Summary of Key Points
The prices of memory chips, especially those for automotive use, have skyrocketed, becoming a “fatal black hole” that squeezes the profits of car manufacturers. Seres reported a pre-loss of 1.5 to 1.8 billion yuan for the first half of the year; Li Bin from NIO has expressed frustration; and the Ideal L6 model had to adjust its price due to increased memory chip costs. Almost all new energy vehicle companies have been affected. The root cause of the price surge is the AI industry’s demand for memory chips, which has diverted production capacity from other sectors. The three major manufacturers (Samsung, SK Hynix, and Micron) have shifted most of their production to higher-profit HBM chips used in AI applications, leaving automotive-grade memory with a smaller market share and lower profits. Car manufacturers already have thin margins (with an average profit margin of only 1.5% in the first half of 2026), and the increase in memory chip prices further reduces their profitability. There are three possible ways for car companies to cope: developing their own chips (though the cost is high), securing supply from suppliers (the effectiveness is uncertain), or relying on domestic alternatives (which is the most realistic option). In the long run, achieving self-sufficiency in the supply chain is crucial.
1. How Crazy Are the Memory Chip Price Increases? An Additional $3,000 per Car
The increase in memory chip prices is not a minor fluctuation but a “tsunami-level” phenomenon:
- Amazing Growth: Since September 2025, the price of automotive-grade DDR5 chips has increased by six times (as of June 2026), and DDR4 prices have risen by 150% overall. From March to June 2026, automotive-grade memory prices rose another 180%, with some high-end DDR5 chips increasing by more than 300%.
- Direct Impact on Car Manufacturers: A smart car requires 4 to 16 DRAM chips and 2 to 6 NAND chips for storage, resulting in an additional cost of $500 to $3,000 per vehicle. For car manufacturers with already thin margins, this means losing money at every turn. A technical expert from a car company stated that they cannot afford to lower prices on new models and can only offer more features instead.
- Impact on the Entire Industry: Companies like Seres are suffering losses, NIO is under pressure, and even the Ideal L6 has had to raise its price. Almost no new energy vehicle company has been spared, with Huawei also raising the prices of its products due to increased memory costs. Car manufacturers have become the next victims of these price hikes.
2. Why the Price Increases? AI Is Competing for Memory Chips
The current price surge is not simply a matter of supply and demand imbalance but an “siphoning effect” by the AI industry:
- Manufacturers Shifting to Higher-Profit Chips: More than 90% of global memory chips are monopolized by Samsung, SK Hynix, and Micron, which have redirected 70%-80% of their advanced production capacity to HBM chips used in AI servers. Why? Because HBM chips are much more profitable: one HBM4 chip sells for $560, with a gross margin of 75%, resulting in a profit of $420, compared to only $46.20 and a 18.5% margin for automotive-grade DDR4 chips.
- Lack of Influence from Car Manufacturers: The automotive industry accounts for less than 10% of the memory chip market, and its costs account for less than 1% of the total vehicle cost, making it difficult to influence suppliers to adjust production. SK Hynix has clearly stated that it cannot adjust its HBM and standard DRAM production lines in 2026, while Samsung plans to triple HBM production. Car manufacturers can only watch helplessly as production capacity is taken over by the AI industry.
- Longer Development Times: Automotive-grade chips require rigorous AEC-Q100 certification (to function in extreme temperatures and with various environmental conditions), taking 2 to 3 years to develop, while AI chips can be developed in just one year. Naturally, suppliers prioritize producing the more profitable products.
3. Car Manufacturers’ Thin Margins: The Final Strain
Car manufacturers are already struggling, and the increase in memory chip prices is a major setback:
- Recordly Low Profit Margins: In the first half of 2026, the average profit margin for Chinese vehicle manufacturing was only 1.5%, far below the national industrial average of 6.1%.
- Marginal Profit per Vehicle: From January to May 2026, Chinese car manufacturers had revenue of 343,000 yuan per vehicle, but costs reached 305,000 yuan, resulting in a profit of only 38,000 yuan per vehicle. A 3,000 yuan increase in memory chip prices equates to nearly a 10% loss on each vehicle sold—losing money on every sale is no exaggeration.
- Worse Than Lithium Mine Price Increases: While previous lithium mine price increases were due to supply and demand imbalances that resolved after capacity expansion, the current situation with memory chips is caused by long-term AI demand, with no immediate solution in sight.
4. Three Possible Ways for Car Manufacturers to Cope: Which One Is Feasible?
There are three approaches, but their feasibility varies greatly:
1. Developing Own Chips: Elon Musk plans to build a TeraFab semiconductor factory with an investment of $119 billion, but this requires massive funds that most domestic car companies cannot afford (the first phase alone would cost 55 billion yuan, equivalent to several years of their net profits).
2. Securing Supplier Capacity: Companies like Ideal and Xpeng have signed direct supply agreements with Micron to ensure stable prices and joint chip development. However, the details of these agreements are not publicly available, so it’s unclear how much capacity they can secure.
3. Relying on Domestic Alternatives: ChangXin Memory (China’s only company capable of mass-producing DRAM) has developed LPDDR5 chips that meet automotive standards and are used in Tesla and BYD vehicles. Additional chip production facilities in Hefei and Beijing are operating at full capacity, with another plant in Shanghai set to start production in 2027. Although domestic production currently accounts for less than 5%, it can gradually reduce reliance on foreign manufacturers.
5. The Long Run: The Battle for Chips Between AI and Automotive Intelligence Just Begins
Cars are becoming “data centers on wheels,” with L2-level autonomous driving requiring 8-12GB of memory and higher-end models needing 64-256GB. As the AI industry continues to grow, demand for memory chips will only increase, making this battle for chips a long-term issue.
For car manufacturers, the short term solution is to secure supply through contracts, while the long-term goal must be to achieve self-sufficiency in the supply chain. Those that can reduce their dependence on foreign components will have a better chance of surviving this cost crisis and thriving in the competition for automotive intelligence.
This news report highlights the challenges faced by the automotive industry in the AI era: as cars become more intelligent, limited chip production capacity is being directed towards higher-profit applications. To avoid losing profits, car manufacturers must shift from passive acceptance to proactive planning, with domestic alternatives potentially being the key to breaking this deadlock.