Last week Intel held their Investor Day. The presentation was full of information about a lot of topics we have been discussing lately. And then when we were listening to this episode of the Transistor Radio podcast, we got a great run down of all things that Intel has to offer. (As a side Note, you should subscribe to this podcast. It is worth the time for Doug and Dylan’s insights. Also, subscribe to their respective newsletters for even more semis goodness.)
One thing that really stood out for us through all that content was the breadth of Intel’s offerings. CPUs, GPUs, FPGAs, Optical Interconnect and much more. And while all of that impressive, it begs the question what does Intel really need to achieve strategically with Intel 2.0 and its ongoing restructuring?
Let us put that in context. When Satya Nadella took over the reins at Microsoft, he pushed the company towards a cloud strategy. He realized that for years Microsoft had been protecting Windows, that no longer made sense as a strategic objective. If we had to sum up Microsoft’s ultimate strategic objective today we would frame it as maximizing Microsoft’s share of enterprise IT wallet, whether that is in the cloud or on-premise. Microsoft has become adroit at leveraging its customer base’s built-in relationships with Microsoft to maintain that position. Other large, older tech companies have made similar shifts in strategy, or at least in focus – Cisco withstood the move to commodity switching with a shift to software licensing, and Oracle has done something similar albeit less easy to describe. (Yes, we are oversimplifying, but the point remains.)
Seen in that light – what is Intel’s ultimate objective? And the reason we ask is because the obvious answer does not seem achievable. From a financial perspective, Intel’s long-term goal has always been to keep their fabs fully (or at least optimally) utilized. They fell off that track several years ago, and with the yawning gap in their manufacturing capabilities, there is no way they can fill their fabs any time soon, especially with major new installations shaping up in Ohio and Arizona now, with other locations likely on the horizon. Yes, they are supply constrained on legacy processes today but that is driven as much by the (temporary) global supply chain crunch as it is by delays in their manufacturing process (if they cannot build to leading edge, they are partially filling the gap with older processes). Either way, their current utilization levels are artificially high and due for a reset in the near future. Everyone believes that Intel Foundry Services (IFS) is a play to tap third party demand to help them keep their fabs full. But IFS is years away from delivering meaningful demand. Few customers will use IFS until Intel can close the manufacturing process gap, and that is not scheduled until at least 2024. Factor in the inevitable growing pains, and just basic sales methodologies, and there is no way IFS will move Intel’s fab utilization meaningfully until towards the end of the decade.
At the same time, Intel is moving meaningful capacity to TSMC. This is the only way the design side of Intel can survive – by getting someone else to manufacture competitive chips. Almost any way you do the math, it should be clear that Intel cannot fill its fabs for a very long time.
Intel is full of smart people, not least their new CEO. So they must be aware of this challenge. So rather than assume they are chasing a false goal (not impossible, but unlikely, we hope), what else can they really hope to achieve?
And here is the problem – it is hard for us to come up with a clear answer. Maximize profits? Yes, but with what strategic leverage. Tread water until manufacturing catches up? That is the strategy that delivered them here, and will still leave them with underutilized fabs. Capture maximum share of customer wallet? That goal is on a timeline beyond everything else above.
At the risk of sounding cynical, we did come up with one path, albeit one that is going to be difficult for them to articulate clearly, or at all. Intel is essentially optimizing to give itself options. Right now, it has very few. With enough time, conditions could improve enough that the company has some maneuvering room, but until then they have to pursue multiple goals simultaneously. Hence the Investor Day focus on all their new products. None of them alone could save Intel, so the company has to push all of them. This is feasible, the company has the resources to accomplish this, but they are not a solution in their own right.
More to the point, the company’s fab strategy becomes an important piece, but not for the obvious reasons. Ask almost any analyst and they will all come to the same conclusion – Intel should split in two. No other company is left pursuing the IDM model, encapsulating both chip design and leading edge manufacturing. Intel refuses to do that. And as much as they dress it up, the reality is that the fab side of their company would not survive today. The design side would be immensely valuable, but the fabs would sink and quickly get acquired. So the company appears to be pushing on this front to get the fab operations to the point they could stand on their own. The problem is this is immensely expensive. And so they are looking for all the government support they can get. If the company split in two today, neither side would merit much in the way of government attention. The design side would not need government support, and the fab side would look too precarious for any government to
subsidize invest in. Taken together, they look like a healthy, cash-generative business which can maximize the return on any government largesse, an Investment In Manufacturing.
Our opinion is that over the long-term, the most likely scenario is that Intel is eventually split in two. There is just no way that Intel can support their fabs, IFS is a decade away from being meaningful in this. By delaying that split, the management team is making the bet that they can maximize value in a way that splitting today cannot.
The fact that Intel doesn’t have a broader strategy like Microsoft or Cisco could be said for many in the fabless sector, right? Few companies have some real unfair advantage (Apple the iOS/Mac, Qualcomm has cellular, Nvidia has software, Intel had/s x86, Samsung has the Galaxy), but unless you’re in a niche market where most of your competitors gave up (I’m no expert in RF, but it seems the case, right?), you’re playing knife fight forever.
Another way to point is: what could be the sources of leverage for Intel? All their options are not great: x86, IDM, being big, being American…
My cynical take is that Intel strategy is going into many knife fights in hoping to win a few: bring talent, work harder and just make better chips than AMD.