There is a lot going on behind the scenes in the semiconductor industry today. There are numerous high profile acquisitions underway, the Trade War, the restrictions on sales to Chinese companies, and multiple facets of antitrust scrutiny across the industry.
One reader asked us why there seems to be a sense that many competitors in the industry seem to be doing their most to apply pressure to others in the industry. He pointed to a fairly large number of companies lining up to comment on matters like Qualcomm’s case in front of the FTC and others lining up to oppose Nvidia’s acquisition of Arm.
We will save our comment on Qualcomm for some other date (not least because the FTC process confuses us to no end). But the case of Nvidia/Arm has some important strategic implications beyond the merely legal.
In general, we think the semis industry is opposed to Nvidia buying Arm. The big chip companies are increasingly consolidated and so are increasingly competing against each other in ever more segments. Nvidia has come a long way from supplying GPUs to become a big player in the data center for AI, industrial and automotive sectors. But this is exactly the market many of the other chip companies are going after as well. No one wants to rely on one of their largest competitors for the kind of key technology that Arm owns. So at first pass, they have many reasons to oppose this deal. Maybe enough reasons to actually mount a real opposition with the regulators.
However, for several decades anti-trust regulations in the US have been fairly hands-off. A year or two ago this deal would have gone through with ease, and while we are not lawyers our guess is this deal has fairly good odds of getting approval in the US.
At the same time, no one thinks Nvidia can really do a good job with Arm. It is a big acquisition of a very complicated company that has not been particularly well run for several years. If you are competing with Nvidia is there an advantage in having them burdened with years of integration and restructuring? We can think of many examples of this, where companies were happy to let their competitors overpay for the wrong deal. So if anti-trust approval is likely, then maybe this helps Nvidia’s competitors.
In many situations we could see the logic of this argument, but there is a key difference here – Arm is too important to fail. All these Nvidia competitors are dependent on Arm processors for their own products. The last thing they want is for Arm to be mismanaged or restructured into a write-off down the road.
One of the biggest critiques we have heard about Arm is that is underinvests in its ecosystem. Arm powers so much of the technology industry and yet it has a poor track of record of nurturing key stakeholders that could really help more things run on Arm. This is one of Intel’s great successes of the past 20 years, they invested heavily in Linux and then dozens of other software companies to make sure those systems run really well on x86 processors. This remains one of their key (albeit shrinking) competitive moats. Want to compete with Intel in the data center? You are going to need a few hundred software engineers to move all the BiOS, databases, firmware, middleware, applications and dozens of others to run on your silicon. Arm has done almost none of this work.
Part of the problem is that Arm probably undercharges for its core products and thus cannot afford the massive investments needed to really grow the ecosystem. But if there is one thing US anti-trust rules look at is companies who acquire a target and then immediately raise prices for the acquired products. It seems unlikely that Nvidia will be able to raise prices for Arm products any time soon.
A common M&A tactic is to reduce costs at the acquired company right away. Again, this would be a huge blow to Arm’s customers. They depend on Arm advancing their processor engines, and this means a heavy continued investment in R&D. When Nvidia gets Arm and hits that first air pocket, finds that gaping hole they missed in due diligence or lose a slew of customers, their first instinct will be to cut R&D. Nvidia is run by smart people, so they may not go down that path, but if you are an Arm customer you have to be feeling nervous. The history of large M&A deals is littered with failed deals, but Arm is too important to risk to this.
Arm’s technology is woven into almost every electronic device we use today. Eventually, companies can reduce that dependence but it will take a decade of rewriting and re-optimizing code to run on whatever alternative they find. There are huge dependencies at stake here.
So Nvidia’s competitors have two worries. They have to worry that Nvidia is too successful with the acquisition and use that success to squeeze others out of every market they are counting on for growth. At the same time, they also have to worry about the opposite – what if Nvidia is not successful and Arm limps along for years. That is an even bigger problem for the industry.