Goldman Sachs sees a boom: storage shortages surge and reach 2028, Samsung and SK Hynix collectively raise prices

Goldman Sachs predicts this upward memory cycle will last even longer, with DRAM and NAND supply and demand becoming tighter in 2027 than in 2026.
(Background summary: MiniMax prepares for A-share IPO on the Sci-Tech Innovation Board! Listing simultaneously with Zhipu to attract AI capital)
(Additional context: Dell surges 39%: AI market expansion reaches the brand hardware supply chain)

Table of Contents

Toggle

• Main focus: Storage shortages to last until 2028, three companies collectively upgraded
• AI computing power chain: from chips to optical modules to data centers
• Targets not on the AI mainline but also mentioned
• Macro mainline: AI boom collides with energy crisis

On June 1, Goldman Sachs launched the daily Asia-Pacific stock review “The 720,” with the cover featuring Samsung, SK Hynix, Kioxia, MediaTek, Lenovo, BYD, and a long list of names. It looks like a comprehensive shopping list, but upon reading, it reveals an absolute core: memory chips.

The most significant judgment in this issue from Goldman Sachs is that this current storage upcycle “will last longer” (higher for longer), with shortages continuing into 2028, and the market has significantly underestimated its duration. The evidence is in valuation: most storage stocks are still trading at single-digit to mid-single-digit P/E ratios, as if the market considers this just another normal cycle rebound. Goldman Sachs disagrees.

Below, we break down the key points, ending with a quick overview table of targets.

Goldman Sachs compares this cycle with past ones and concludes this time is different. There are three reasons: higher visibility of AI server demand, limited supply growth, and long-term supply agreements becoming more rigid (locked-in orders and prices). The combination of these factors means that in 2027, DRAM, NAND, and HBM supply and demand will be tighter than in 2026, with shortages extending into 2028.

Main focus: Storage shortages to last until 2028, three companies collectively upgraded

The most straightforward is Goldman Sachs’s DRAM supply-demand chart. Negative numbers indicate supply shortages; the deeper the gap, the more price support. Goldman Sachs has lowered its forecasts for 2026 to 2028 into deeper shortage zones, with 2027’s forecast changing from -2.5% to -5.9%, nearly doubling. In plain language: Goldman Sachs predicts storage manufacturers will face increasing shortages next year and the year after, meaning price increases can last longer.

Focusing on specific companies, three have been collectively upgraded:

• Samsung Electronics: target price raised by 12 months to 48,000 KRW, maintain buy.
• SK Hynix: target price raised by 12 months to 3.5 million KRW, maintain buy.
• Kioxia: upgraded from hold to buy, new target price 93k JPY.

Kioxia is the only one of this cycle’s ratings upgrade from Goldman Sachs. Its logic is worth a separate look: Goldman Sachs believes this cycle’s profit peak will be higher than previously expected and can be sustained for two to three years, not just a quick spike. Based on this, Goldman Sachs has sharply raised its profit forecasts for Kioxia’s 2027-2029 fiscal years by 16% to 48%, and expects gross margin to stay around 80%. Giving a company in the storage cycle such a strong statement—sustainable high profits over three years—is quite significant.

Beyond storage, this report almost covers China and Asia’s AI hardware supply chain, with a unified logic: global hyperscaler capital expenditures are accelerating, money flowing down this chain.

• MediaTek: Buy, new target NT$5,000. Focus on its transformation from mobile chips to data center and customized AI chips (ASICs). The company aims to reach US$2 billion in data center/AI ASIC revenue by 2026, capturing 10-15% of the US$70-80 billion ASIC market in 2027.

• Eoptolink: Buy, target price raised to RMB841. It makes optical modules, key components for high-speed data transfer in AI data centers. Goldman Sachs expects its 1.6T optical modules to ramp from Q2, accelerate in H2, and expand production in Thailand, raising profit forecasts for 2027 and 2028 by 5% and 6%, respectively.

• Biren: Buy, target price raised to HK$70.7. Domestic AI chipmaker, with its Bili166 rated as secure and reliable. Goldman Sachs expects it to shift to higher computing power AI chips, sell at higher prices, turn profitable in 2027, and has raised revenue forecasts for 2026-2030 by 4% to 28%.

• Huaqin Technology: Buy, a “new coverage” target. A-shares target price RMB149, first coverage of H-shares with a target of HK$127.76. Logic: shifting from ODM consumer electronics to AI data centers, with a compound annual growth rate of 32% from 2025 to 2027.

• Data center giants: GDS maintains buy, but ADR target price lowered to US$49 (slower deployment, monthly service revenue declining, partly offset by higher valuation of overseas DayOne business); VNET maintains buy, target price raised to US$16 (Q1 results exceeded expectations, strong capacity investment outlook).

