Download this article in PDF format.
What industry wouldn’t love for a huge and eager customer segment with deep pockets to emerge seemingly overnight, ready to buy up all of its products, drive innovation of new options and soak up any extra stock that may have been lying around waiting for a buyer? It sounds like a dream come true on the surface, but as most leaders know, this level of demand also comes with some downsides.
Consider the great toilet paper shortage that happened during COVID and left large CPGs scrambling to refill their supply chains, or the Stanley Cup TikTok craze that left retailers short on stock. We’re now seeing a similar thing happening in the semiconductor industry, but with even higher stakes and a considerably longer timeline. With more AI being developed, more data centers coming out of the ground and more of the technology being folded into everything from electronics to software to our kitchen refrigerators, the race is on to produce enough chips to satisfy the growing demand.
The problem is that there are only so many chips to go around, and the infrastructure behind them takes years to plan, build and get up and running. Add in the fact that the industry was already suffering a chip shortage even before generative AI made its grand debut and started being used by everyone, and it’s easy to see why the constraints and even the prospects are keeping chipmakers and procurement teams up at night.
Before the AI Boom
Let's rewind the clock a moment, shall we? The semiconductor supply chain was already struggling with capacity constraints as far back as 2019. Those problems accelerated and turned into a full-blown crisis the following year. “COVID-19 struck,” Fusion Worldwide’s Brittany McCabe writes in Supply Chain Connect.
“Manufacturers in all industries scrambled to meet supply and demand in an unprecedented economic climate,” she continues. “For chipmakers, production for almost half of the year was halted due to government shutdowns and COVID-19 restrictions. By the time production did resume, semiconductor companies had to adjust to increased pressure from various sectors.”
Today, AI fuels much of the demand for advanced chips and memory. According to McKinsey, demand for dynamic random-access memory (DRAM) and high-bandwidth memory (HBM) has grown much faster than manufacturers can produce them. Since early 2023, DRAM prices have increased roughly sixfold, reversing 15 years of price declines and returning average selling prices to 2011 levels.
Different types of chips have come in and out of short supply over the years, but the overall picture is still constrained for some of the most sought-after components. The AI boom has added another layer of demand, and now companies like EMD Electronics are raising concerns about the industry’s ability to keep up.
In a September 2026 survey of 200 semiconductor industry respondents, EMD found that the biggest obstacles to expanding manufacturing capacity go beyond building new plants. Finding enough qualified people is a major concern, with 75% of respondents identifying skilled workforce availability as the biggest barrier to accelerating domestic investment.
Citing Gartner’s numbers, EMD says worldwide semiconductor revenue will reach $1.6 trillion in 2026, a 92% increase over 2025. That’s exciting news for any company that builds, distributes or sells these components. It’s also a double-edged sword, particularly for buyers that face some serious competition as they scour the market for the parts and components they need.
No Quick Fixes
McKinsey sees few viable short-term solutions to the memory shortage. “While companies can reengineer products to decrease DRAM/NAND requirements, this process is time consuming and does not provide immediate relief,” it explains. “Other alternatives include delaying production or moving to a new technology generation.”
Some of the longer-term solutions that companies are using now include:
- Investing in new fabs to increase DRAM and HBM manufacturing capacity.
- Exploring advanced packaging solutions (for instance, taller HBM stacks and 2.5D or 3D integration to increase memory density and bandwidth).
- Increasing the bit density per wafer for DRAM through next-generation technology.
“If these solutions prove successful, memory supply could begin to catch up with demand around 2027 and prices could begin to decline, following the cyclical pattern typically seen within the semiconductor industry,” McKinsey adds. “Even then, memory ASPs [average selling prices] could remain higher than historical patterns would suggest.”
