Goldman Sachs Bullish on Changxin: Will Doubling Production in Four Years Cover Half of China's DRAM Demand?

TL;DR Goldman Sachs has given a "Buy" rating and a 12-month target price of 129 yuan, betting on a resonance of AI demand, capacity expansion, and domestic substitution. Goldman Sachs expects Changxin's monthly production capacity to increase from 270,000 wafers in 2026 to 665,000 wafers in 2030, more than doubling in four years. By 2028, Changxin's supply may cover about 50% of China's DRAM demand, but this still depends on capital expenditure, yield ramp-up, and customer validation. HBM is key to Changxin's profit upgrade, with Goldman Sachs estimating its revenue share will rise from 2% in 2026 to 27% in 2030. The target price of 129 yuan implies not only shipment growth but also sustained high DRAM prices, product structure upgrades, and an aggressive assumption of gross margins rising to 82% by 2030.

Changxin Technology, the leading DRAM manufacturer in China, received its first coverage from Goldman Sachs less than a month after its IPO.

On August 23, Goldman Sachs issued a "Buy" rating for Changxin in its report "CHIPS IV: Accelerating China's Semiconductor Self-Sufficiency Process," setting a 12-month target price of 129 yuan. At the time of the report, Changxin was valued at about 10 times the projected P/E ratio for 2027; Goldman Sachs' target price implies about 24 times the projected P/E ratio for 2027.

This pricing is backed by a growth model extending to 2030: wafer production capacity doubling, traditional DRAM shipments continuing to expand, and rapid growth in HBM revenue, while China's AI computing infrastructure and customer supply chain diversification jointly boost domestic storage demand.

However, this model is also based on a series of optimistic assumptions. In addition to capacity expansion and yield ramp-up, Goldman Sachs expects DRAM prices to remain high in a tight supply environment, with Changxin's gross margin rising from 41% in 2025 to 82% in 2030. Therefore, the target price of 129 yuan bets not only on domestic substitution but also on capacity, prices, and product structure developing favorably.

Changxin Technology went public on the STAR Market on July 27, with the stock code 688825. At the time of its IPO, the total A-share capital was approximately 66.881 billion shares, with about 4.503 billion shares starting to trade. The prospectus indicated that the raised funds would mainly be used for upgrading wafer manufacturing production lines, DRAM technology upgrades, and advanced technology research and development.

Goldman Sachs' new report further extrapolates these investments into a capacity leap lasting until 2030.

Capacity to Double in Four Years, Potentially Covering Half of China's DRAM Demand by 2028

Goldman Sachs expects Changxin's monthly wafer production capacity to increase from 270,000 in 2026 to 447,000 in 2028, reaching 665,000 in 2030, more than doubling from 2026.

The expansion is supported by continuous capital investment. Goldman Sachs estimates that Changxin's average annual capital expenditure from 2026 to 2030 will reach 84 billion yuan, higher than the approximately 50 billion yuan from 2022 to 2025; comparing only to 2024 and 2025, the average annual capital expenditure in the previous two years was about 60 billion yuan.

Driven by capacity expansion, yield improvement, and product specification upgrades, Goldman Sachs expects Changxin's total DRAM supply to grow at a compound annual growth rate of 34% from 2026 to 2030, reaching 9.837 billion GB by 2030.

By 2028, Changxin's traditional DRAM supply is expected to reach 41% and 50% of the corresponding supply from Samsung and SK Hynix, respectively, up from 28% and 35% in 2025.

A more striking prediction is that Goldman Sachs expects Changxin's supply to cover about 50% of China's DRAM demand by 2028. This is a forecast of future supply capability and does not imply that Changxin currently holds half of the market share in China.

Changxin's monthly production capacity expansion trend from 2026 to 2030, with a core forecast increasing from 270,000 to 665,000 wafers.

Comparison of China's DRAM supply and demand; Goldman Sachs expects Changxin to cover about 50% of domestic DRAM demand by 2028.

AI Boosts China's Storage Demand, Leaving Room for Changxin's Expansion

Goldman Sachs expects the scale of China's DRAM market to grow at a compound annual growth rate of 50% from 2026 to 2028, reaching 257 billion dollars by 2028. The growth is mainly driven by AI server shipments, server DRAM, and HBM demand, while smartphones, personal computers, networking equipment, and automobiles also constitute the fundamental demand.

The supply side is also undergoing changes. As global storage manufacturers like Samsung, SK Hynix, and Micron shift more resources towards HBM and other AI-related products, traditional DRAM supply is being squeezed. In an environment of rising prices and limited supply, consumer electronics manufacturers are also more motivated to introduce new suppliers to reduce single-source risks.

Goldman Sachs believes that even though Changxin's technology nodes still lag behind global leading manufacturers by several generations, American and other overseas customers may still validate its mobile DRAM and traditional DRAM products, especially in the smartphone and personal computer sectors.