• Lenovo: Buy, target price raised from HK$27 to HK$31. Focus on AI PC upgrade wave; Goldman Sachs expects its laptop market share to expand to 28% by 2028, AI laptop penetration to 66%, raising overall average prices. Its profit forecasts for 2027 and 2028 are 22% and 25% higher than Bloomberg consensus, respectively.

• China real estate (China Overseas, China Resources Land): Goldman Sachs is assessing whether this rebound in the property sector can hold. It assumes an optimistic scenario where 15 key cities follow Shanghai and Shenzhen in recovering, with housing prices rising 15% by end-2028. Under this, estimates show China Overseas (COLI) and China Resources Land (CR Land) could see cash profits grow by over 30% and 50%, respectively, by 2028. Based on segment valuation, Goldman Sachs sees further upside of 52% for China Overseas and 76% for CR Land, maintaining a positive outlook on these stronger state-owned developers. Key point: this is based on an optimistic assumption, not a baseline forecast.

• BYD: Buy, target price RMB137 / HK$134. Focus on its smart strategy, making “Tian Shen Eye B” city navigation assist (NOA) an optional feature costing RMB12k, with entry models priced at RMB78.8k, making it the cheapest city NOA vehicle in China. Also announced its first self-developed 4nm autonomous driving chip “Xuanji A3,” already mass-produced. Goldman Sachs believes these engineering capabilities will boost high-end autonomous driving penetration, lower costs, and improve margins.

• Japanese semiconductor equipment: Goldman Sachs maintains buy ratings on Lasertec, Ebara, Disco, Tokyo Electron. The only downgrade is Ulvac (6728.T), from buy to neutral, target price lowered to ¥9,400, due to weak orders for high-margin power semiconductors, leading to slower gross margin expansion.

• Panasonic HD: Buy, target price raised from ¥4,000 to ¥4,220, optimistic about generative AI-related businesses (backup power supplies, copper-clad laminate CCL, high-performance capacitors).

• NTT: Buy, target price slightly raised from ¥176 to ¥179, based on domestic IT service demand and a roughly 5% shareholder return rate providing a safety margin.

AI computing power chain: from chips to optical modules to data centers

Connecting individual stocks is Goldman Sachs’s macro judgment: emerging markets are being torn apart by two forces. One is the AI investment boom; the other is the energy supply contraction caused by the Hormuz Strait blockade.

Tech-exporting economies like Korea and Taiwan benefit from surging exports and current account surpluses; energy-importing countries face rising inflation, currency weakness, and fiscal strain from fuel subsidies. Goldman Sachs expects Brent crude oil prices to average $90 per barrel in Q4, continuing pressure on heavily oil-dependent economies, and recommends overweighting Chinese, Korean, Brazilian, and South African stocks. This macro backdrop aligns with recent Iran tensions and oil price movements.

Targets not on the AI mainline but also mentioned

Two other points directly affecting A-share capital flows:

China’s imports surged 23.6% YoY in the first four months of this year, but Goldman Sachs judges this is highly concentrated, with gold and semiconductors accounting for about 65% of the increase, not indicating a sustained external imbalance.

The semi-annual rebalancing of CSI and CNI indices is estimated by Goldman Sachs to trigger over US$48 billion in passive capital flows, with the most inflows into tech hardware, semiconductors, and capital goods (about US$3.1 billion and US$1.4 billion, respectively), and the most outflows from healthcare and banking sectors. Newly highlighted targets expected to see the largest passive net inflows include Huagong Tech, YuanJie Technology, Huahong Semiconductor, GigaDevice, and Chipone. For funds engaging in index rebalancing arbitrage, this is a clear signal.

Finally, Goldman Sachs has embedded a fun Easter egg: its 2026 World Cup winner probability forecast ranks Spain first at 26%, France 19%, Argentina 14%, Brazil 8%, and England 5%. The model deducts points from defending champion Argentina—just for fun.

Macro mainline: AI boom collides with energy crisis

Please note three points when reading:

1. Target prices are analysts’ expectations for the next 12 months (usually), not commitments, and will be adjusted based on company performance and market conditions.

2. Sell-side research reports tend to be optimistic. It’s common for brokerages to give “buy” ratings, and some covered companies have business interests with the brokerages. When reading a “buy” list, keep this context in mind.

3. The value of research reports lies in their main logic and underlying assumptions, not just target prices. If the main logic holds, the related targets make sense; if it’s invalidated, the entire chain of targets will loosen. Focus on the logic, not just the prices.