It is something of an understatement to say that the year 2020 has been a rough one. Like every other sector of the economy, the economic recession brought on by the COVID-19 pandemic and the efforts to stem the spread of the novel coronavirus have significantly impacted the semiconductor industry. In addition, from a personal perspective, the timing of this recession seems tailor-made to give high-tech market research professionals a bad name. What happened to the nearly universal picture painted by analysts of a market on the rise? Amidst our many forecasting tables and figures and detailed market segmentation, have we missed the forest for the trees? Well, it’s not that simple. A short history may give some perspective.
What the years have wrought
After years of lackluster growth, the semiconductor industry had a breakout year in 2018. During a year of double-digit growth, the global semiconductor market posted record sales of $469 billion, according to the Semiconductor Industry Association (SIA). The IC packaging market segment, which New Venture Research (NVR) has tracked for more than two decades, likewise did very well, growing by more than 13% and generating close to $70.3 billion.
Beginning near the end of 2018 and through most of 2019, however, the semiconductor industry faced an abrupt reversal, with global sales unexpectedly contracting by double digits. There were, perhaps, a number of factors that created the downturn, including general uncertainty in global economies. While the U.S. economy was in the midst of a record-breaking expansion, many other countries were not doing so well. The global trade war set off by the Trump Administration had been particularly destructive to markets in Asia and especially China, where worldwide IC manufacturing is concentrated. Coupled with such external factors, were a number of internal supply and demand issues. The greatest immediate factor in the contraction was oversupply of devices, with memory leading the downward spiral. Naturally, the IC packaging market was also impacted, but packaging is a more dynamic market, especially among advanced packaging products, and the market segment continued to grow, but at a modest 3%, reaching $72.3 billion.
Fortunately, by the end of 2019, the overall market appeared to be recovering, despite continuing uncertainty in trade relations between the U.S. and China. The sharp decline of the first half of 2019 had softened, and we (and others) expected to see modest to strong growth return in 2020, followed by even stronger growth in the years to follow. For example, at the end of 2019, the SIA’s respected WSTS forecasted that the semiconductor industry would grow by 5.9% in 2020. We were not quite so optimistic; NVR’s reports, forecasted IC devices would grow by only about 2%, and that the worldwide packaging market segment would grow by 4.7% in 2020. However, regardless of the specific numbers, there was a general expectation that 2020 would be the start of a new expansion in the entire electronics sector.
Alas, the expansion was not to be. With a suddenness that shocked the global economy, the novel coronavirus created a whole new reality. First striking in China, and quickly spreading around the globe, the COVID-19 pandemic forced countries to effectively close up shop one after another. As with virtually every sector of the global economy, the impact on the semiconductor industry was quite significant, and wreaked havoc on the carefully structured forecasts of market researchers everywhere.
The silver lining, if it exists, is that some (but by no means all) economists are expecting a return to growth in the second half of the year – resulting in a so-called V-shaped recession. In terms of the semiconductor market, as China and other Asian countries, where the majority of devices are manufactured, reopen their economies, industry observers agree that growth will definitely be slower this year than what was originally anticipated. How much slower, however, is still open for debate. In June, the WSTS published its forecast for the semiconductor industry that anticipates an annual growth of 3.3% to $426 billion (down from 5.9%). Other analyses are not quite so sanguine. Gartner Research, for example, published a forecast in April that forecasted that the market will decline by –0.9% percent in 2020, down from their previous forecast of 12.5% growth for the year. We suspect that the reality will be somewhere in between the optimists (SIA) and pessimists (Gartner).
So, what went wrong with all those rosy forecasts? Why could the people who are expected to know where the market is heading not anticipate where the market would head? And more importantly, why should anyone believe our predictions, now? I’m going to let you in on a secret known only to analysts in my field: market research is not rocket science. OK, this is obvious. Markets do not obey fixed laws of physics; outside or unexpected events—such as the emergence of a new disease that leads to a global recession—can never be factored in before the fact. Even without having something quite so catastrophic take place, there are numerous factors that can only be understood in the perspective of hindsight. With this understanding about the reliability of market research forecasts in these uncertain times, the remainder of this article will present some thoughts and, yes, some updated forecasts from NVR’s own research into the IC packaging market.
