The 2021-2022 automotive chip shortage cost the car industry an estimated $200 billion in lost production. Factories sat idle. New car lots were empty for the first time in decades. Used car prices spiked to absurd levels. Rental car companies were selling their fleets because they could not source vehicles to replace them.
That crisis was caused by a COVID-19 demand disruption that nobody planned for. The next one is different. It is structural, it is deliberate, and it is already beginning.
According to The Drive, AI's appetite for semiconductors is triggering a new automotive chip shortage. Unlike 2021, this is not a supply chain accident. This is chip fabrication capacity being consciously allocated to higher-margin AI customers at the expense of automotive suppliers.
Here is how the economics work. Semiconductor fabs — primarily TSMC, Samsung, and a few others — have finite production capacity. They can choose who gets that capacity. A hyperscaler building AI data centers buys chips in enormous volumes, on long-term contracts, at premium margins. An automotive supplier buying microcontrollers for engine management systems is a smaller, more fragmented customer with tighter margin expectations.
If you are a fab and demand from AI customers is exceeding your capacity, you do not expand to serve automotive. You charge the AI customers more and allocate your existing capacity to the most profitable buyers. Automotive customers get allocated less, wait longer, or pay more. All of those outcomes eventually show up as production delays or higher vehicle prices.
The complication is that automotive chips are not the same chips as AI chips. Modern vehicles use hundreds of different semiconductor components — microcontrollers for HVAC systems, infotainment processors, radar chips for driver assistance, power management ICs. These are not the H100 GPUs that Nvidia sells to data centers.
But they share fab capacity with less exotic chips that AI infrastructure also needs. Power management chips. Memory controllers. Networking ASICs. When those supply lines tighten, automotive feels it.
EV adoption adds a second layer to the problem. Electric vehicles use significantly more semiconductor content than internal combustion vehicles — roughly two to three times as many chips per vehicle. At the exact moment that EV adoption is accelerating, the supply environment for the chips EVs need most is tightening.
The automotive industry has been trying to improve its supply chain practices since 2021, building deeper relationships with fab partners and holding more inventory. But the fundamental supply constraint is the same: there is not enough advanced semiconductor manufacturing capacity in the world, and AI data centers are currently the most attractive customer for what does exist.
"Just build more fabs" is the instinctive policy response. It is correct in the long run. TSMC, Intel, and Samsung are all investing in capacity expansion. The CHIPS Act in the United States and similar legislation in Europe is funding new domestic capacity. But semiconductor fabs take three to five years to build and qualify. That is a 2028-2030 solution to a 2026 problem.
In the meantime, the next time you are waiting months for a specific vehicle model or paying more than expected for a new car, you will have a partial answer. The AI boom is not just an abstraction. It competes with your car for the same fundamental infrastructure, and right now, it is winning.
