AI data centers' voracious demand for DRAM and NAND is crowding out automotive supply, forcing carmakers to absorb massive cost increases. General Motors warned that it expects costs to rise by $1.5 to $2 billion in 2026, with CFO Paul Jacobson citing sharply higher DRAM prices as the primary driver. Chinese EV maker BYD has already acted, raising prices on its driver-assistance packages by 20% to offset the memory cost surge.
The root cause is structural reallocation of semiconductor capacity: Samsung, SK Hynix, and Micron are prioritizing high-bandwidth memory (HBM) for AI servers, where margins are 3–5× higher than automotive DRAM. DRAM spot prices jumped 450% between September 2025 and January 2026, with Kearney estimating further 60–75% increases in the second half of 2026. Modern EVs already require 278 GB of memory on average (up from 90 GB in 2023); Level 4 autonomous vehicles need 300+ GB of DRAM alone, all subject to 2-year AEC-Q100 certification that prevents quick substitution.
For supply-chain operators and auto-tech teams, this is a structural, not cyclical, shortage. New DRAM fabs won't add meaningful capacity until 2027+, while EV electrification and autonomous features continue to push memory-per-vehicle higher. Workarounds—downgrading autonomy levels, consumer-grade memory in non-safety systems, long-term fixed-price contracts—are already underway. Expect feature deletions and price increases across the industry through 2027 unless AI infrastructure spending suddenly contracts.