Author: Tidal Research
On June 1st, Goldman Sachs released its daily Asia-Pacific stock summary, "The 720," featuring a long list of names on the cover including Samsung, SK Hynix, Kioxia, MediaTek, Lenovo, and BYD. It appears to be a comprehensive shopping list, but a closer read reveals its absolute core: memory chips.
One of the most significant judgments in this Goldman Sachs report is that the current memory upcycle "will last longer" (higher for longer), with the shortage expected to persist all the way until 2028, a length vastly underestimated by the market. The evidence lies in valuations: most memory stocks are still trading at mid-single-digit price-to-earnings ratios, as if the market believes this is just another ordinary cyclical rebound, but Goldman Sachs sees it differently.
Below, we break it down by importance, concluding with a quick-reference table of targets.
Main Event: Memory Shortage to Last Until 2028, Three Companies Collectively Upgraded
Goldman Sachs compares this cycle to past ones and concludes this time is different. The reasons are threefold: higher visibility of AI server demand, limited supply growth, and increasingly firm long-term supply agreements (locking in orders and prices). Combined, these factors suggest that DRAM, NAND, and HBM supply-demand will be even tighter in 2027 than in 2026, with shortages extending into 2028.
The most intuitive evidence is Goldman's DRAM supply-demand chart. Negative numbers represent supply shortage; the deeper the deficit, the stronger the price support. Goldman has now revised its forecasts for 2026 to 2028 deeper into shortage territory. Specifically, the 2027 forecast was revised from an earlier -2.5% to -5.9%, more than doubling. In plain language: Goldman believes memory manufacturers will face increasing shortages in the next few years, meaning price increases can be sustained longer.
Regarding specific companies, three were collectively adjusted:
- Samsung Electronics: 12-month price target raised to 480,000 KRW, Buy rating maintained.
- SK Hynix: 12-month price target raised to 3.5 million KRW, Buy rating maintained.
- Kioxia: Upgraded from Neutral to Buy, new price target set at 9,300 JPY.
Kioxia is the only rating upgrade in this report, and Goldman's logic is worth noting separately: it believes the profit peak of this cycle is higher than previously anticipated and can be maintained for two to three years, rather than surging and then falling back quickly. Based on this, Goldman has raised its operating profit forecasts for Kioxia for fiscal years 2027 to 2029 by 16% to 48% in one go and expects gross margins to remain at a high level around 80%. Predicting three years of sustainably high profits for a company in the strongly cyclical memory business is a very strong statement.
AI Computing Chain "Full Package": From Chips to Optical Modules to Data Centers
Beyond memory, this report touches upon nearly the entire AI hardware supply chain in China and Asia, unified by one main thread: global cloud hyperscaler capital expenditures are accelerating, and money is flowing down this chain.
- MediaTek: Buy, price target set at 5,000 TWD. The highlight is its transition from mobile chips to data centers and custom ASICs (AI chips tailor-made for specific clients). The company's target is to achieve $2 billion in data center/AI ASIC revenue by 2026 and capture a 10% to 15% share of the $70-80 billion ASIC market by 2027.
- Eoptolink: Buy, price target raised to 841 RMB. It manufactures optical modules, critical components for high-speed data transmission in AI data centers. Goldman is bullish on its 1.6T optical modules ramping up starting Q2 and accelerating in the second half, along with capacity expansion in Thailand. It raised profit forecasts for 2027 and 2028 by 5% and 6%, respectively.
- Biren: Buy, price target raised to 70.7 HKD. A domestic Chinese AI chip manufacturer, its Bili166 product has received a first-tier security and reliability rating. Goldman expects it to turn profitable by 2027 as its product portfolio migrates to higher-performance, more expensive AI chips. Revenue forecasts for 2026 to 2030 were raised by 4% to 28%.
- Huaqin: Buy, a newly covered target in this report. A-share price target set at 149 RMB, with simultaneous first-time coverage of its H-shares, target set at 127.76 HKD. The logic is its move from consumer electronics ODM into AI data centers, with expected revenue CAGR of 32% from 2025 to 2027.
- Data Center Duo: GDS maintains Buy rating, but its ADR target is lowered to $49 (due to slower move-in rates, lower monthly service revenue, partially offset by higher valuation for overseas DayOne business); VNET maintains Buy rating, target raised to $16 (due to Q1 earnings beat, strong execution of capacity ramp-up, and removal of strategic investor overhang).
- Lenovo: Buy, price target raised from 27 HKD to 31 HKD. The bet is on the AI PC refresh cycle. Goldman expects its notebook market share to expand to 28% by 2028, with AI notebook penetration reaching 66%, lifting overall average selling prices. Its profit forecasts for Lenovo's FY2027 and FY2028 are 22% and 25% higher than the Bloomberg consensus, indicating significant divergence.
