In the first half of 2026, memory chip manufacturers and automakers are experiencing extreme profit divergence. Shenzhen Longsys Electronics saw net profit surge over 740-fold year-on-year, while GigaDevice's quarterly earnings exceeded three times its entire previous year total. Meanwhile, SERES, GAC, and JAC have issued profit warnings. The explosive demand for high-bandwidth memory from AI servers is squeezing the automotive industry's memory chip supply space.
Data Highlights: Extreme Profit Divergence Between Chips and Automakers
In July 2026, China's automotive industry is witnessing a dramatic "fire and ice" profit polarization.
Staggering performance from memory chip makers:
- Shenzhen Longsys Electronics: First-half net profit surged over 740-fold year-on-year
- GigaDevice: Single-quarter earnings exceeded three times its entire previous year's total
Profit warnings from automakers:
- SERES: Cited in its performance report that "sharply rising memory chip prices pushing up production costs" was a primary driver of losses
- GAC: Issued a profit warning
- JAC: Similarly facing profitability pressure
Behind this extreme divergence lies a massive supply chain reshuffling triggered by AI computing demand.
AI Era Ignites Memory Chip "Super Fuse"
The explosive demand for High Bandwidth Memory (HBM) from AI training servers is the core driver of this cycle.
According to CCTV Business reports, a single top-tier AI training server packs over 100GB of HBM — dozens of times the DRAM used in a standard server. To meet this explosive demand, Samsung, SK Hynix, and Micron have shifted 70% to 80% of their advanced process capacity to HBM and premium server DDR5 memory.
The direct consequences of this capacity shift:
- General-purpose DRAM and NAND capacity is systematically squeezed
- Supply-demand mismatch ignites a price spike
- The automotive industry — claiming only about 3% of the global DRAM market — has been pushed to the back of the supply queue
| Memory Type | AI Server Usage | Standard Server Usage | Multiplier Gap |
|---|---|---|---|
| HBM | Over 100GB | None | Dozens of times |
| DRAM | Dozens of times standard | Standard config | Dozens of times |
| Advanced capacity allocation | 70%-80% | 20%-30% | 2-3x |
Automotive Supply Chain: Forced to Absorb "Spillover Costs"
Against the backdrop of AI server orders swallowing advanced capacity like a black hole, automotive buyers face severe challenges:
- Supply priority decline: The automotive industry holds only about 3% of the global DRAM market, placing it at a disadvantage in capacity competition
- Procurement costs soaring: Sharply rising memory chip prices directly push up automakers' production costs
- Automotive-grade chips even tighter: Automotive memory chips require higher reliability and temperature ranges, making capacity shifts more difficult
SERES's wording in its performance report is telling — listing rising memory chip prices as one of the primary drivers of losses. This reveals a harsh reality: when the AI industry and automotive industry compete in the same semiconductor pool, the smaller, lower-margin automotive side often absorbs the "spillover costs."
Industry Insights: A New Test of Supply Chain Resilience
This divergence raises new considerations for the global automotive supply chain, especially for Central Asian and Russian markets dependent on Chinese supply chains:
- Supply chain diversification: Single-source dependency amplifies risk during capacity shortages
- Long-term agreement locking: Signing long-term procurement agreements with chip suppliers may become key to stable supply
- Domestic substitution acceleration: The rise of Chinese chip design companies (such as GigaDevice) provides more options for automotive supply chains
The supply-demand landscape for memory chips has been fundamentally reshaped in the AI era. For automotive industry professionals, understanding this structural shift and adjusting procurement strategies has become a survival necessity. For more in-depth industry analysis, visit EX1000.COM.
