For the past year, semiconductor industry sales and semiconductor industry stocks have rocketed higher.
Industry insiders and analysts alike have offered numerous reasons why, from the impact of Covid-19, to the digital transformation of the world, to big government spending to secure national supply chains.
Amid all the fuss, the looming question is: How long can the boom last?
Which leads to related questions: Are the trends listed above real or overblown? Are they long term or transitory?
When chip production equipment maker ASML held its earnings call last Wednesday (7/21), CEO Peter Wennink, said the underlying growth trends are very real.
“I think the future for the industry looks bright… semiconductor makers currently have a combined sales number of about US$500 billion. That could be a trillion dollars by the end of this decade,” he said.
His comments were echoed a day later by Pat Gelsinger, the CEO of Intel, the world’s largest chip maker.
“We are only in the early innings of what is likely to be a decade of sustained growth across the industry,” he said during the company’s second quarter earnings call.
That’s a bold prediction for the notoriously cyclical semiconductor industry. It generally follows the business cycle, with shortages prompting the build out of new factories, leading to over exuberance and overproduction, then falling prices.
Mark Liu, chairman of TSMC, the biggest chip foundry, has taken a different view of the industry. In an April speech, he said the shortages were caused first by the U.S.-China trade war, as some companies pulled orders from China chip suppliers, such as SMIC, which was blacklisted for allegedly working with China’s military.
U.S., European and other companies that had chips produced in China feared their suppliers might be next, so shifted orders out of the country.
That left idle semiconductor factories in China.
Further, Liu said, the electronics boom caused by the work-and-study-from-home/play-at-home economy during the pandemic added strong demand to a supply chain already in disarray from the trade war.
The idea that the shortages will last for years and that technology trends will provide a decade long tailwind doesn’t mesh with Liu’s view.
He believes there’s plenty of unused capacity and says TSMC has seen double-booking in the industry – caused by a fear of being unable to get enough chips.
Double-booking is normally a sign the semiconductor cycle is turning down. But normal times these are not, what with trade wars and the pandemic.
Finally, he argued that if nations globally try to build their own semiconductor supply chains, it would lead to a large amount of unprofitable capacity.
Liu’s viewpoint offers a different way to see the situation than Wennink and Gelsinger, a sort of counterbalance.
The reality is that nobody can tell the future. It’s all about sifting through the evidence and making the best possible moves for the future.
ASML, for one, is in a special situation. It’s been in the spotlight this year as the mainstream media talks up its EUV machines. No other company in the world makes them. You cannot make cutting edge semiconductors without them.
The reason for all the attention is simple. Chips are hot right now. Semiconductor equipment sales have soared alongside the fortunes of chip manufacturers.
The pandemic lifted sales of dozens of electronic items, while at the same time disrupting the supply chain.
The work-and-study-from-home trend saw consumers around the world buying new laptops and PCs, smartphones, Zoom-related devices (WiFi gear, webcams, microphones) and more, while travel difficulties and lockdowns meant play-at-home sales boosted everything from Xboxes and PlayStations to wall-sized flat screen TVs.
It’s been a bonanza for the electronics industry, and that means big sales for the semiconductors that make them work.
The pandemic also put on display how much semiconductors are becoming a part of everyday life. The auto industry, born from mechanization and oil, has been brought to its knees over semiconductors. Some companies today can’t ship $50,000 cars for want of a few $1 chips. The shortage remains and automakers say they’re missing out on billions of dollars in sales.
The fact chips are now common in automobiles is the low-tech part of the digital transformation. Chips are everywhere now, in refrigerators, washing machines, and even small, inexpensive items like toasters, blenders and fans.
The high tech part of the digital transformation also includes cars, autonomous and electric with LiDAR, and other computing systems to aid drivers. This category also means smartphones, 5G networks, data centers and more.
A number of other emerging technologies are fueling these changes as well, including next-generation 6G wireless, pervasive connectivity, cloud infrastructure, artificial intelligence (AI), high performance and distributed computing, and more.
“All those distributed systems that need a level of compute next to the sensors. That is a secular trend that will not go away very soon. Actually we think it will continue throughout the rest of the decade,” Wennink said.
