AI Demand Echoes the Late-’90s Tech Surge

Memory chip shortages and PCB order spikes signal strength, but history suggests caution amid the hype.

It appears that possibly, as baseball legend Yogi Berra would have said, it’s déjà vu all over again.

Reuters reports that the AI (artificial intelligence) frenzy is creating a major shortage of memory chips, used in everything from consumer electronics and industrial equipment to computers and even the most advanced military systems. Concurrently, the Global Electronics Association (IPC) reported that printed circuit board bookings and shipments surged during October. While both announcements suggest that the electronics industry is alive and well, could they also be a forewarning that we are heading toward the same euphoric economic environment that occurred in the late 1990s?

For the electronics industry, the late 1990s was the best of times, until they weren’t. PC power was growing by leaps and bounds, creating constant demand for the latest iteration. The internet was in its infancy, creating an insatiable demand for servers and all the infrastructure required to harness emerging technology and enable users to take full advantage of the nascent web. The dot-com frenzy was upon us. Add to that the fear that Y2K, the millennial bug, would render older computers and analog technologies inoperable, which prompted further demand for new, compliant equipment. Demand for all things electronic, and the printed circuit boards and assemblies that the new technology was built on, was skyrocketing, and company revenues and profits were soaring.

Regrettably, the euphoric ’90s came to a stunning halt in 2001. Companies with year-long order books had the rude awakening first that demand was flattening, and then decreasing. Inventories spiked. As a white-hot industry cooled, overcommitted companies began to cancel orders. For the electronics industry, especially the circuit board industry, the bottom fell out and many, many businesses foundered. The best of times became the worst of times.

Many will say, that was then, but this time is different. True, we do not have to contend with the uncertainty of a new millennium; we are now a quarter of the way through it. What is similar – if not identical – however, is the economic frenzy an emerging technology is creating, coupled with the resultant demand for electronics. As with the dot-com environment of the late ’90s, euphoria over AI now drives the electronics industry. Every business “must” either make or use AI – even if they are not sure how. Consumers are equally enchanted with the likes of ChatGPT or Google AI Overview, which are being played with and deployed for a host of trivial and major activities.

As with many exciting new technologies, a mixture of science, wishful thinking and a lot of hype have swirled together in the story. Since the dot-com era, other “must-have” technologies have been championed. Electric vehicles are a recent example that has proven exciting and technologically successful, but perhaps disappointing when it comes to worldwide demand. And while EVs are here to stay, many auto companies are scaling back sales projections, production, new product development and, most importantly, are writing off huge investments that will not meet prior forecasts.

The question is, will demand for all things AI come to the same abrupt demise as the dot-com bubble, or just fizzle, as is occurring with EVs? More importantly, will today’s electronics companies take what was learned back then and apply those lessons to avoid potential future calamity?

Short term, the future of AI may serve our industry well in creating demand, filling plants and providing the profits to continue investing in our businesses. Clearly, serious money is being invested at all levels of the AI supply chain. And with any new technology, the goal should be to participate at some level and to take advantage of the opportunity at hand. The challenge is not losing objectivity amid all the hype.

When dealing with anything new, a jaundiced eye can be helpful. Upward projections of demand are fine. Unrealistic or euphoric projections should be questioned and challenged, however. When planning capacity utilization or considering expanding capacity, also consider the business response if projections do not materialize. What if the exit ramp is straight down? Possibly most importantly, diversify your customer base, so all your business eggs are not in a single end-market basket. 

AI is the next exciting technology. Just be aware that our industry has been there before many times and, in the past, when there has been excessive hype, the end game has not necessarily gone well. As we all navigate the excitement, use lessons from the past to reduce some of the inevitable risk.

Peter Bigelow has more than 30 years’ experience as a PCB executive, most recently as president of FTG Circuits Haverhill; peterbigelow@msn.com. He is vice chair of the PCEA PCB Management Symposium, taking place April 28 at PCB East.