Soaring Memory Chip Prices Drive Up Auto Costs, Shifting Profits Across Supply Chain

In July 2026, China's automotive industry exhibited stark divergence: memory chipmakers reported explosive profit growth, while numerous automakers plunged into losses. Longsys (Jiangbolong) projected its first-half 2026 net profit attributable to shareholders at RMB 9.2–11 billion, surging over 740-fold year-over-year. GigaDevice forecasted a net profit of approximately RMB 6.9 billion for the same period, up nearly 11-fold YoY. Companies such as Shannon Core, ChangXin Memory Technologies, and Hongxin Yu also recorded profit increases ranging from several-fold to dozens of times. Meanwhile, Seres anticipated a net loss of RMB 1.5–1.8 billion for the first half of 2026, GAC Group projected a loss of RMB 4.06–4.57 billion, and JAC Motors expected a loss of around RMB 740 million. This divergence stems from a structural shortage of memory chips driven by surging AI computing demand. The sharp rise in demand for high-bandwidth memory (HBM) and premium DDR5 from AI servers has prompted leading manufacturers—including Samsung, SK Hynix, and Micron—to redirect 70%–80% of their advanced-node capacity toward AI-related products, squeezing out capacity for commodity DRAM and NAND. The automotive sector accounts for only about 3% of the global DRAM market, placing it at a disadvantage in capacity allocation; even with price premiums, automakers struggle to secure sufficient automotive-grade chips. In Q2 2026, contract prices for conventional DRAM rose 58%–63% quarter-over-quarter, while NAND Flash prices climbed 70%–75%. According to UBS data, global monthly memory chip sales reached a record-high USD 74.6 billion in June. The industry’s total revenue is projected to hit USD 992 billion in 2026 and could surge to USD 1.76 trillion in 2027. Automakers are facing significantly heightened cost pressures. Seres reported that the unit price of memory chips soared from RMB 20 to nearly RMB 100, while lithium carbonate prices jumped from RMB 80,000 to RMB 180,000 per ton, adding RMB 15,000–20,000 to the cost per Aito vehicle. Brands including BYD, XPeng, and Jetour have already raised prices for optional intelligent driving packages. NIO stated that raw material cost increases alone added over RMB 10,000 per vehicle. To mitigate these cost pressures, some automakers are accelerating the adoption of integrated cabin-and-driving-domain architectures, which reduce DRAM usage through shared memory. Companies such as Zhuoyu, Horizon Robotics, Black Sesame Smart, and BYD have already launched relevant chips or mass-production solutions, expected to save RMB 2,000–3,000 per vehicle in memory costs. Market expectations point to continued near-term price increases for memory chips. ADATA Chairman Simon Chen predicted that Q3 2026 DRAM contract prices would rise another 20%–30%, with NAND Flash increasing 35%–40%. UBS forecasts that DRAM supply-demand tightness will persist through the first half of 2028, though TrendForce anticipates that price hikes will gradually moderate. A new variable is emerging: humanoid robots are being viewed by Micron as the next-generation engine for memory demand, with estimated storage capacity ten times that of L2+ autonomous vehicles. If this segment achieves scale deployment around 2030, it could further extend the current high-growth cycle for memory chips. Analysts note that the current shortage of automotive memory chips is not a short-term cyclical fluctuation but a structural mismatch resulting from overlapping demand surges from both AI and intelligent vehicles. Automakers must shift from reactive measures to long-term supply chain collaboration and technological restructuring to enhance cost control over critical components.