Digital nationalism is the final big trend most pundits agree on. China, for example, wants to build something nobody else has, an entirely self-sufficient semiconductor industry. Other places, such as the E.U, Japan and U.S., want to shore up semiconductor manufacturing to ensure supplies. The issue has become a national priority in some places, partly due to the shortage among automakers.
The pandemic has shone a spotlight on how politically important the auto industry is. Thousands of good paying jobs have been threatened by a chip shortage at the very moment of mass unemployment caused by a pandemic.
Counterpoint Research earlier this year also highlighted a problem of under-investment in the semiconductor industry. The factories needed to make most automotive chips are far less advanced than the ones required for cutting edge ones.
Low prices that may have been good for auto makers have not allowed enough profit to entice chip makers to expand those lagging edge factories.
The situation has led to years of under-investment, Counterpoint says.
Chip makers don’t want to expand production without long term contracts with customers that help pay for the new lines.
Taiwan’s UMC announced such a deal in April. It will invest $3.5 billion in a chip factory because customers made deposits to secure their long-term chip supply. The money will mainly pay for production line machinery.
Such deals are necessary because making semiconductors costs so much.
The semiconductor industry is the most expensive industry in the world. TSMC, the world’s leading chip manufacturer, figures it will spend $20 billion on its latest 3-nanometer chip factory, a staggering sum.
Other industries don’t even come close. The newest aircraft carriers for the U.S Navy, called Gerald Ford Class carriers, cost $13 billion each.
TSMC could buy one-and-a-half aircraft carriers instead of its new chip factory. Or two nuclear power plants ($9 billion each for a 1,100MW plant), or five fully integrated, state-of-the-art steel mills ($4 billion each).
Production equipment is the most expensive part of the semiconductor industry. ASML’s EUV machines cost around $150 million each. Each.
No wonder the company’s stock has done so well this year. The shares closed Friday (7/23) at $748.14, up 53.4% so far this year.
Sales for equipment makers have been so brisk, their main industry association, SEMI, has had to sprint to keep its forecasts relevant.
In its recently published mid-year forecast, SEMI raised its full year 2021 equipment sales estimate to $95.3 billion, up 34% over last year.
Thirty-four percent is a huge jump in a multi-billion dollar a year industry, and marks the highest rate of growth in years. It’s also quite a leap from SEMI’s previous forecast from last December, $71.9 billion.
Next year will be even bigger, SEMI says. It forecasts 6 percent growth to $101.31 billion, the first time ever over $100 billion. Back in December, the 2022 forecast was $76.1 billion.
The numbers are rising so fast, they might be confused for irrational exuberance. Can these forecasts really keep growing? Can equipment makers really produce that much machinery?
ASML’s orders-on-hand for EUV machines already cover 80% of next year’s output. It seems to be hitting the limits of growth.
Wennink said in the near term, aside from selling out inventories, the way ASML is boosting output is by speeding up work along its entire supply chain.
He was able to raise ASML’s full year sales guidance to 35% growth, from 30%, by squeezing every drop of inventory out of his supply chain. Getting everyone moving faster takes more time, but improvements are coming along.
Over the next 12 to 18 months, ASML will focus on boosting production at existing factories by adding workers and machines in needed areas.
Farther out, 2 to 3-years, the company is building new factories to keep up with long term demand.
“So this is what we’re discussing together with our supply chain,” he said.
When asked what would happen if he’s wrong, if the digital transformation is a hoax, and the party ends once the pandemic shortages are gone, he replied that the underlying growth trend is there.
“I’m not concerned about building that capacity," he said. "We will use it.”
A number of other chip makers will host conference calls in coming weeks (Late July, Early August). It should be interesting to hear what they have to say.
_________________________________________________________________________
Disclaimer: Blog posts and other information on TriOrient Investments' web site (3-orient.com) do not constitute investment advice. TriOrient Investments is a private company and does not accept outside funds for investment, nor does it divulge trading activity, nor provide recommendations of any kind to buy or sell any kind of investment product. This material is provided for informational purposes only. The views expressed regarding market, economic, industry or corporate trends are those of the authors and are subject to change at any time based on market and other conditions and there can be no assurances that countries, markets, sectors or firms will perform as expected.
Comments