AI capital expenditure is tied to the lifeline of South Korea's economy, and the central bank warns that the decline in exports could reach US$100 billion! Why does Goldman Sachs still remain bullish on South Korean stocks and the two memory giants?

The latest core risks to South Korea's economy revealed in a Bank of Korea research report show: if U.S. tech giants slow AI capital expenditure next year, South Korean semiconductor exports could be US$50 billion to US$100 billion lower than originally forecast, dragging down 2027 GDP growth by about 0.4 percentage points. However, this is a calculation and outlook under a pessimistic scenario, not an export loss that has already occurred, nor is it a baseline forecast that exports will decline by US$100 billion year-on-year. At the same time, although Wall Street financial giant Goldman Sachs cut its KOSPI target by 25%, it still believes memory chip valuations are overly pessimistic; South Korean retail investors, meanwhile, are buying back Samsung Electronics, SK Hynix, and those triple-leveraged ETF products closely related to the chip theme. Together, these three signals constitute the current divergence in the AI computing power trading themeat the macro level, some investors remain continuously worried about a series of major negative shocks brought by cooling AI investment among North American tech giants under increasingly elevated AI financing cost pressure, while the stock market has begun to judge whether such pessimistic risk has already been overpriced as valuation metrics for South Korea's two major memory chip giants have fallen sharply. "South Korea cannot afford to lose the AI memory competition" has a realistic basis. South Korea's third-quarter GDP grew 1% quarter-on-quarter, exceeding the market expectation of 0.7%; the Bank of Korea estimates that semiconductors contributed about half of economic growth, while private consumption fell 0.2% quarter-on-quarter. The chip boom has already deeply affected exports, corporate investment, and overall economic performance. This is also why, in the view of South Korean President Lee Jae-myung and Bank of Korea officials, South Korea needs to consolidate its high-end memory advantage, expand real AI applications and customer sources, and at the same time reduce the economy's dependence on a single capital expenditure cycle. The underlying mechanism of export pressure lies in the fact that shipment volume and average selling price jointly determine export value. South Korean media disclosed that in the pessimistic scenario, chip export volume may still grow by about 10%, but below the original expectation, while price performance is weaker than the baseline assumption. Therefore, even if demand for AI computing power resources continues to expand, as long as procurement growth and prices are below the previously extremely high expectations, it is enough to create a huge export gap. For memory stocks, the key question therefore becomes: will future earnings merely slow in growth, or will they suffer a deep decline under the recession cycle implied by current pessimistic valuations? Bank of Korea warns: if AI spending slows, South Korean chip exports could be reduced by as much as US$100 billion versus the original forecast Bank of Korea data show that if large U.S. technology companies slow artificial intelligence capital expenditure next year, South Korean semiconductor exports could be US$50 billion to US$100 billion lower than previously forecast. In a written response submitted to lawmaker Ahn Do-geol, the Bank of Korea constructed a "pessimistic scenario" to assess the economic impact of U.S. tech giants cutting large-scale infrastructure investment. Under this scenario, exports suddenly falling short of expectations would make South Korea's 2027 current account surplus significantly lower than the central bank's initial forecast of US$430 billion, and would reduce next year's gross domestic product (GDP) growth by about 0.4 percentage points. The scenario assumes a 10% decline in South Korean semiconductor unit prices. This latest Bank of Korea assessment highlights deepening market concerns over South Korea's heavy reliance on chip exports. Chips currently account for about half of South Korea's total exports. The report said that combined capital expenditure by five large U.S. technology companiesAmazon, Google, Microsoft, Meta, and Oraclegrew rapidly, from US$40 billion last year to a conservative estimate of US$80 billion this year, and a conservative estimate of a staggering US$110 billion next year. However, the Bank of Korea cautioned that there remains considerable uncertainty regarding the trajectory of demand related to frontier AI technologies and the overall memory chip business cycle fully driven by AI computing power demand. Bank of Korea calculates export gap, while Goldman Sachs seeks expectation gap in memory chip stocks A Goldman Sachs research report shows that the institution cut its 12-month target for South Korea's benchmark stock index, the KOSPI Composite Index, from 12,000 points to 9,000 points, mainly reflecting more conservative valuation assumptions; based on the report's timing, the new target still implies more than 60% upside. Goldman Sachs estimates that current semiconductor valuations imply a roughly 65% earnings decline in 2027, a scenario that is overly pessimistic for Goldman Sachs analysts and indicates market pessimism has gone severely overboard. Samsung Electronics and SK Hynix were each trading at only about 3x forward price-to-earnings at the time; the implied return on equity corresponding to the memory chip sector's price-to-book ratio was only about 23%, clearly below the market consensus expectation of 45%. These figures reflect the distance between market pricing and earnings expectations. The implied 