This forms two main lines for Changxin's expansion: on one hand, meeting domestic DRAM demand and domestic substitution; on the other hand, filling the gap in traditional DRAM left by global manufacturers shifting capacity to HBM.

However, whether overseas customers can form large-scale purchases is still affected by geopolitical and trade restrictions. The willingness to validate products does not equate to guaranteed orders, which is also one of the main risks listed by Goldman Sachs.

HBM Determines Whether Changxin Can Transition from Scale Expansion to Profit Upgrade

Traditional DRAM provides the scale foundation, while HBM determines whether Changxin can truly capture the high-value increments of AI storage.

HBM provides higher bandwidth, capacity, and energy efficiency for AI accelerators through vertically stacked multiple layers of DRAM. However, compared to ordinary memory, HBM manufacturing involves multiple processes such as front-end DRAM chip production, high-precision silicon vias, advanced node logic wafers, thermal management, and reliability, with significantly higher technical and supply chain barriers.

Goldman Sachs expects Changxin's HBM products to start contributing revenue from the fourth quarter of 2026, with HBM revenue share rising from 2% in 2026 to 27% in 2030. Its HBM supply is expected to grow at a compound annual growth rate of 207% from 2026 to 2028, reaching 1.179 billion GB by 2028.

However, this is also the part of the entire model with the greatest uncertainty.

Goldman Sachs clearly points out that Changxin's HBM technology maturity is still in the early stages, and it is expected to be difficult to penetrate the supply chain of American customers in the short term. In contrast, overseas customers have a stronger willingness to validate mobile DRAM and traditional DRAM.

In other words, Changxin can quickly increase traditional DRAM supply capacity through capacity expansion, but to enter the high-value HBM market, it still needs to solve issues related to manufacturing processes, local packaging ecosystems, advanced logic wafers, thermal reliability, and customer certification.

HBM manufacturing processes and key bottlenecks, including front-end DRAM manufacturing, high-precision TSV, advanced node logic wafers, thermal management, and reliability.

129 Yuan Target Price Bets on More Than Just Shipment Growth

Goldman Sachs expects Changxin's net profit to grow at a compound annual growth rate of 47% from 2026 to 2030, with traditional DRAM revenue growing at a compound annual growth rate of 34%, and HBM revenue growing at a compound annual growth rate of 166%.

The target price of 129 yuan is not directly derived from multiplying 2027 profits by a P/E ratio of 24. Goldman Sachs first assigns Changxin a target P/E ratio of 16.6 times for 2030 based on industry valuation and profit growth relationships, and then discounts it to 2027 at a cost of equity capital of 12.7%, ultimately arriving at the target price of 129 yuan. This price implies about 24 times the projected P/E ratio for 2027.

The factors supporting the valuation increase mainly consist of three layers.

First, the expansion of China's AI infrastructure brings demand for server DRAM and HBM; second, the global traditional DRAM supply is squeezed by HBM capacity expansion, prompting consumer electronics customers to accelerate supplier diversification; third, Changxin's rarity as a large-scale DRAM manufacturer in China allows it to obtain a certain domestic semiconductor valuation premium.

However, the truly aggressive part of this valuation lies in the profit margins.

Goldman Sachs expects that as DRAM prices remain high, shipment volumes expand, and product structures upgrade to DDR5, LPDDR6, and HBM, Changxin's gross margin will rise from 41% in 2025 to 82% in 2030, while the operating expense ratio will decrease from 27.4% to 7.9% during the same period.

This means that the target price of 129 yuan not only requires wafer factories to produce as planned but also demands that new capacity smoothly translates into shipments, that DRAM remains in high demand, that HBM's share continues to increase, and that it ultimately translates into a significant improvement in profit margins.

From Expanding Wafer Factories to Realizing Profits, There Are Many Hurdles to Overcome

Memory is still a typical cyclical industry. When demand is strong, capacity expansion can simultaneously boost revenue and profits; however, when new capacity is concentrated and released, the reversal of supply and demand can quickly depress prices and profit levels.

For Changxin, risks mainly come from three aspects.

First, global manufacturers such as Samsung, SK Hynix, and Micron are still expanding DRAM capacity and developing next-generation products, and stronger-than-expected competition may depress Changxin's shipments and profits. Second, Goldman Sachs' profit margin model is based on the premise that DRAM demand and prices remain strong; if AI or consumer electronics demand falls below expectations, both shipment volumes and gross margins may come under pressure. Finally, the geopolitical environment may limit Changxin's access to the global customer supply chain, weakening overseas growth potential.

Changxin has completed the leap from domestic DRAM "from zero to one" to entering the public market. The next more challenging part is to turn the expanding wafer factories into stable output and extend the scale advantages of traditional DRAM to the performance, yield, and customer certification of HBM.

Therefore, the target price of 129 yuan does not represent a result that has already been realized, but rather a concentrated bet on the simultaneous realization of capacity expansion, high memory demand, domestic substitution, and HBM upgrades.