Opinion: Semis Top Five
Qualcomm, Nvidia, Intel, Texas Instruments, and who else?By Jonathan Goldberg
Editor's take: The industry has changed a lot in the eight years since we wrote our first analysis on the top five chip companies. We anticipated semis were no longer a growth industry and the only way for companies to keep growing was to win market share (hard) or buy other companies. This is especially true in semiconductors because most of these companies outsource their manufacturing to foundries like TSMC and GlobalFoundries.
Nearly a decade later, much of the consolidation has taken place and there are few obvious deals left to be done. So you might think our list should be largely unchanged, but that is not the case, but the reasons for the changes are not the same as they were in the past.
Guest author Jonathan Goldberg is the founder of D2D Advisory, a multi-functional consulting firm. Jonathan has developed growth strategies and alliances for companies in the mobile, networking, gaming, and software industries.
Ben Bajarin wrote a similar post not so long ago on the five "most important" semiconductor companies. This struck a chord as we have written similar analysis in the past. Bajarin has a solid list, and we thought his inclusion of Apple was smart and important. However, we have a different take.
Bajarin's criteria for inclusion on the list differs from ours. He looks at companies that are driving or controlling compute platforms. By contrast, our list is based on which companies will survive the ongoing industry consolidation, which is not the same thing. So we won't include Apple or Google, as they are not subject to the same industry conditions, but do merit honorable mentions.
Here is our list:
- The analog duopoly of Texas Instruments and ADI
- A Chinese chip company – TBD
- The smoldering ruins of Intel
Texas Instruments and ADI are easy entries for the list, but are often overlooked. Both companies make a huge array of products that most of us never think about. With ADI's acquisition of Maxim, there are now no other analog companies of their scale. There are lots of smaller companies carving out specific niches which may go on to years of independent profitable growth, or end up as targets of one of these two. Either way, there does not seem to be anything on the horizon to displace these two.
At one point, there were serious concerns that Qualcomm might not be around that much longer. But they have now survived a hostile takeover and embarked on a smart new strategy which likely means they will be a major player for many years to come. We should probably add MediaTek to this list as well, they seem to be in a solid position, but we already stretched the rules of the list with the two analog companies, and MediaTek operates under a very different set of corporate and geopolitical conditions.
Nvidia is the third hold over from our last list, and if anything seems to have only extended their relevance. This is built not only on their dominance of the AI market but with their very ambitious plans to extend their reach throughout the data center.
Before we flesh out the rest of the list, a quick word on two companies that are not on the list. The first is Marvell. We think highly of Marvell, they have been executing well on a solid strategy for many years, but there is now the very obvious question of what do they want to do next? We have seen arguments that they could be either predator or prey in the semis consolidation. Do they continue their path of acquisitions or package themselves up for sale? Neither option is great, there are not many good targets left, nor are there many motivated acquirors. If we had to guess, our sense is that they have done a great job and now want to exit, the alternative is going to require a lot of hard work.
The other company missing from the list is Broadcom. They were on it last time, but now we have to question how much longer they want to be in the semis business. As we have argued, at heart they are less of a semis company and more of a private equity fund that used to focus on semis but is now focused on software. We would not be surprised if somewhere down the road they start divesting chip assets. There are so many more targets in software…
Taking their place on the list is a "To be determined" Chinese chip company. We do not know which one, they may not even have been founded yet, but ten years from now there will be a global-scale, highly competitive Chinese chip company that everyone has to pay attention to. Of course, geopolitics could throw a wrench in that vision, but absent a drastic further escalation we think it is very likely that at least one of the thousands of fabless companies in China today survives the gauntlet to emerge as a world player.
And that brings us to Intel. We are increasingly of the view that Intel cannot survive in its current form. We are not happy about it, but our feelings do not factor into the cold, hard reality of the business. Of course, there is still tremendous value in what Intel has, they have so much talent, there are some key assets that will survive. Whether through some miracle the current company makes a comeback, or more likely they are split up and acquired by others, that asset will eventually generate value for someone.
This list has a fairly narrow focus – fabless chip design companies. The broader ecosystem is already fairly well covered elsewhere. The whole world now realizes how irreplaceable TSMC and ASML have become, and so we have not looked at the wafer fabrication equipment (WFE) space, which looks unlikely to change any time soon.
Similarly, we have not touched on the memory sector because it has been fairly stable for a decade. That may be changing now, with Samsung seemingly breaking the long-held truce in the sector maintaining its capacity expanding capex while its peers are cutting sharply, and Western Digital struggling to digest its acquisition of Sandisk. Coupled with the abrupt curtailment of China's memory companies (notably YMTC), we may see some change in memory soon, but we will leave that beyond our scope today.
Finally, we need to touch on the subject of all the non-chip companies designing their own chips.
Apple remains the best run semiconductor company on the planet, albeit with signs of recent stumbling. Similarly, Google removes the most innovative semiconductor company on the planet, whose efforts to broaden the pool of semis designers could alter the industry entirely somewhere down the road. Lastly, Amazon's AWS is another contender for this list having done more than anyone else to bring Arm CPUs to the data center, with the heft to alter the platform dynamics of the whole industry.