Semiconductors – Limiting Factors; Supply Chain & Talent- Will Limit Stock Upside

– If chips are “as good as it gets” so are the stock prices
– Are we at a near term ceiling that stocks have bounced off of?
– If growth slows do valuations also slow?
– Are we in a holding pattern waiting for a down cycle?

Second order derivative investing

As we have said many times in the past, investors don’t just look at growth rates they look at whether the rate of growth is increasing or decreasing ( the second order derivative) especially in cyclical industries such as semiconductors, as way of trying to gauge where we are in the cycle and whether or not it is beginning to turn.

While its hard to imagine a down cycle when things are so amazingly good, we have to keep reminding ourselves that all good things (and cycles) come to an end. The semiconductor road is littered with the bodies of management and analysts who at one time said “the industry is no longer cyclical”, or “its different this time”.

While we are not saying the cycle is over, we are saying that the second order derivative , or rate of increasing growth may be slowing due to a number of limiting factors.

ASML analyst day – Maybe perfect isn’t good enough

ASML is in the best position of almost anyone in the industry. A virtual monopoly with huge demand with all cylinders firing. Then why did the stock falter? Its clear that investors were looking for ever higher guidance and ever faster growth and there was no blood left in the stone to give.

There is clearly a finite limit on how fast the company can grow given its hugely complex supply chain and need for more incredible talent to make it happen.

The reaction of the stock should be a bit of a warning to other companies in the industry as to investors reaction to slowing growth or at the very least, lack of accelerating growth.

Semiconductor supply chain is the most complex in the world

As compared to any other industry on the planet, the manufacture of semiconductors is by far the most complex, interconnected and widespread. We have also found out more recently exactly how delicate the supply chain can actually be.

A $10B fab has tools and parts and materials and workers from every corner of the globe. There is no country that can build semiconductors without the help and cooperation of many other countries suppliers. The auto industry is very, very simple by comparison.

That supply chain has depended upon an unimpeded flow of goods and services to keep the industry running….that is until COVID hit. Surprisingly semiconductor production hasn’t seen the hit that auto production has seen and in fact has grown versus last year (which is an unfair comparison but even without the unfair compare).

JIT -“just in time” and Kanban…all but dead

The concept of having components arrive just in time as needed by production was started by Toyota in the 1940’s and hit its peak a while back in the electronics industry. It works when the supply chain is very reliable and flexible. It reduces waste (inventory) in the supply chain and improves overall efficiency. The auto industry did a great job with this method.

We have now heard that aside from double and triple ordering of parts that manufacturers are stocking in some cases up to a years supply of components so they don’t get bit again.

This stocking up is just another form of over ordering that hurts the overall system and will cause a huge letdown when it ends.

This also hurts the overall supply chain as it has to accommodate this new cyclicality and uncertainty and there’s a price associated with it.

Fighting over limited supplies

As we have previously pointed out the auto industry shouldn’t be too shocked when they are a low priority for chip makers. Making low margin, 50 cent chips, is not a turn on nor priority for chip executives.

In the near term , prices will go up as manufacturers take advantage of the shortages…but not forever.

Semiconductor value in the bill of materials for an auto exceeded the value of steel in an auto years ago and EVs accelerate that trend.

Tesla appears to have a better handle on chip supply perhaps because its headquartered in silicon valley and understands the technology supply chain better than people in Detroit who are closer to the mid western steel mills.

Limited supply limits company successes

We had pointed out in a recent note that other types of supply chain limits may limit the future of some companies.

Most notably Intel and EUV tools supplied by ASML. If Intel wants to retake the Moore’s Law lead and be a real foundry player its needs those EUV tools that are in short supply.

Right now we would guess that Intel has about 6 (or so) viable EUV litho tools while TSMC has more like 60ish tools in its fleet , or roughly ten times Intels capacity.

ASML will likely ship about 50 EUV tools in 2021. This means that Intel would have to buy up an entire years production of ASML with others, such as TSMC, Samsung and memory makers buying zero, just to catch up to where TSMC is today….

The math simply doesn’t add up for Intel……..Intel can’t catch TSMC’s capability unless it can catch up in EUV tools. With plans for $100B in spend, TSMC isn’t going away any time soon.

While Intel could catch TSMC in Moore’s Law technology they can’t really catch TSMC in leading edge capacity due to the disadvantage in EUV tool count…and thats what’s needed for foundry dominance.