Current trends in IC packaging: tentative forecasts, at best
Modern IC packaging products are extremely diverse, encompassing numerous packaging products ranging from simple 3-pin plastic transistor outline (TO) packages—originally designed for discrete devices, but sometimes used for analog ICs such as amplifiers and voltage regulators—to plastic or ceramic IC packages equipped with hundreds of solder “bumps” arrayed across the surface of the package (such as ball grid arrays [BGAs] and fine-pitch BGAs [FBGAs]), to complex multi-chip packages with a thousand or more I/O connections (e.g., system in package (SiP) devices).
As noted earlier, the growth rate for the worldwide IC packaging market was accelerating coming into 2020. But the emergence of COVID-19 put the brakes on the growth during the first half of the year. Still, as the economies of China and other countries begin to restart, we expect at least some of that lost momentum will return. At the end of 2019, NVR forecasted the packaging market would grow at a healthy 4.7%, in this year. However, based on current trends, we anticipate that the overall packaging market will grow by just 2.5% (Figure 1). Presumably, the second half of the year will see improvement, but real recovery will not take place until 2021, as the global economy emerges from the effects of the current recession.
Of course, the situation is still very much in flux, and this forecast presupposes that a vaccine for COVID-19 will have been developed and that pent-up demand takes hold to return to “normal” economic conditions. In truth, the full impact of the coronavirus is still unknown. There are countries where the disease is still not under control – most notably and ominously, at the time of this writing, in the United States. Unless and until a viable vaccine is widely available, no one can predict with certainty what may happen in the coming months, if then. Also, there is debate whether the recession will be V-shaped, with a rapid recovery once the coronavirus has been mastered, or U-shaped reflecting hidden weakness in the economy that predated the pandemic. Moreover, disease is certainly not the only cataclysm that can – without notice – negatively impact the marketplace. One will recall the earthquake and tsunami that struck Japan in 2011 and led to significant months-long adverse effects in the electronics supply chain.
On the other hand, it would be incorrect to conclude that current (and even previous) forecasts are without value. The point is that while we can rely, generally, on the direction that the market is taking, it is times such as these that require closer attention to the details—and not merely seeing only the forest, as it were.
Naturally, total market forecasts provide value; they point out the direction that the industry is moving. But the most valuable information lies in the details (or to carry the analogy further, in the trees) in the many market segments that make up the total market. In the case of IC packaging, the trend (or more to the point, the money) is in advanced packaging types. Take for example the BGA/FBGA market segment. First developed in the 1980s, this substrate-based packaging format is both more flexible, and less costly per I/O than other formats and has come to be the most widely used. The packaging type accounted for nearly a quarter of all IC packaging shipments in 2019, and as shown in Figure 2, nearly two-thirds of all revenues generated in the market.
Moreover, the more advanced FBGAs are particularly dominant, owing to their breadth of possible applications. FBGAs have an I/O-count ranging from four I/Os to more than 2,500 I/Os, and their size is quite small relative to the number of I/Os that they are capable of supporting. These characteristics alone make FBGAs an ideal packaging type to be used with the most advanced 3-D packages, multi-chip modules and system-in-packaging (SiP) formats. No wonder, despite the size of the market segment (with shipments exceeding 60 billion units in 2019), they continue to capture an increasing market share. In fact, only wafer-level packaging is growing at a faster pace than FBGAs, and for much the same reason: chip-scale package (CSP)-size, relatively inexpensive to manufacture, relatively high I/O-counts, and with the added factor of starting from a smaller installed base. These segment trends, as well as many others, on top of the broader industry trends continue irrespective of the wider economic impact of COVID-19 or the recession it has initiated. Helping to understand these trends within trends is the function that market research—with its many tables and figures—provides.
