Five years ago we wrote a piece forecasting that there would eventually be only five chip companies left in the US in a few years. Given the ongoing turmoil in the chip market we want to revisit that thesis here.
The US and European semis industry has been consolidating for over a decade. We have gone from something like 400 chip companies to roughly 100 today. This has been good for the industry, allowing those still standing to raise prices and move away from the perennial knife’s edge margins that saddled the industry after the dot com crash.
From a macro perspective, the case for consolidation still holds, albeit we are now entering its final stages as there are so few companies left to consolidate.
Put simply, there are two reasons to acquire a competitor – growth and cost savings. For the chip companies, both theses hold. Most semis end-markets are now in mature, slow-growth stages. Mobile phones grow in line with GDP, PCs are basically flat, TVs have been largely commoditized. Data center and networking are still growing nicely, but only with a very narrow set of (demanding) customers. AI is a hot spot, but as we note, it feels like the last frontier, and is already a crowded space. And new technologies like AR/VR and autonomous vehicles are still years away. In short, the main avenue for growth for chip companies lies through M&A.
On the cost side, semiconductor manufacturing has strong returns to scale. This is especially true at the leading edge processes like 7nm where only very large companies can afford the upfront costs (>$100m) to participate. In addition, fabless chip companies’ core competency rests in design, but they all need large operations teams to take those designs and move them to manufacture. The more chips you can push through that ops team, the lighter the cost per unit of every chip.
As we have detailed previously, semis economics have improved to the point that it now makes sense for non-semis companies to build their own chips in some cases. The obvious example here is Apple (and here), but also worth noting are companies like Amazon disrupting server CPUs. And this in itself is another driver of consolidation -there are only so many customers left.
And as with so many things today, there is a geopolitical aspect to consider. China is heavily subsidizing its semiconductor industry. There are something like 1,300 chip companies in China today. Most of them will fall under some future wave of consolidation, but some of them will emerge as full-scale global competitors. US companies will need the scale to compete.
There are a lot of reasons for chip companies to acquire other chip companies.
The one question we are always asked about this by investors is who will be the next target? That question is tricky to answer as there are so many random variables that affect the timing for a company putting itself for sale. That being said, a quick survey can at least help understand who will be doing the acquisitions.
- Intel – This company obviously has its own share of problems right now, but they still lead the industry in many metrics. For the first time, we are starting to question whether Intel will remain independent. They have to make such large strategic changes, it may be easier to sell the company to someone who is more willing and able to make those changes.
- Qualcomm – Right after we wrote our original piece on this subject, and listed Qualcomm as one of the survivors, Broadcom tried to buy them. Qualcomm is doing well now, but we are in the sweet part of the transition to 5G, a time when Qualcomm does best. As we have repeated, the 5G roll-out is very different than 4G, and Qualcomm’s prospects look a bit more challenging. This is especially true if mmWave 5G deployments take longer than expected (and here). Broadcom is now a US company, and we suspect they may come back. On the other hand, Qualcomm has immense wealth and may eventually return to acquiring others.
- The Analog Duopoly – Texas Instruments and ADI/Maxim will both be too large to acquire the other. We suspect that many smaller analog companies will end up with one or the other of these two.
- Broadcom – Broadcom is an interesting case in that they are really a private equity company, and no longer a chip company. Under their current model, there are really only two semis targets left that can move their needle (i.e. the first two names on this list). If anything, we would not be surprised to see Broadcom shed some chip assets to free up funds for more software purchases.
- Nvidia – After our last post, we kicked ourselves for forgetting Nvidia. No way to forget them now, as they are or are close to being the most valuable chip company on the planet. They have done some M&A lately and are likely to do more. Bold prediction – if Intel cannot turn itself around, Nvidia would be a natural buyer. There are a lot of assumptions built into that statement, our point is that Nvidia has the ability, and likely the appetite, to do something big.
Others – We included NXP on our list five years ago, and then they went an tried to become part of Qualcomm. That did not happen, but we have to question how long they will remain independent. They are on the border between predator and prey, and will probably need to pick one side or the other in coming years.
Another one on the border line is AMD. That company has benefited more than anyone from Intel’s implosion, and its stock reflects that. But they are also on border and will also need to decide soon whether they need to acquire or be acquired.
There are another dozen mid-cap semis stock – ST Micro, Infineon, Skyworks, Qorvo, etc. – who all face their own set of market conditions which are likely to push them into someone’s arms eventually. With stock prices high and debt cheap, we would not be surprised to see another wave of deals soon.