The semiconductor industry has strong fundamentals, so the state of the economy is what will really drive the rebound. "Rebound is a certainty, but its timing is not."
Speaking at his mid-term semiconductor industry forecast seminar in London this week, Malcolm Penn, chairman and CEO of industry analyst Future Horizons, assured attendees that industry fundamentals were sound, and after a fall of around 15% in 2019, the industry will rebound with around 4% revenue growth to $414 billion in 2020.
He said, “The fundamentals are sound. In terms of IC unit growth, fab capacity and average selling price, they are all in in good shape. It’s the timing of the upswing in the economy that puts it into doubt.” He added, “Rebound is a certainty, but its timing is not.”
Penn offered his analysis on everything from trade wars to fab capacity and industry trends. He believes the US-China trade war will end soon, since the industry won’t be able to sustain it long-term. In particular, he cited the food chains of companies like Apple. Despite whatever moves companies might be making to rebalance their production centers, he said China is still vital for Apple and others in the US economy. He added that what America is doing now is what Europe tried to do 20 years ago, which didn’t work then. “The trade war is addressing the wrong problem.”
He doesn’t see China being self-sufficient in ICs any time soon. “But in 10 years it will be very different to what we see today. I don’t see any reason why it cannot be a major force. It has the size, it has the talent, it has the money.”
Industry fundamentals have been consistent for decades
Penn dismissed those who suggest that the emergence of new markets and huge opportunities from the internet of things (IoT), artificial intelligence (AI) and 5G means we’re coming out of the pattern of semiconductor industry boom-bust cycles. He said, “Industry continues to be a cyclically driven market, and nothing has changed. It’s cyclical by nature, it’s built into the DNA of the industry. IC growth rate has been a consistent 10% a year. It’s the fundamental heartbeat of IC unit demand.”
This year’s sharp correction of -8% is because of memory overbuild and inventory overbuild. “A bounce-back is inevitable for 2020 — it’s a good omen for next year,” he added. He also noted that chip vendors plan for these corrections, so that they can ‘take it on the chin’ in the down cycle.
The other big consistent factor throughout the history of the semiconductor industry is the average revenue per square centimeter of silicon. “It has always been constant at $9 per square centimeter,” according to Penn.
However, one of the challenges for TSMC is what’s next in its roadmap. Penn said, “TSMC is running out of classic market as its market share as a foundry business is probably as high as it can get. They’ll be under pressure to grow. Maybe they could get into new product markets — such as embedded memory.” Penn emphasized TSMC hasn’t run out of steam, but it needs to think more out of the box for future development.
As for Moore’s Law, he said it is far from dead, despite lots of commentary on its demise. He highlighted continual innovation and said silicon will continue to be the baseline for the next decade. Citing 3D NAND flash as an example, he said Moore’s Law never said you halve transistor size every two years — upwards is just as valid using smaller transistors. The number of layers is limited only by how accurately you can etch deep cylindrical wells for the channel holes through all the layers. He added that Moore’s Law is still going strong at many of the major players such as GlobalFoundries, Samsung and TSMC.
Technology and industry trends
Penn also gave his perspectives on some of the industry trends and on Europe. On AI, he said this will move on from being GPU-centric, since they are not power-efficient. While Nvidia has been a key beneficiary of AI interest to date, Intel has been making numerous acquisitions in this space, and nVidia will need to respond, possibly by acquiring one or more of the leading AI chip startups.
On robotics and automation, Germany’s Industry 4.0 initiative is a good example of how IoT is applied, and the country leads in advanced automation. Meanwhile Japan leads in robotic devices used in factories worldwide to make other products. Penn suggested that President Trump’s attempts to get manufacturing back into the USA could trigger a new wave of robotics.
On automotive, Penn said, “The automotive industry has probably been one of the biggest drivers of the semiconductor industry ever," apart from mobile phones in the past. He highlighted how it’s good news for the semiconductor industry as driver assistance systems are now moving down from high-end vehicles to mass-production low-end vehicles. In addition, the European Commission mandating of various safety systems has driven prices down; ETSI has also published a standard for radar systems so these are probably the next to be mandated in all vehicles.
The result is that sensors, processors, actuators and motor drivers for automotive applications are one of the fastest growing semiconductor markets. Bosch and Infineon are doing particularly well in this market, although NXP and STMicroelectronics also do well in the non-mission critical segments, notably in infotainment systems; Renesas and Texas Instruments are the other major players.
Electric vehicles struggle, hydrogen green credentials
The one bleak sector of automotive is that the fully electric vehicle market is still struggling, Penn said. Apart from some bright spots like Norway, the cost of electric vehicles plus the lack of charging infrastructure is still prohibitive in most markets. He added that Tesla has recognized this and reduced staff levels whilst trying to engineer cost out of their vehicles.
Hydrogen could be the dark horse in all of this — it has a lot of practical benefits, is more efficient, and inherently much cleaner. A delegate from California highlighted anecdotally the visibility of more and more hydrogen cars on roads in the state.
Mike Bryant, CTO of Future Horizons, suggested the green credentials of electric vehicles may not be so green, especially since the electricity comes from LPG-burning electricity generators. And then there is another waste product with electric cars nobody wants to discuss — old batteries. He said, “Eventually there will be too many of them to use up, and the reused ones will be totally dead and need reprocessing. The lithium is worth extracting but the rest will go to landfill.”
On the other hand, hydrogen is totally viable provided you have lots of sunlight, and the cost is almost zero once you install the infrastructure in the world’s sunniest places. Bryant comments, “Saltwater from the Arabian sea/Indian Ocean/Panama Canal/Atlantic/Northern Australia Sea can be pumped a few kilometers inland using solar power to biotech plants that split the water into hydrogen and oxygen again using sunlight. The hydrogen is then pumped into the same ships which are already used to bring vast amounts of LPG. The oxygen can be used for industry or released to the atmosphere. So, it’s totally viable provided you have lots of sunlight. The plants would sit behind the housing on the seafront so don’t use up valuable sea-front acreage. Also, by recombining the hydrogen and oxygen you have an instant supply of pure water for the local community.”
Europe losing ground?
Penn suggested that while Europe might still be leading the way in terms of fundamental R&D, it just doesn’t have the formula for commercializing it. He commented, “Generally Europe still seems to under-invest in the technology to get it to scale.” In addition, he said some of the strategic European projects aren’t anywhere near as aggressive as the original goals of the projects 20-30 years ago. “Semiconductors are not really of national strategic interest. Now we are much more applications driven than technology driven. A lot of the programs and structures in the EU have been disbanded.” Because of this, Europe is a shadow of what it was, he suggested.
Penn added that Europe missed the boat on autonomous vehicles, indicating there were too many groups, too many activities, and too much duplication of effort. “It could have been a GSM moment if we did have a European mandate for companies working together on a single system.”