Even the world’s best archer has trouble hitting the bullseye if the target won’t stand still.

That must be what it feels like to be a manufacturer of high-tech products today, struggling to accurately forecast demand for microprocessors.

Just one year ago, the supply picture for microchips across the board was dire. Automotive manufacturers, however, were especially feeling the pain, having cut back on commitments to suppliers when consumer sales cratered, then being unable to re-acquire the necessary supply when demand unexpectedly recovered.

The crisis is far from over, but the supply picture today is mixed. Coming into 2023, chip demand was expected to decline by as much as 30% from its high in the second quarter of 2022 (although that’s still three to four times higher than pre-pandemic levels). And in some industrial sectors, that scenario pretty much played out as predicted. Medical and industrial industries experienced a drop of 20% to 30%, down from “boom-time respectable numbers,” says Frank Cavallaro, chief executive officer of A2 Global, a distributor of electronic components. And sales of consumer electronics, having soared during the lockdown period of 2020-2021, plunged when people were freed from their homes and allowed to congregate in public once more.

Things are far from back to “normal,” however. Order lead times for chips today range from 26 to 52 weeks, versus 10 to 16 weeks prior to the advent of supply constraints, Cavallaro says, adding that manufacturers have simply had to adjust to the delays.

Demand remains high in the automotive industry, which has nevertheless “done an excellent job adjusting to the new supply chain paradigm,” he says. Manufacturers are allocating resources to the most profitable, high-end vehicles. And wherever possible, they’ve sought to reduce their dependence on microchips, such as Toyota shipping some new models without key fobs.

Automakers “have gotten really smart really quick,” Cavallaro says. Still, wait periods for some of the most popular models are lengthening, with some orders being placed today for cars that won’t be delivered until 2025 or even 2026.

The growing popularity of electric vehicles, which require far more microprocessors than those powered by internal combustion engines, is placing an additional strain on supply. That trend, coupled with higher prices for new cars, could serve to dampen demand in the months to come, Cavallaro says.

Certain industries that require more sophisticated chips, such as mobile phones, internet of things (IoT) devices, and defense and aerospace, are also expected to keep generating strong demand through the rest of this year. Cell phone makers need the chips to complete the rollout of 5G connectivity. And while the appetite for IoT devices on the consumer side isn’t especially strong at the moment — most buyers have yet to be seduced by the prospect of a “smart” refrigerator, Cavallaro says — demand in the industrial sector continues apace. Popular applications include systems for tracking freight and transportation equipment across multiple modes. Fortunately, the slackening of demand on the consumer side has allowed for some supply to be reallocated for industrial use.

Defense and aerospace manufacturers have also been caught short by unexpected demand for microchips, Cavallaro says. Forecasts made three years ago didn’t account for Russia’s invasion of Ukraine, and the resulting surge in military spending. “That’s going to be hard to source,” he says.

The world’s major chip producers are scrambling to accommodate the increase in demand, at the same time that original equipment manufacturers are looking to diversify global sourcing in order to reduce the risk of supply disruption. In the U.S., the recently enacted CHIPS and Science Act promises to pump $52 billion into the construction of fabrication plants, or fabs, in the U.S. But such plans require years and huge investments to realize, and will do little to ease the short-term pain that key high-tech industries continue to suffer.

Getting the arrow to line up with the target will never be easy. Cavallaro says cycles of microchip supply and demand repeat every four to five years, although the highs and lows could become even more extreme as the need for sophisticated new technology, such as devices employing artificial technology, explodes.

For the foreseeable future, “that’s going to prevent the balance that we’re all looking for,” Cavallaro says, “unless we call 26- to 52-week lead times for parts acceptable.”