremely useful information buried in the details of a company’s 10K Risk section. You just need to search for it and recognise it for what it is when you see it. Now, let’s dig into the two newest Risk section additions outlined above….
Keep reading with a 7-day free trial
Subscribe to Semicon Alpha to keep reading this post and get 7 days of free access to the full post archives.