Stocks Mentioned Outside the AI Main Theme
- Chinese Real Estate (COLI, CR Land): Goldman is assessing the sustainability of the recent real estate sector rebound. It assumes an optimistic scenario where 15 key cities follow Shanghai and Shenzhen's property price recovery, leading to a 15% price increase by the end of 2028. Under this premise, it estimates cash profits for COLI and CR Land could expand by over 30% and over 50%, respectively, by 2028. Using a sum-of-the-parts valuation, Goldman suggests further upside potential of 52% for COLI and 76% for CR Land, maintaining a positive view on these stronger state-owned developers. Key point: this calculation is based on an optimistic assumption, not the base case forecast.
- BYD: Buy, price target set at 137 RMB / 134 HKD. The highlight is its "Tianshen Eye B" City Navigation Guided Pilot (City NOA) becoming a 12,000 RMB option across its lineup at its Intelligent Driving Strategy Conference, making its entry-level City NOA-equipped model the cheapest in China at 78,800 RMB. It also unveiled its first self-developed 4nm intelligent driving chip, "Xuánjī A3," which is already in mass production. Goldman believes these engineering capabilities can boost high-end intelligent driving penetration, lower costs, and improve margins.
- Japanese Semiconductor Equipment: Goldman maintains Buy ratings on Lasertec, Ebara, Disco, and Tokyo Electron. The only contrarian move was downgrading vacuum equipment manufacturer Ulvac (6728.T) from Buy to Neutral, lowering the target to 9,400 JPY, citing weak high-margin power semiconductor orders and slower-than-expected gross margin expansion.
- Panasonic HD: Buy, price target raised from 4,000 JPY to 4,220 JPY, bullish on generative AI-related businesses (backup power, copper-clad laminate CCL, high-performance capacitors).
- NTT: Buy, price target slightly raised from 176 JPY to 179 JPY, focusing on domestic IT service demand and the safety margin provided by a shareholder total return rate of around 5%.
A Macro Theme: AI Boom Clashes with Energy Crisis
Stringing together the individual stocks is Goldman's macro judgment: Emerging markets are being pulled in two by two opposing forces. On one side is the AI investment boom, and on the other is energy supply contraction due to the blockade of the Strait of Hormuz.
Tech-exporting economies like South Korea and Taiwan benefit from surging exports and current account surpluses; while energy-importing countries face rising inflation, weaker currencies, and fuel subsidies straining their finances. Goldman expects the Q4 average Brent crude price to be $90/barrel, continuing pressure on heavily oil-import-dependent economies. It recommends overweighting stocks in China, South Korea, Brazil, and South Africa. This line connects with the recent macro background of the Iran situation and oil prices.
Two other points directly affecting A-share capital flows:
Chinese imports surged 23.6% year-on-year in the first four months of this year, but Goldman believes this is a highly concentrated phenomenon, with gold and semiconductors accounting for about 65% of the import growth increment, not indicative of a continuous worsening of the external balance.
Regarding the CSI and CNI index semi-annual rebalancing, Goldman estimates it will trigger over $48 billion in two-way passive fund flows. The technology hardware & semiconductor and capital goods sectors are expected to see the largest inflows (approximately $3.1 billion and $1.4 billion, respectively), while healthcare and banks see the largest outflows. Newly added constituents "expected to receive the largest net passive inflows" include Huagong Tech, Yuanjie Tech, Huahong Semiconductor, GigaDevice, VeriSilicon, etc. For index rebalancing arbitrage funds, this is a clear signal.
Finally, as usual, Goldman left an Easter egg: 2026 World Cup winning probability forecast: Spain leads at 26%, followed by France at 19%, Argentina at 14%, Brazil at 8%, and England at 5%. The model penalized the defending champion Argentina; take it as a bit of fun.
Quick Reference Table of Targets
This article is TechFlow's compilation and interpretation of a third-party brokerage research report. The ratings, price targets, profit forecasts, and related judgments cited herein represent the views of the report's analysts and their affiliated institutions only. They do not represent the views of TechFlow and do not constitute any investment advice.
Please note the following three points when reading:
1. Price targets are analysts' expectations for a future period (typically 12 months). They are forecasts, not promises, and are subject to frequent adjustment based on company performance and market conditions.
2. Sell-side research reports are inherently bullish. It is common for brokerages to issue "Buy" ratings on covered companies, and some covered companies may have investment banking or other business relationships with the brokerage. A list predominantly featuring "Buy" ratings should be read with this inherent bias in mind.
3. The value of a research report lies in its mainline logic and the underlying assumptions it depends on, not any single price target. The logic for related targets holds only if the main theme holds; if the main theme is disproven, the entire chain of targets will come loose together. Focus on the logic, not just the price.
Markets involve risks; decisions should be made independently. This article should not be used as the basis for buying or selling any securities.
