Semiconductor industry blues: pips squeaking as chips are peaking
The semiconductor industry is used to disappointment. Now product shortages are driving up prices, not demand.
Rampant in the 1990s, the double-whammy of the dotcom and financial bubbles took a severe toll on the expectations of people in the industry and an even worse toll on the industry’s that serve the chipmakers.
Take the wafer makers, the companies that supply the blanks on which the chipmakers assemble their ever tinier and denser circuitry. A report from SEMI, the group that brings together wafer and processing-equipment suppliers, makes the position clear.
According to SEMI, wafer shipments measured by total area increased 10 per cent in 2017 over 2016. That growth more or less tracks the consistent unit year-on-year growth in chip shipments the industry has seen for decades, barring the odd blip during recessions. They now ship almost 50 per cent more silicon than they did at the peak of the market for them ahead of the financial crash of 2008. The bad news is the money they earned from chipmakers is only two-thirds of what it was in 2007. In 2007, the wafer makers took $12bn in revenues in a semiconductor market worth $270bn. Ten years later, the semiconductor suppliers invoiced for more than $410bn. The wafer makers collected $8.7bn.
This year, according to market analyst Future Horizons, the chip market could surge to more than $500bn. This may not be due to the wafer makers ramping up production even more. Instead, unless the economy suffers a sudden shock, shortages are likely to drive unit prices higher as customers scramble for supplies.
Some of that scrambling has already taken place. The larger OEM and manufacturing outsourcing (EMS) companies do not typically buy from catalogue distributors – those distributors mostly serve a market for prototypes and low-volume manufacturing. When things get tight, manufacturers go where they can. In the autumn of 2017, Graham Maggs, vice president of marketing, Mouser Electronics, said: “We do not promote volume sales. But, in the last year where people are looking for inventory, customers have looked for larger quantities. With OEMs and EMS providers, when they come to us, it’s an indication that there are shortages in the market.”
Growing shortages of stocks have forced prices up solidly for the first time since 2014, according to data from WSTS, to highs not seen since 2005 and towards levels reminiscent of the mid-1990s before a series of crunches took the wind out of the chipmakers’ sails. The intervening years were punishing for the industry and a collective sense of imminent failure is proving infectious, according to Malcolm Penn, president of Future Horizons.
Penn believes a repeat of 2017’s double-digit growth in semiconductor revenues is not only possible, it’s more or less inevitable. It’s worth remembering that a similar shortage-driven surge in revenue was on the cards for 2015, but was choked off by poor economic results. There are fears of the US economy overheating and plunging into recession. The question is whether that might happen quickly.
A number of analysts have taken a more cautious approach than Future Horizons. Gartner and IC Insights and the industry’s own WSTS are looking to a return to single-digit growth for this year, although most upgraded their forecasts during 2017 and into 2018. The caution is infectious.
Samsung, for example, now the biggest buyer of chip-making equipment driven largely by the push towards 3D Nand flash, is expected to curtail spending this year, Penn says.
“People are saying, no matter how good  was, this year will be horrible,” he adds. “No-one wants to take a risk and put their head above the parapet and say it’s going to be a good growth year.”
The problem, Penn says, is that if sales do ramp up as Future Horizons expects, manufacturers will be unable to meet demand for more than a year – risking the overheating of the overall market: “The supply chain can’t react in less than a year.”
This is as apparent in the domain of the wafer makers as in the equipment used to process them into finished chips. It is having curious effects on a business that has the image of pushing technology to its limits each and every year. A number of suppliers are dusting off old 200mm equipment, having originally planned to focus on the larger and, as long as you are making products that supply many large customers, more economic.
“200mm is quite a good wafer for a lot of product types,” Penn says. The question they face is actually getting equipment from suppliers who are wary of investing heavily in new production in case they get hammered the next time things go bad.
Suppliers such as Samsung are bringing up processes such as 65mn on 200mm wafers for devices such as IoT controllers. Since the early 2000s and the 130nm, pretty much all equipment was designed for the 300mm wafer size. No-one has attempted to make a product line that processes 200mm wafers to that level.
“You can buy secondhand pieces of kit for 200mm, that’s workable. But those are now rarer than hen’s teeth. Yet people are saying they are building a 200mm fab. The only thing you can really do is buy redundant 300mm kit and run 200mm wafers on it. So the amount of installed capacity has gone but no-one has invested in it. The equipment people do not see a market in it,” Penn explains, noting there is a similar problem in the supply of 200mm wafers, which is capacity-limited.
Wafer-makers need only look at the intense price pressure they have suffered under more than a decade of low expectations to understand that investing in extra 200mm capacity may simply open the door to another round of gouging when the wider economy goes bad. Without a re-evaluation of the relationship between the players in the semiconductor, they are going to plan for the next recession, not an imminent boom. Like the stock market in recent weeks, this could be a classic parabolic blow off, as prices spiral out of control.