On July 9, South Korea's KOSPI Index fell 5.4% in a single day. The benchmark has now retreated around 20% from the record high it reached last month, officially entering a technical bear market. Just a few months ago, the KOSPI was one of the world's best-performing major stock indices, with its year-to-date gain peaking at 116%.

Leading the decline were the two companies that had previously driven the market rally—SK Hynix dropped 5.7%, while Samsung Electronics fell 6.3%.

Ironically, Samsung had just reported a 19-fold year-on-year surge in quarterly profit earlier this week. Ian Samson, Portfolio Manager at Fidelity International, highlighted the core issue: "A lot of the volatility stems from uncertainty around the fundamentals. We do see that AI-driven semiconductor demand is real and enormous—but it is ultimately supported by around US$1 trillion in capital expenditure controlled by only a handful of major technology companies."

Should this spending begin to slow, downside risks could emerge rapidly.

The South Korean market also faces a structural vulnerability. A large number of retail investors have built leveraged ETF positions heavily concentrated in semiconductor stocks, amplifying declines when market sentiment reverses. Meanwhile, foreign investors have continued to pull out. Since the beginning of the year, global funds have sold more than US$100 billion worth of South Korean domestic equities.

Capital flows are already telling the story: so far this month, the Hang Seng family of indices has been the best-performing benchmark in Asia, while the KOSPI has been the worst.

Why Did Goldman Sachs Publish This Report?

Reason 1: China's AI Valuations Are Significantly Undervalued

Since the end of 2022, AI-related stocks worldwide have collectively added US$34 trillion in market capitalization. China's AI-related market capitalization stands at approximately US$4 trillion. According to analysts, this figure is "clearly undervalued" relative to China's actual position in the global AI industry.

China accounts for 10% of global AI-related market capitalization and 16% of AI-related revenues. However, as of January 2026, global mutual fund managers allocated only 1.2% of their portfolios to Chinese technology stocks.

This allocation is near historical lows, suggesting substantial room for future inflows. Goldman Sachs also estimates that the potential economic benefits generated by AI through efficiency improvements and new profit creation could be 50% to 100% higher than what current AI stock prices imply.

Reason 2: China's Structural Advantages Are Being Underestimated

China's competitive advantages in AI supply chains, infrastructure, power, and semiconductors have yet to be fully reflected in market valuations.

In another report, Goldman Sachs noted that China is accelerating the construction of its nationwide computing power network. Related infrastructure projects are expected to attract RMB 7 trillion in investment in 2026, while data center investments are projected to reach approximately RMB 2 trillion over the next five years.

The report forecasts that domestically produced AI accelerators could capture more than 50% of total shipments in 2026, with Huawei and Alibaba's T-Head expected to lead the domestic market with 20% and 7% market shares, respectively.

Reason 3: China's AI Still Trails the U.S., Leaving Room for Catch-Up

Goldman Sachs Asia equity strategist Timothy Moe further broke down China's market structure, describing it as "three Chinas."

Offshore software-heavy indices have lagged, while onshore equities and AI hardware-related stocks have rallied sharply. The strongest earnings momentum is concentrated in smaller technology hardware segments.

In other words, the investment opportunity in China's AI sector lies not at the index level, but within specific hardware companies and supply chain segments.

Within the A-share market, Goldman Sachs' top three stock picks are Canadian Solar, Orbbec, and LONGi Green Energy.

Meanwhile, Goldman Sachs maintains an overweight recommendation on Chinese equities within Asia. It has upgraded technology hardware from Neutral to Overweight and remains optimistic on smartphones, AI servers/data centers, semiconductors, and physical AI.

China's Market Is Already Beginning to Validate the Thesis

On July 8, Hong Kong's technology sector surged across the board.

Alibaba jumped 13%, marking its biggest single-day gain in nearly a year. Xiaomi and Kuaishou each gained more than 8%, while Baidu and Lenovo rose nearly 9%. Semiconductor Manufacturing International Corporation (SMIC) climbed 8.2%.

In the A-share market, computing power-related stocks also strengthened significantly. Sangfor Technologies, Langway Technology, and Wangsu Science & Technology each hit the 20% daily limit, while Inspur Electronic Information and several others also reached their daily limits. The Wind Cloud Computing Index advanced 4.05%.

Industry fundamentals also provided support.

On the evening of July 7, Inspur Electronic Information released its earnings guidance, forecasting first-half net profit attributable to shareholders of RMB 2.6 billion to RMB 3.1 billion, representing year-on-year growth of 226% to 288%.

Meanwhile, weekly API calls to Chinese AI large language models have ranked first globally for ten consecutive weeks. However, China's AI industry still faces three major risks:

First, the performance gap in domestic AI chips. Goldman Sachs noted that while domestic chips require 40% to 50% less capital expenditure per unit of IT power consumption than imported chips, their performance gap means capital expenditure per unit of computing power is still two to four times higher. Moreover, computing power generated per unit of power consumption is only 10% to 30% of imported alternatives.

Huawei's Ascend 910B/910C servers produce only around one-sixth to one-third of the daily token output of Nvidia's H800. This performance gap suggests that domestic substitution may progress more slowly than previously expected.

Second, crowded positioning within the AI sector itself. After leading gains in the first half of the year, AI-related technology stocks have recently become more volatile. Themes such as CPO and memory chips have experienced notable pullbacks, with leading names such as Zhongji Innolight retreating more than 20% from their recent highs.

Some institutional investors believe the correction has been primarily driven by liquidity pressures rather than any deterioration in the AI industry's long-term fundamentals.

In addition, China's Interim Measures for the Administration of AI Anthropomorphic Interactive Services will officially take effect on July 15, potentially introducing new compliance requirements for AI application providers.

Third, the sustainability of global AI capital expenditure. The primary catalyst behind the South Korean market's sharp decline was growing concern over whether AI-related capital expenditure can remain sustainable.

That concern does not disappear simply by shifting to another market.

Whether the valuation re-rating of China's AI value chain can continue will ultimately depend on whether Chinese technology companies can deliver on both capital expenditure and earnings growth.