Editor’s response: Intel faces three sequential problems: get production right, start designing competitive products again, and win some customers for IFS (Intel Foundry Services). Then they have to keep it up. Not impossible, just very challenging.
We’ve been getting a lot of questions lately about the current state of Intel. Since this topic seems to fascinate much of the semiconductor world, we thought it might help to put our thoughts here.
Intel is currently facing a series of successive problems. Sequential in the sense that each problem must be resolved before the next can be effectively addressed. The company is working on all of them, but they are so dependent on each other that the company’s prospects become confused until they are all resolved.
Guest author Jonathan Goldberg is the founder of D2D Advisory, a cross-functional consulting firm. Jonathan has developed growth strategies and alliances for companies in the mobile, networking, gaming and software industries.
The first problem Intel needs to overcome is the manufacturing process. The company is structured around its IDM (Integrated Device Manufacturing) model, with internally controlled factories. Although they have outsourced some of their production to TSMC, approximately 70% of their revenue still comes from their own factories.
The company fell off the path of Moore’s law several years ago and is now racing to catch up. The ambitious goal of moving five nodes forward in four years has become the company’s official slogan. This issue is of existential importance to Intel and if left unaddressed could spell a bleak future for the company.
As far as we can tell from the outside, they are on track to achieve that goal, but it will take more than a year for these advancements to be fully implemented. It’s an incredible challenge to achieve this, but if we had to make a prediction today, we believe they could reach a level where their products can regain competitiveness, although it seems less likely to outperform TSMC in the near term.
This brings us to the second problem. Assuming Intel’s manufacturing process reaches an acceptable level, they will face competition from their products. Here they face a litany of difficulties. At the top of that list is AMD, which performs like the proverbial machine. AMD’s latest CPU portfolio looks very attractive to customers.
They have introduced significant innovations to the CPU market such as their chiplet architecture and its packaging. AMD states that their current products perform better and offer a better Total Cost of Ownership (TCO) than Intel. This gap is likely to widen as Intel continues to address its production issues, a process likely to be completed by late 2024 at the earliest.
In addition, the market is shifting. In data centers, customers are moving away from CPU-centric systems to heterogeneous computing with CPUs, GPUs, and accelerators. While Intel offers GPUs and AI accelerators, their impact on the market is modest to say the least. It seems that Intel is putting so much effort into survival that they are not updating their roadmap to adapt to this new reality. We are being generous here, because many think the situation is much worse.
Simply put, when Intel entered its current funk in the middle of the last decade, they had one CPU competitor, but today they face a dozen. A revitalized Intel can meet all these challenges. They still have a solid foothold in the PC market and long-standing relationships with the entire ecosystem, but they need products that can stand on their own.
And then we come to the third problem: they have to keep all this going. That means investing heavily in improving their production process. Even if they can get to a 5 node version in 4 years, they have to go beyond that. Moore’s law economics is punitive, only companies with huge revenues can sustain the required pace of investment.
The third challenge is the continuous commitment to improvement. This means that they invest heavily in improving their production process. Even if they can reach their goal of five nodes in four years, they need to evolve. The harsh reality of Moore’s Law is that only companies with substantial revenues can sustain the necessary pace of investment.
This problem is compounded by the fact that Intel’s factories produce so much less than TSMC. In order for Intel to keep up the R&D pace, it needs to fill its factory with more than just their own products. This is a matter of combining. Part of the reason TSMC has become the leader in semi-finished manufacturing is that they produce so much volume, which means they learn faster than everyone else, a critical part of it all.
So in order to sustain itself, Intel needs to build Intel Foundry Services (IFS) into a bona fide foundry competitor in the future. Customers will only consider IFS if they believe it can compete with TSMC. Therefore, we view Intel’s challenges as sequential. Despite ongoing rumors of fabulous companies considering IFS, we don’t foresee it becoming substantial until Intel proves its manufacturing process is both competitive and sustainable. We believe that IFS will not make a significant contribution to revenue until the end of the decade.
Essentially, Intel’s biggest challenge is cultural. The company must recognize and adapt to the changing world. Over the years, Intel has created huge blind spots by focusing too much inwardly, losing sight of its diminutive stature compared to its heyday. That’s why it was somewhat disturbing to hear Intel claim during their recent analyst update that IFS is “the second largest foundry in the world,” a calculation that included their internal products. This is akin to taking your hot older sibling to prom. Technically it’s true, but it doesn’t portray the image they think it does.
While this may seem insurmountable, there are a number of factors favoring the company. They have tremendous internal talent, which seems both dedicated and amplified. Plus, they don’t have to excel in every aspect. For production, they don’t need to surpass TSMC, they just need to get close enough that the process no longer undercuts their products. They don’t have to be the second largest foundry in the world (excluding their own product), they just need to bring in a few fables to outside customers.
Difficult, but not impossible.