To The Moon

Summary

• Intel is managing to lose market share with superior products.
• Zen 4 versus Raptor Lake will be more difficult for Intel.
• Failures in servers and GPUs are stunning and inexplicable.

JHVEPhoto/iStock Editorial via Getty Images

This year has obviously been a difficult year for the semiconductor companies, especially recently. But, for Intel (NASDAQ:INTC), it's exposed to so many problems and failures, it makes one wonder if they are competent enough to continue to compete inthis industry, which has gotten much more competitive. This is not to say Intel has failed at everything, but it has in some ways that are just difficult to understand, and it failed spectacularly.

A few months ago, I rated Intel a strong buy, now I rate it a buy. Maybe. I don't have the answers, I'm just trying to understand the questions. So, let's go over them.

Why is Intel losing market share to AMD? Tiger Lake (Intel's 11th mobile) has always been better than Zen 3, although not dramatically so. Alder Lake is far better performing in mobile and desktop usages than Zen 3, but Intel lost market share. How did this happen? Why did this happen?

How will Intel's Raptor Lake and Meteor Lake compare to Zen 4. First, to be clear, Zen 4 is a much bigger generational leap than Raptor Lake is compared to Alder Lake. It has to be, Zen 3 is so far behind. But the kicker is, Zen 4 is built on TSMC's 5nm process, Zen 3 was on their 7nm process. You can be pretty sure, at least as time moves on, the cost per transistor is going to be less on the newer node, because it's a lot more dense. Raptor Lake is on Intel's "7" node, just like Alder Lake, but it's bigger and better. But it's not as much better as Zen 4 is compared to Zen 3. Given that Intel lost market share to AMD (for reasons I can't begin to understand, outside of incompetence) when Alder Lake competed against Zen 3, competing against a product that is closer in performance, while being likely less expensive to produce than its predecessor, when your product is more expensive to product doesn't sound that attractive a situation. How does Intel maintain market share?

Then we have Meteor Lake. This will almost certainly be out before AMD has a response to it, but how much does it matter? Intel moves to their "4" process, and will be using EUV for the first time in making the CPU. But, early reports indicate the architecture is close to Raptor Lake, so will not be a huge improvement in performance. While not certain, by any means, it also wouldn't be surprising since Intel is doing a lot of new things, like moving to a new node and using EUV as part of the process for the first time, not to mention using "tiles" for the first time. It's a lot to handle. And while the CPU will be made on Intel "4", everything else that's active will be made at TSMC, including the GPU, and IO. CPUs are not a particularly large part of the die, so this raises even more questions. The first is, how is this going to save costs, and be able to compete on price, given Intel still has to foot the costs for developing new nodes, and the incremental costs of building on those nodes should be a lot less than farming out the work. And the other part is, what are they going to be using their capacity for, when the greater part of the processor module is made by TSMC? Now, you might easily think they would use this capacity for their upcoming foundry customers, except there are none for Intel 4. Intel 3 will be of greater interest, but comes later, while Intel 4 is pure high performance, without high-density libraries, so mainly attractive for Intel's CPUs. So, while they roll out Meteor Lake and it replaces earlier generations, and given Intel can't make GPUs on I4 for the reasons given above, what will the fabs be doing? Making old products? That's not really ideal.

Of course, the biggest question is, how is Intel so incredibly incompetent when it comes to server parts, and consequently why is Sapphire Rapids constantly being delayed? Isn't this a company that knows how to make server chips? I mean, aren't they the leader in it, and have been for decades now? How did this happen not only for Ice Lake, but now for Sapphire Rapids? It's not just one delay, it's one after another, after another. We're at the point where we have to start asking ourselves, is Intel really competent enough to compete in this segment anymore? All evidence points to the contrary. Every chance they've had to dispel that, has only confirmed their incompetence.

Of course, their discrete graphics processors are another fiasco, and very difficult to explain, outside of incompetence again rearing its ugly head. But, to their defense, it's not a bizarre as their complete failure with server chips, since it's a new market, and they don't have a history of success there. And unlike the server humiliation, it looks like their GPUs are beginning to get some good news. They are still very bad products, but they've moved up from horrible. And the speed the drivers are improving is encouraging, although this generation is a lost cause in any event. NVIDIA and AMD are coming out with next generation products that obviously are going to be better than their last, which already limit Intel to the middle to lower spectrum. However, it might bode well for their next generation, which is what they are working on now.

There are some positives with the company, but their level of incompetence and failure may cast serious doubts on their leadership. You rarely find such catastrophic failures follow others, but Intel manages to do this well. Everything in this industry takes a long time, so credit or discredit sometimes hits the wrong person, but for the first time I am having doubts about Pat Gelsinger. He's at least partially responsible for Sapphire Rapids being so delayed, and also why it took so long to try a different approach with the GPU drivers. Should he have known better? Maybe not, he's new at the job, and certainly didn't have the expertise to understand the GPU situation, for example, but we'll get a better picture going forward whether he is learning quickly and reversing the inertia of decay that Intel was in before he took over.

I am still cautiously optimistic, because they have so many advantages over companies that don't have fabs, or have fabs and don't make their own parts. There are so many opportunities available for the company, but does any of it matter if they aren't effective in executing on them?

I guess it comes down to one real question. That is, is Intel competent? Obviously, they aren't. In servers and GPUs, as well as process technology, they have enjoyed failure and delay, again and again. In mobile and desktop parts, they managed to lose market share with better products. We have good promises, on this node, and this product, or that product. But with what's real, it's been one type of failure after another. Maybe the better question is, can they become competent? I don't know, we'll see. I'm optimistic by nature, and frequently disappointed, but I'm going to give Pat another six months before I pass a final verdict. If there's improvement, I'll keep the stock, if there's more bumbling and failure, and market share losses, there are much better places to put my money than this giant.

Source: Seeking Alpha

$Intel(INTC)$