We also remain concerned about Intel’s current pace in trying to catch TSMC on the technology side as we think that Intel’s yields in its 1278 process (whatever that has been renamed to) are in the single digits. Gelsinger is on the right track but needs to step even harder on the accelerator as the honeymoon period will be over soon.

Supply Chains woes are clear but talent woes are not yet recognized

The press, analysts and the market in general have gotten the message about the supply chain problems. Those problems have been there for a very long time but no one paid any attention until they rose to the surface by the Covid Crisis which was the catalyst that lit the fuse.

A larger and more difficult to solve problem in the semiconductor industry, particularly in the United States, is that of lack of talented manpower needed to push Moore’s law and the industry to its limits. Supply chain issues can eventually get fixed but talent issues are much harder and longer term in nature.

Back , many years ago, when I was graduating as a newly minted electrical engineer, getting a job offer from Intel was a dream come true and the pinnacle of the profession coveted by all. Today, not so much.

If someone graduating with a technical degree today had a choice of working at a hot quantum computing start up in Palo Alto, that could change the course of computing, or work inside a fab in Portland, babysitting an expensive temperamental litho tool, the choice is pretty clear.

The average tenure of a fab worker in Asia is several times that of one in the US.

Many years ago when working on the IPO of SMIC in Shanghai I was at the fab on a “hiring day” when there was a huge long line of people (mainly female) stretched around down a hallway. Everyone was clamoring for a job and an opportunity to live in dormitory next to the fab so they could roll out of bed to their jobs. It was heaven as compared to the more rural areas of China that most had made the trek from. SMIC had their pick of the best and brightest.

One of the key problems that Global Foundries had that contributed to its failure was attracting semiconductor talent to the rust belt area of upstate New York. New York did a great job with building CNSE around SUNY Albany and trying to build a critical mass of technology know how. But Global Foundries essentially killed off most of that when it gave up on advancing Moore’s Law, fired all its advanced R&D people and sold off its EUV tools, shutting down the future. It would be difficult if not impossible for GloFo to try to get back in the race as the most talented in the industry would much rather go to Arizona or Texas to companies in the race already.

The fight for talent in Arizona is going to be very, very tough as applicants will have two fiercely competing juggernauts, Intel & TSMC to woo them.

Obtaining great semiconductor talent will be harder than the fight for EUV tools.

Morris Chang, the founder of TSMC in Taiwan recently was in the news for making negative comments about US talent by saying “The United States stood out for cheap land and electricity when TSMC looked for an overseas site but we had to try hard to scout out competent technicians and workers in Arizona because manufacturing jobs have not been popular among American people for decades,”

While many in the US were offended by those comments he is not entirely wrong……..we just haven’t realized it yet…..

Even if we could get our hands on the EUV tools needed, we may not be able to get the right people to operate them…..

The Stocks

The reaction of the market to ASML’s analyst day underscores our concern that the semiconductor sector is past its peak.

This is not even taking into account the current “tech wreck” of the overall tech stock sector.

Although there’s still a chip shortage and lots of chips and chip tools to be sold, investors may have already tired of the group as a whole or at least tired of paying high growth multiples for a group whose growth is as good as it gets and isn’t getting any better from here.

I see no near term or even medium term reason to own Intel as we don’t think there is any positive news versus where we are now and the downside beta is higher than the upside beta as we could start to hear news of the renewed effort running into problems.

ASML still has European investors willing to support the stock but the upside may be lacking a investors clearly voted with their feet on less than stellar guidance from analyst day. We don’t expect a more positive tone coming out of the September quarter report so again likely more downside beta.

We think that equipment companies such as AMAT, LRCX and KLAC are going to be hard put to come up with positive enough outlook or guidance to continue to support the higher valuations.

Chip shortages will continue to stay in the news flow but chip and equipment companies will not see their valuations driven up further as the salvation of those shortages as we are over that story.

Over the longer run investors need to be careful that chip and chip equipment companies may finally be caught up in the supply chain (and talent) shortages that thus far has only impacted their customers and not them.

Also Read:

ASML is the key to Intel’s Resurrection Just like ASML helped TSMC beat Intel

KLA – Chip process control outgrowing fabrication tools as capacity needs grow

LAM – Surfing the Spending Tsunami in Semiconductors – Trailing Edge Terrific

Share this post via:

• 
• 
• 
•