Morgan Stanley issued a stark warning on January 22, 2026, flagging dynamic random-access memory (DRAM) shortages as a looming operational risk for global automakers that could dwarf previous semiconductor disruptions in both duration and structural impact. The note, led by European autos analyst Javier Martinez de Olcoz Cerdan alongside the firm's global semiconductor team, arrives as spot DRAM and NAND prices have surged triple-digit percentages since mid-2025, threatening to add US$100–400 per vehicle in incremental costs while forcing potential platform redesigns by 2028. Chinese electric vehicle manufacturers, heavily reliant on memory-intensive smart cockpits and autonomous driving systems, appear disproportionately exposed to the crunch.

AI Reallocation Triggers Supply Crunch

The root cause is familiar to anyone tracking the semiconductor space: artificial intelligence. Morgan Stanley's cross-sector team identifies AI-driven capacity reallocation toward data centers as the primary culprit behind tightening automotive memory supply. Unlike the 2021 chip shortage centered on analog components and microcontrollers, this cycle targets DRAM and NAND flash—components concentrated in the most compute-intensive vehicle systems including digital cockpits, advanced driver-assistance systems (ADAS), and over-the-air update infrastructure.

The bank estimates total memory silicon content per vehicle currently ranges from US$60–100 for internal combustion engine vehicles and US$150–180+ for battery electric vehicles, representing roughly 0.5% of total vehicle cost of goods sold. However, with DRAM and NAND spot prices more than tripling since mid-2025, incremental cost headwinds could reach US$100–200 per ICE vehicle and US$300–400 per BEV on current spot assumptions, posing meaningful downside risk to already compressed automotive margins.

Chinese EV Makers in the Crosshairs

Premium and software-intensive vehicles face disproportionate exposure, a category where Chinese EV manufacturers have aggressively staked their competitive positioning. BYD, NIO, XPeng, and Li Auto have differentiated their products through advanced infotainment systems and autonomous driving capabilities—precisely the features that consume the most DRAM. According to estimates from Micron Technology, the average vehicle sold in 2025 contained 16GB of DRAM and 204GB of NAND, representing roughly 3x more DRAM and 4x more NAND than vehicles sold in 2021.

Morgan Stanley notes that while this exposure creates near-term vulnerability, it could paradoxically accelerate strategic partnerships between Chinese automakers and global chipmakers, expediting semiconductor localization efforts that underpin China's long-term global automotive ambitions. The faster iteration cycles of Chinese EV platforms may also provide an advantage in migrating to newer LPDDR5 and GDDR6 memory standards, potentially turning a short-term headwind into a medium-term competitive edge.

No Quick Resolution in Sight

The bank's semiconductor analysts expect supply-demand gaps to widen for NOR flash, DDR4/DDR3 DRAM, and SLC/MLC NAND, with bottlenecks likely extending through year-end 2026 and potentially into 2027. Recent channel checks suggest DDR5 pricing will double year-over-year for automotive customers by 4Q26. NAND supply faces similar constraints as Samsung Electronics ramps down MLC NAND production with limited prospect of reversal into 1H27.

Automotive DRAM supply remains highly concentrated among three suppliers—Micron, Samsung, and SK hynix—all gradually phasing out legacy capacity, amplifying structural risk. While these manufacturers have signaled continued automotive support for several years, the uncertainty has already triggered precautionary buying by OEMs and Tier-1 suppliers, according to S&P Global data cited in the report.

2028 Redesign Risk Looms

Perhaps most concerning is the structural challenge beyond 2027. Morgan Stanley flags that suppliers indicate DDR4/LPDDR4 availability through end-2027, but beyond that timeframe, the challenge shifts from financial to structural. Legacy automotive DRAM availability is expected to decline sharply regardless of price, yet most vehicles planned for 2028 production—including the majority of top cockpit and ADAS designs—still rely on DDR4/LPDDR4 architecture.

This leaves the industry with roughly two years to orchestrate coordinated migration toward LPDDR5, requiring synchronized action across system-on-chip vendors, Tier-1 suppliers, and OEMs. Delays raise the risk of forced redesigns, launch slippage, or constrained production later in the decade—a scenario that would be particularly damaging for Chinese manufacturers whose product roadmaps emphasize continuous feature upgrades.

Uneven Cost Absorption Expected

Near-term visibility remains mixed. Volkswagen AG indicated current coverage under supply contracts, while Hyundai Motor Company and Kia Corporation stated they do not see significant semiconductor supply risk at this stage, reflecting improved inventory discipline following prior shortages. However, Morgan Stanley cautions that historical precedent suggests near-term impacts can deteriorate more quickly than anticipated.

The bank expects uneven cost absorption across the supply chain, with OEMs better positioned than Tier-1 suppliers given the imbalanced margin profile observed over the past 12 months. Realized impact will depend on inventory buffers, contract coverage, hedging strategies, and OEM pricing power—variables that differ significantly between established Western automakers and newer Chinese entrants.

For Chinese EV makers specifically, the memory shortage arrives at an inopportune moment as they push international expansion while navigating tariff uncertainties and intensifying competition. Combined with ongoing raw material and foreign exchange headwinds, persistent chip disruptions could force a more cautious tone on 2026 guidance, particularly if shortages extend beyond spot markets into contracted supply. The question is whether their architectural flexibility and closer proximity to Asian semiconductor supply chains can offset their higher memory content exposure—a test that will likely define competitive dynamics through the remainder of the decade.