Application markets: the change will be small
The societal impact of COVID-19 has been most obvious in the use of widespread shelter-in-place orders in many countries. People are working from home as never before. Students are getting their schooling online, rather than in the classroom. Cars are left in driveways and garages everywhere, and rush hour (for those who still must drive for their essential services) is a breeze. All these changes have also had a small but measurable impact on applications for IC devices.
An important segmentation of our packaging reports has been the applications for which ICs are ultimately used. In our reports, we identify five major application categories: consumer products, automotive applications, computers and peripherals, communications, and industrial and other applications (Table 1). In recent years, the main drivers of the growth in the IC market have remained largely consistent, although we have monitored gradual changes over time. For example, while demand for once-hot mobile products like smartphones and tablets appears to have leveled off (and even declined, in the case of tablets), other product applications are driving semiconductor manufacturers in new directions. The automotive industry has seen an uptick in demand as auto manufacturers have begun to look at futuristic features such as autonomous vehicles and sophisticated safety and comfort features. In the communications sector, networking providers have finally begun to expand into 5G networks, which may eventually lead to a resurgence not only for the communications infrastructure, but for a new generation of smartphone and other mobile devices. And further out, the Internet of Things (IoT), artificial intelligence (AI), and machine learning (ML) are the markets that will drive future growth in the semiconductor industry.
All of these macro trends pre-date the current crisis and in many ways will reassert themselves once the industry returns to “normal.” Meanwhile, we have observed some striking consequences of the virus-caused recession. The computer and peripherals application category, while an obviously dominant application for ICs, and therefore IC packaging, has not been particularly dynamic in recent years, as, say, communications have been. Yet, with the sharp increase in stay-at-home workers and students, there has been a small, if temporary, increase in this segment (we estimate close to 2% in 2020), mostly at the expense of automotive and consumer markets, both of which have seen a drop in sales due to the abrupt halt to economic activity.
The long-term trend remains, however, and computer applications are gradually losing market share, primarily to communications and industrial/other applications (Figure 3). Nevertheless, the impact is measurable and will endure for several years to come. We expect the market share of computers to be about a half percent higher in 2023, than we had forecasted in our pre-COVID-19 forecasts.
Considering all the “adjustments” NVR (and other organizations) have made to estimates for the industry in the past six months, can it be said that our initial forecasts for 2020 were wrong? Well, yes. In the hard glare of reality, that fact cannot be denied. Moreover, looking at the record of high-tech market research, market forecasts are often wrong. But here’s the thing: despite the presentation of detailed tables and figures showing multiple years of growth for any given market, in the final analysis, they are merely estimates based on known historical trends coupled with assumptions about future supply and demand, technological progress, and a myriad other factors that go into selling products and services. They are, in effect, educated guesses that (and here’s the important thing) tell a story about how the product will grow over time. And when an invisible foe suddenly appears, or looking farther back in time, when the bottom suddenly falls out of the housing market, or when terrorists strike at the heart of our financial industry, when any of these unexpected calamities occurs, it throws a large wrench into the workings of our carefully thought out forecasts.
So, if forecasts are so easily upended, does that mean that market research is not useful? As I just pointed out, the point of market research forecasts is to tell a story. Events big or small may well change individual forecasts, but in the long term, what is most valuable are the trends and long-range impacts that are revealed in those forecasts—in other words, the narrative that emerges from the data we present. Product managers and executives rely (or should do so), not merely on whether the IC packaging market will grow by 4.7% or 2.5% in 2020. Rather, what is more crucial to long-term success is understanding how the various market segments interact, thereby allowing us to see both the forest and the trees.
Jerry Watkins is Senior Research Analyst at New Venture Research, Nevada City, CA. He has more than 20 years of experience as both writer and manager at leading market research and consulting firms such as Frost & Sullivan, Lucid Information Services, and NSI Research. He has authored numerous syndicated reports on a variety of technologies, including telecommunications, office automation, merchant embedded computing and more recently in semiconductor packaging markets. He holds a BA in History and an MA in International Relations and International Management. Email firstname.lastname@example.org
SOURCE: Chip Scale Review