The broad chip shortage that crippled industries from 2020 to 2023 is largely over, but pockets of tightness remain, and a clean “back to normal” date doesn’t exist. Some chip categories are now in oversupply, with warehouses sitting on months of unsold inventory. Others, particularly automotive-grade components and AI-related processors, still face constrained supply heading into late 2025. The answer depends entirely on which chips you’re asking about.
Where the Shortage Has Eased
For most consumer electronics, the worst is behind us. The pandemic-era scramble for basic logic chips, memory modules, and processor components has cooled significantly. Inventory levels across major chipmakers tell the story clearly: AMD is sitting on roughly 146 days of inventory, Texas Instruments on 222 days, and NXP on 148 days. For context, a healthy semiconductor company typically holds 60 to 90 days of inventory. These elevated numbers signal oversupply, not scarcity, for many standard chip types.
Memory chip makers like Micron and SK Hynix also hold well over 120 days of inventory each. If you’re buying a laptop, a smartphone, or a gaming console in 2025, chip availability is not the bottleneck it was in 2021 and 2022.
Automotive Chips Are Recovering, Slowly
The auto industry was hit hardest by the original shortage and has been slowest to recover. At the peak in 2021 and 2022, automakers waited over 50 weeks for microcontrollers and 40-plus weeks for power semiconductors like the transistors used in electric vehicle drivetrains. Those wait times have come down considerably. By mid-2024, automotive microcontroller lead times had dropped to roughly 20 to 30 weeks. By early-to-mid 2025, some standard 8-bit and 32-bit microcontrollers from major suppliers like Infineon and NXP were quoted at 10 to 26 weeks.
That’s a major improvement, but it’s still not pre-pandemic normal, when lead times for most automotive chips ran 8 to 14 weeks. The auto sector uses older “mature node” chips, typically manufactured at 28 to 90 nanometers, and fewer factories produce them because the profit margins are lower. A 2024 report from the U.S. Bureau of Industry and Security found that for nearly a third of the chips U.S. companies had fabricated at Chinese foundries, no alternative factories were available to make them. That kind of concentration makes the supply chain fragile even when overall capacity looks adequate.
AI Demand Is Creating a New Kind of Squeeze
Just as the original shortage faded, a new pressure emerged. The explosion in AI infrastructure spending is absorbing enormous amounts of manufacturing capacity for advanced chips. Nvidia’s CFO warned in 2025 that boosting shipments of consumer graphics cards would be difficult “for multiple quarters” because data center operators are buying GPU compute and memory at scale, often willing to pay premiums to lock in supply.
This doesn’t just affect finished GPUs. It pulls on upstream components and manufacturing capacity that consumer products also rely on: advanced packaging, high-bandwidth memory, and cutting-edge fabrication slots. In a supply-constrained environment, the highest-value demand wins the allocation. AI chips generate far more revenue per wafer than a graphics card for a gaming PC, so foundries and suppliers prioritize accordingly.
The result is a two-track semiconductor world. Commodity chips are plentiful. The most advanced chips, anything tied to AI training and inference, remain tight and will likely stay that way through at least 2026 as data center buildouts continue accelerating.
Older Chip Factories Remain a Weak Spot
Most of the global investment pouring into new fabrication plants is aimed at leading-edge manufacturing, chips at 5 nanometers and below. That’s where the margins are, and that’s where government subsidies are pointed. Under the CHIPS Act, the United States is projected to produce at least 20 percent of the world’s leading-edge logic chips by 2030, up from zero percent in 2022.
But the chips inside your car’s airbag controller, your washing machine, or a hospital ventilator are built on older technology nodes. Investment in expanding capacity for these mature chips has been comparatively modest. The pandemic exposed how fragile that supply chain is: prices spiked across automobiles, consumer appliances, and medical devices because a handful of factories, many in China, were the only ones producing certain chip types. That structural vulnerability hasn’t been fully resolved. New mature-node capacity is being added, particularly in China, but it takes years for a fabrication plant to go from groundbreaking to volume production.
Raw Materials Add Another Layer of Risk
Chip manufacturing depends on a surprisingly narrow set of specialized materials, and some of them have their own supply problems. Helium, used in cooling and quality testing throughout the fabrication process, is classified as a critical mineral by the EU and Canada. It’s a finite resource with production concentrated in just a few countries, notably Qatar and Russia. Geopolitical tensions can disrupt supply at any time, and historically low helium prices have discouraged investment in new extraction or recycling technologies.
As chip manufacturing moves to smaller, more complex nodes to meet AI demand, the semiconductor industry’s reliance on helium is actually increasing, with no viable alternatives available. Neon gas, another essential input for the lasers used in chip lithography, faced its own crisis when Russia’s invasion of Ukraine disrupted supplies from major Ukrainian producers. New sources have come online since then, but the episode illustrated how a single geopolitical event can ripple through the entire chip supply chain.
The Taiwan Factor
Any honest assessment of semiconductor supply risk has to account for Taiwan. TSMC, headquartered there, manufactures the vast majority of the world’s most advanced chips. Researchers modeling Chinese aggression scenarios have identified a quarantine of Taiwan, selectively restricting shipments of raw materials, chemicals, and natural gas, as the most likely near-term threat, with particular vulnerability before 2027.
Even a partial disruption to Taiwan’s chip output would trigger shortages that would dwarf the 2020-2023 crisis. This is a major reason governments are spending billions to build domestic chip fabrication capacity, but those plants won’t reach meaningful production volumes until the late 2020s at the earliest. In the meantime, the global chip supply remains heavily dependent on a single island in a geopolitically volatile region.
What This Means Practically
If you’re waiting for chip shortages to stop affecting the price or availability of a specific product, the timeline depends on the category. Consumer electronics like phones, laptops, and TVs are already back to normal availability and pricing. New cars are widely available again, though some specialty models with advanced driver-assistance features may still see occasional delays into late 2025. High-end GPUs for gaming or creative work will remain tight through at least mid-2026 as AI demand continues to dominate allocation.
The broader semiconductor industry is unlikely to return to the comfortably oversupplied state that existed before 2020. Demand is growing faster than capacity in the most advanced segments, geopolitical risks are higher than they’ve been in decades, and critical raw materials face their own constraints. The era of assuming chips would always be cheap and available is over. What’s replaced it isn’t a permanent shortage, but a permanently tighter market where disruptions hit harder and take longer to resolve.