Gelsinger must take the long view to re-engineer Intel
Some will observe that with the re-employment of Gelsinger Intel is swapping a finance executive for one with a technology background. Others may observe that Intel has reacted to the complaints of an activist investor with alarming haste. But as is often the case in semiconductors, the situation is more complex than it first appears.
The roll call of Intel CEOs reads:
Robert Noyce, Gordon Moore, Andy Grove; these were effectively the three technologist founders of Intel in 1968 and were CEOs in turn. Seen with the benefit of hindsight it may seem like they did no wrong. They, of course, helped drive the creation of the semiconductor industry with Intel as its leading disaggregated company but there were some strange and unsuccessful diversions into ventures such as LED wristwatches.
Then in 1998 another engineer, Craig Barrett, took the helm. Barrett had been responsible for the copy-exact strategy that had let Intel increase manufacturing capacity to meet rapidly increasing demand for microprocessors for personal computers.
Paul Otellini took the top job from 2005 to 2013. He had started with Intel in 1974 in sales and finance and, for some, the consumerization of semiconductors made Otellini the right choice. Others said it was a turning away from Intel’s necessary deep connection with semiconductor manufacturing.
This may be seen as the reasoning behind the appointment of an insider and a manufacturing guy, Brian Krzanich in 2013. But in 2018 it did not end well (see Intel CEO resigns over past relationship) and it was on Krzanich’s watch that Intel’s manufacturing stall accelerated. The company had numerous delays in standing up its 10nm manufacturing process and fell behind TSMC and Samsung.
In 2018 Bob Swan, CFO, was appointed interim CEO. Some have complained that it took Intel 50 years for Intel to fall from being a creative company led by technologists to a moribund entity led by a finance guy. In Swan’s defence he has played the cards he has been dealt as well anyone could. It is not clear his successor Gelsinger will be able to do any better in the short- or medium-term. But Intel’s problems are long-term.
One could argue that Intel’s problems with manufacturing started under Otellini and were not obviously addressed under Krzanich.
Next: Not years but decades
Because in semiconductors time-lines are measured in decades and often require consistent application of strategy over periods that are longer than the time executives and directors occupy their comfortable chairs. In Gelsinger’s time as CTO under Otellini he always struck me as much more of marketer than a traditional CTO. Intel liked to roll him out to speak but he would rarely talk about technical detail for a company that was increasingly pitching to consumers rather than engineers.
So what is the problem that Gelsinger has to begin to address?
In essence Gelsinger must decide whether to double-down on manufacturing because that allows Intel to make maximum margin on products, or to exit manufacturing and save the eye-watering R&D and capital expenditure and accept sharing the margin with foundries. Some may feel that the appointment of Gelsinger, an executive with an engineering background, means that decision has been taken in favor of maintaining chip manufacturing. But hold hard.
In 2008 long-time x86 rival Advanced Micro Devices took the decision to spin off its manufacturing operations to form Globalfoundries Inc. In recent years AMD has reaped the benefit of that decision, gaining market share as Intel struggles. AMD’s Lisu Su has an engineering background but she is leading a successful fabless chip company.
In 2020 the hedge fund Third Point took a $1 billion position in Intel because of its financial underperformance and then at the end of the year CEO Dan Loeb wrote a highly critical letter to Intel’s chairman complaining that the company had lost its lead in manufacturing, while overpaying executives and called for the divestment of failed acquisitions.
The whirlwind that Intel is now reaping was sown years ago. Certainly, Intel has paid some of its CEOs multimillions of dollars a year across multiple forms of compensation, which might seem an unnecessarily large amount of money for anyone to be paid. But Third Point is pointing out the obvious and offering no specific strategy; only calling for an independent advisor to be appointed to help Intel consider its strategic options.
From a financial perspective that is only likely to lead in one direction; towards the further separation of manufacturing and design and the possible divestment of chip manufacturing; following in the footsteps of AMD. But Third Point is trying to have its cake and eat it. It complains that Intel has lost ground to Apple-TSMC and the fabless-foundry axis the pair represent, and then calls for strategic solutions.
And there is also a geopolitical perspective to this. The US government appears ready to subsidize on-shore manufacturing and will be keen to keep a domestic champion in the chip technology race.
Intel’s problems were created when it was so successful in the x86 processor space it could not only command a high price for its processors but could dictate to Taiwanese motherboard makers and computer companies the design, features, price and introduction timetable of computers made with those motherboards.
At the same time Intel got into the habit of overpaying in many walks of life when it could afford to do so because its semiconductor manufacturing leadership was like the proverbial license to print money. Meanwhile it made several failed attempts to get into other markets, including foundry.
Through its Intel Capital wing, Intel has also been a main contributor in creating a booming startup culture. While the billions of dollars of investments that Intel has made may have benefitted society, it is not clear how much they have benefitted Intel shareholders.
Intel has overpaid to acquire businesses to try and leverage it into new sectors and then seen the promise squandered. Now it is no longer a technology leader and some – Apple being the perfect example – no longer require its product and services, it has to change its culture.
Next: There is a tide . . .
Intel’s situation has certain similarities to the ebbing tide of semiconductor manufacturing in Europe, although markedly less acute (see Opinion: Money’s not the problem for Europe’s semiconductor rebuild).
For both Intel and Europe, catching up with the leading-edge in semiconductors before there is some major change to the materials and manufacturing requirements at 2nm or 1nm is unlikely. Indeed, a major change in the architecture of electronics/photonics/spintronics is the best opportunity for re-engagement for both, as it can level the playing field for all would-be participants.
Gelsinger spent 30 years in his first spell at Intel during which he led numerous microprocessor design programs. If he is to make the changes that shareholders and the board of directors would like to see, his vision – if not his time as CEO – will need to be of the same order of time. And he may yet be forced to conclude that Intel has to quit leading-edge semiconductor manufacturing in the short term with a view to catching the flooding tide and coming again in the post-CMOS era.
Related links and articles: