Intel – Back to basics.

Intel rides the perfect storm.

• Intel has made maximum use of both tensions with China and the current semiconductor shortage to push its new foundry strategy which is going to take some time to come to fruition and left the really big question unanswered.
• Intel held its Intel unleashed event at which it detailed how its current roadmap has been fixed as well as its 2^nd stab at creating a foundry business.
• The main details are:
  • First, Foundry 2.0: Intel will spend $20bn to build two Fabs at its Arizona facility to be the foundation of its new Intel Foundry Services business division.
  • This division will be almost entirely autonomous where its CEO will report directly to Pat Gelsinger in order to maintain its independence.
  • The exact process nodes have not been disclosed at this time, but it is clear that they will need to be leading edge as this is where the demand is and where Intel needs to be if it wants to win its crown back from TSMC.
  • Further expansion into the EU and RoW will follow in time.
  • Second, Advanced manufacturing: Intel reiterated that it had fixed its problem with 7nm manufacturing and gave details around the strategic choices it made that led to this problem.
  • Intel also stated that the majority of its products would be on 10nm by the H2 2021 putting it in line with TSMC’s 7nm process.
  • TSMC and Intel grade their processes differently but in real terms, TSMC’s 7nm and Intel’s 10nm are broadly equivalent as they both produce 90m transistors per square millimetre.
  • Intel expects to have 7nm (TSMC 5nm) up and running in 2023 which could also coincide with the two new fabs that are breaking ground in Arizona this year.
  • Third, Q1 2021 guidance: where Intel expects to exceed both top-line and bottom-line guidance comfortably thanks to continued demand for laptops driven by the homeworking and home-schooling trend.
  • Intel has kept full year 2021 unchanged due to the uncertainties being caused by the widespread semiconductor shortage which could slow sales of products that use Intel components.
• Intel’s move to build leading-edge manufacturing in the USA and Europe is pushing against an open door and has been very well received.
• Alavan Independent’s research indicates that Taiwan is a major flashpoint with China’s now unambiguous desire to reunify by any means necessary.
• This means that it is now in the interest of the USA and its Western allies to ensure that critical semiconductor manufacturing can also be carried out outside of China’s sphere of influence.
• Consequently, I suspect that Intel has managed to win both federal and state incentives to build Fabs in the USA.
• Furthermore, the current shortage (which I think will be alleviated in a few quarters) has also provided the perfect atmosphere to launch this new strategy and I agree with Pat Gelsinger that the environment and Intel’s commitment to foundry is completely different compared to the last time it took a swing at foundry.
• I also do not see a competitive problem as Samsung has shown for many years that it is quite possible to have different divisions that compete against and sell to the same companies at the same time with no real issue.
• This is why Intel Foundry Services is a completely independent unit of Intel which is very similar to how Samsung does it.
• While I think that this is the right move for Intel, the real question remains unanswered which is: has Apple rendered the x86 processor design obsolete?
• Intel of course says absolutely not but only a side-by-side comparison of the two designs manufactured on equivalent semiconductor processes will begin to answer this question.
• When Apple launched the M1, it demonstrated superior performance and there is a strong possibility that it was able to do this largely because Intel’s missteps meant that its best silicon was 2 generations behind the TSMC 5nm process that Apple had used.
• Qualcomm’s recent acquisition of NUVIA (see here) and the fact that Apple is trying to stymie the progress of this company is an indicator that the wonders it has pulled off with the M1 are more about the 5nm process and skill in chip design and much less about any proprietary IP that Apple has developed based on Arm.
• However, the ball is firmly in Intel’s court to prove that its 7nm (TSMC 5m equivalent) chips will retain the performance edge that has made Intel the market leader for the last 30 years.
• This is a critical question and I suspect that Intel’s future as an integrated device manufacturer (IDM), as opposed to a foundry, hinges on this crucial question.
• There will not be an answer to this for a little while yet and the share price is likely to seesaw between these two outcomes for some time.
• I missed Intel at $44 as a result of this uncertainty and given where it is now, I am chalking this one up to a missed opportunity and looking elsewhere.