The big news in semis yesterday was Intel’s announcement that it was ending its bid to buy Tower Semi. As with so many deals in recent years, China’s anti-trust regulators threw the wrench in the works, blocking the deal.
It should come as no surprise that no one is really clear why China blocked Intel’s deal. On the face of it these companies have almost no overlap, with Tower providing foundry capacity to trailing edge chips, while Intel has zero trailing edge capacity. Tower is tiny compared to Intel, nor does it have any magic technology, so there is no way to see the combined companies dominating anything in any anti-competitive sense. The deal sailed through Western regulators.
The seemingly obvious reason that China blocked the deal was that they have been blocking all the Western M&A deals that they can. This policy began during the Trump-era trade war and is almost certainly a deliberate retaliatory practice from China which does not have many good options for retribution. There had been some hope that China was walking back its no-approvals policy with last week’s announcement that they would allow MaxLinear to acquire Silicon Motion (news which tanked Max Linear’s stock given the very bad state of Silicon’s Motions latest financials). So if geopolitics is still the overriding policy of China’s anti-trust regulators, maybe Silicon Motion was just a fluke.
However, there is another theory making its rounds in some circles. As much of China has been cut off from access to TSMC’s advanced manufacturing, the one area where China has some level of self-sufficiency in semis is in the trailing edge processes which Tower provides. China actually has as much trailing edge capacity as the US, and at some nodes has significantly more. So maybe China is willing to allow US M&A again, but does not want to see US companies add to their trailing edge capabilities, thus allowing China to maintain its edge. This is full Tin Foil Hat territory, and there is no way to gauge if anyone in China’s government really thought this way. However, if true, it is a largely meaningless gesture. Tower is based in Israel, a firm US ally, and if you add up the trailing edge capacity of all US allies, combined they have farm more capacity than China.
Regardless of the reasons behind it, Intel is not going to be able to get its hands on Tower, and our apparently contrarian position is this is a good thing. Intel does not need Tower and it would have proved a distraction for a company that has much bigger problems to solve. One of the three existential problems Intel faces is the need to build up its 3rd party foundry business, Intel Foundry Services (IFS). As we wrote about years ago, the biggest challenge IFS faces is building a customer service capability. Intel’s fab managers have never needed to serve outside customers and lack the capacity to do so. They can learn it, they can build it, but it will take time. As much as Intel had any rationale for acquiring Tower it seemed to be that Tower could help IFS accelerate those capabilities, with rumors that they planned to put Tower’s CEO in charge of IFS. To put it diplomatically, we think there was little chance of that plan working. Tower’s fabs operate at a far remove from Intel’s practices. Blending the two companies’ operations would have proven beyond challenging. We are not disparaging either side, it is just that the gap is too wide.
For years, a key tenet of Intel’s operations was to build and run all of its fabs in exactly the same fashion. And while this had certain benefits in scaling up new processes quickly it also mean that each time the company advanced a process node it abandoned the previous node. As such, Intel has no trailing edge capacity today. [Feel free to insert any joke about Intel’s current process being trailing edge.] By contrast, when TSMC launched new processes it built new fabs and kept hold of all the old processes. As a a result they have a sizable trailing edge business, larger than their leading edge business. Intel’s CEO recognized the importance of this, and probably saw Tower as one way to jump start that side as well. Again, this seems misplaced as the gap is just too wide.
All in all, we think China did Intel a big favor by blocking this deal. It was unlikely to achieve the results Intel intended and would have become a major drag on resources and management mindshare.