65% earnings decline is an excessively pessimistic market expectation calculated based on KOSPI trading levels, not Goldman Sachs' own forecast and outlook. The Goldman Sachs analyst team noted in the research report that Samsung Electronics and SK Hynix are currently both trading at about 3x forward price-to-earnings, the lowest level in history. Based on the relationship between price-to-book ratio and return on equity with an R-squared (R) goodness of fit of as high as 72% revealed by the past 20 years of monthly data compiled by Goldman Sachs, the current sector price-to-book ratio of 2.6x implies a 2027 ROE of only 23%, while the market consensus forecast for ROE is as high as 45%. This divergence is equally prominent at the index level: the KOSPI is down about 40% from its June high, while forward 12-month earnings forecasts over the same period have risen about 22%; the KOSPI's record-low forward price-to-earnings of about 4x is at an extreme low among the 18 major trillion-dollar markets covered by the relevant statistics. However, the memory industry's low price-to-earnings ratio may also include concerns that earnings at the cycle peak are unsustainable. The key to Goldman Sachs' bullish judgment on South Korea's two major memory chip giants and South Korea's benchmark stock index KOSPI is: as long as the actual magnitude of earnings downgrades is smaller than the degree of pessimism implied by share prices, combined with a decline in risk premium, South Korea's stock market and South Korea's two major memory chip giants, which account for 50% of South Korean stock market capitalization, may achieve a strong valuation recovery during a period of slowing earnings growth. The Goldman Sachs research report also shows that South Korean stock market funds have already begun betting on this expectation gap. From October 1 to 23, South Korean individual retail investors heavily net bought 14 single-name high-leverage ETF products related to Samsung Electronics and SK Hynix, totaling KRW315.4 billion, of which SK Hynix-related products accounted for KRW226.3 billion and Samsung-related products accounted for KRW89.1 billion. From October 1 to 22, South Korean investors heavily net bought about US$4.8 billion of U.S. market stocks, of which about US$3.548 billion flowed into the triple-leveraged long U.S. market Philadelphia Semiconductor ETFthe U.S.-listed ETF with ticker "SOXL"accounting for 73.9%. This shows that bottom-fishing funds are highly concentrated in the memory chip and semiconductor industry chain driven by AI computing power demand, but products with daily-reset leverage also amplify volatility losses, and the return of funds itself does not prove that the market has bottomed. The industry-chain basis for the global memory chip sector's counterattack is that, as frontier AI agents such as Muse become popular worldwide, demand for computing power resources at the AI high-performance computing level is spreading to more memory and larger-capacity NAND enterprise-level storage tiershigh-bandwidth memory (HBM) serves accelerator computing, server DRAM carries data and task environments in operation, and enterprise SSDs preserve indexes, task records, and persistent state. From the perspective of AI inference system architecture, more complex AI tasks led by Meta Muse and OpenAI's GPT-6 Astra often include longer contexts, multi-turn model calls, tool execution, and result verification: prefill needs to process input, decode continuously generates output, and the key-value cache (KV Cache) occupies more HBM memory as context and concurrency scale expand, while computing throughput, memory bandwidth, and enterprise-grade high-performance storage capacity need to improve in coordination. SK Hynix has disclosed that HBM4 began mass shipments in the second quarter and that it has established long-term supply agreements with about ten customers; Bank of America, after recently communicating with Micron management, still emphasized supply constraints and the support of long-term agreements for demand visibility. Market research firm TrendForce expects fourth-quarter conventional DRAM contract prices to rise 10%15% quarter-on-quarter on a high base, and NAND contract prices to rise 15%20%. On the DRAM side, agent workloads are driving procurement of supporting general-purpose servers and memory, while advanced processes, packaging and testing, and capacity allocation face constraints; on the NAND side, real-time retrieval, caching, and vector databases are expanding demand for enterprise solid-state drives (Enterprise SSD). TrendForce expects enterprise SSD bit demand to grow more than 80% year-on-year in 2026, with most of the new supply already booked, so the price increase for this category in the fourth quarter may even expand compared with the third quarter. TrendForce also sharply raised its 2027 HBM blended average selling price (Blended ASP) forecast to a year-on-year increase of 121%, based on the fact that HBM and conventional DRAM are competing for limited advanced process and wafer capacity, yield ramp-up for new-generation products takes time, and at the same time the share of higher-priced HBM4 is rising and HBM4e is gradually ramping. During the morning session on October 27, South Korea's KOSPI rose more than 4%, after the Philadelphia Semiconductor Index rose about 2.8% overnight, and Anthropic's official IPO filing for a U.S. listing and Meta's equity financing plan also improved market expectations for the sources of AI construction funding. The core factor that will allow the large-scale counterattack in the U.S. semiconductor sector on Monday to continue will be the formation of persistently stronger procurement expectations after financing is implemented, which is expected to ultimately earnings and cash flow for memory manufacturers.