The Fork in Intel’s Road

At an industry level, what happens with Intel going forward has significant implications. That is why, if you follow industry commentary, what is going on with Intel, their CEO shift and the larger discussion about semiconductor manufacturing in the US is getting attention.

It’s worth digging into the debate as to whether Intel should spin-off their foundries, as has been suggested, but I’ll save that for another article. What I want to tackle is more specifically why this is a tough decision for Intel and why they are still standing at the fork in the road.

For context, it bears repeating that foundries specialize in proprietary process technology and chip design companies (i.g Nvidia, Qualcomm, Apple, AMD, etc.) specialize in proprietary architecture. Foundries compete on trying to create differentiated process solutions that customers find attractive. Chip companies focus on creating architectures that differentiate their end products from competitors. Intel is in the unique position of doing with the caveat that their process is largely only used for their architecture.

What makes Intel’s challenge unique is their process and their architecture are both quite good. Despite their process technology not being on leading-edge nodes, the process design itself is quality. Similarly, their architecture is very good. This is highlighted with consistent performance benchmarks that show Intel’s architecture is competitive with others, in performance, even while their process is behind. The only major area Intel loses is in power consumption, which for things like smartphones, and portable PCs matters. But even in some specific designs, Intel is able to keep performance per watt relatively close to competitors even with a process generation that is behind. This would be evidence for Intel to continue its conviction to stay the course in the process given it is not years competitively behind even though their process/manufacturing is years behind.

If it was extremely obvious that their process manufacturing was simply the albatross and unrecoverable then I think this is a much easier decision for Intel. But, for now, that is not the case and performance benchmarks Intel and others keep seeing against the competition is enough to keep them optimistic.

The other wrinkle, that I think is quite interesting, is let’s say Intel does shift to TSMC. Then Intel architecture would compete with others on equal footing. If this happened it really would become a battle of who can design the best chips, not necessarily who can manufacture the best chips. Which, if my logic about the quality of their architecture is true, then Intel would be quite competitive if all things were equal with process technology. But the tradeoff will come with an undifferentiated process technology, which strategically could be an issue if everyone uses TSMC process.

Lastly, Intel staying relevant, if not cutting edge in manufacturing is strategically important for the United States. Semiconductor technology very well could be, one of the most important strategic technology assets for the digital future and if that moves off US soil, there a significant risks involved. Especially, when China is having a hard time with their own proprietary technology manufacturing with SMIC, and if they catch onto how important this is, TSMC becomes a political and national target. Having a plan B, in case anything happens to TSMC should be the utmost concern for all US companies, including the US Government, and as of now Intel is that plan B and ideally, Intel can become Plan A. I do not believe spinning off Intel’s foundries has even a glimmer of hope of making Intel’s process cutting edge again. It could certainly still be a big and viable business but, strategically, Intel process needs to be cutting edge.

Tim mentioned our friend Jim’s article today, where he suggested out of the box thinking about Intel buying Global Foundries. Any move like this has to have with it the direction to getting back to cutting edge process and likely making Arm chips a priority as well.

This, in my mind, is one of the most important dynamics for the US and tech industry at large at the moment. The next few years will be critical for Intel.