Goldman Sachs Research estimates: The potential economic benefits of China’s AI, through efficiency gains and new profit creation, could be 50%–100% higher than current stock prices imply.

Article author and source: 0x9999in1, ME News



TL;DR

• On July 9, 2026, Goldman Sachs’ Thematic Investing team released the report “Investment Strategy: Long China’s AI Value Chain,” officially adding China’s AI sector to its core recommendations and simultaneously launching the Goldman Sachs China AI Value Chain portfolio (GSXACART).
• On the day the report was released, the Korea Composite Stock Price Index retreated approximately 20% from its June historical high, entering technical bear market territory; the Hang Seng China Enterprises Index surged as much as 4.5% intraday, marking its largest single-day gain since February 2025.
• Goldman Sachs' core argument: Since the end of 2022, the global market capitalization of AI-related public companies has increased by $34 trillion, while China's AI sector currently has a market cap of only about $4 trillion; China contributes 10% of global AI market capitalization and 16% of global AI revenue, yet global mutual funds allocate just 1.2% of their holdings to Chinese technology.
• Goldman Sachs Research estimates: The potential economic benefits of China’s AI, through efficiency gains and new profit creation, could be 50%–100% higher than current stock prices imply.
• In May, China's chip sales increased by 111% year-over-year; Yangtze Memory Technologies' global NAND market share rose from 8% to 13%, with Q1 revenue up 445% year-over-year, and CXMT is projected to exceed $50 billion in revenue by 2026.
• This is not a simple rotation, but a structural repricing of capital—from the old narrative of "AI = NVIDIA + SK Hynix" to the new narrative of "AI = a complete, self-controlled industrial chain."

A report that caused the entire Asian market to reverse course that day

First, the conclusion.

The importance of Goldman Sachs' report lies not in the novelty of its views, but in the timing.

At凌晨 on July 9, 2026, analyst Louis Mille from Goldman Sachs’ Thematic Investment team sent a report titled “Investment Strategy: Long the Chinese AI Value Chain” to global clients—with just one sentence: “China’s AI industry has officially entered our radar.”

In the past, we didn't hold large positions; starting today, we will go heavily weighted.

On the same day, the Korea Composite Stock Price Index closed lower, entering a technical bear market with a cumulative decline of approximately 20% from its June historical high. Samsung Electronics and SK Hynix—the two storage giants elevated to iconic status by the global AI narrative—experienced their first collective decline in over a year.

What about the opposite side? The Hang Seng China Enterprises Index surged as much as 4.5% intraday, posting its largest single-day gain since February 2025. The Sci-Tech Innovation Board in mainland China rallied sharply in the afternoon, while the A50 Index accelerated upward.

This is no coincidence. This is the sound of a gun's trigger being pulled.

For over a year, Goldman Sachs has been a staunch bull on Korean equities. Its models, roadshows, and client recommendations all centered around the "HBM cycle + NVIDIA ecosystem." When an institution like Goldman Sachs begins to rebalance its portfolio, it signals that even the most conservative capital on Wall Street has finally turned the corner.

You ask: Why today?

Because the story of South Korea’s AI has run its course. SK Hynix’s IPO in the U.S. was over-subscribed sevenfold, appearing booming—but it was the final peak, not the start of a new cycle. The market’s shift from belief to doubt about whether the storage supercycle can continue took just one quarter.

What about China’s AI sector? ChangXin Memory passed its review, Unigroup Guoxin submitted its listing application, Yangtze Memory’s NAND market share surged, DeepSeek’s models underwent iterations, and ByteDance and Alibaba increased their capital expenditures—all of these events occurred precisely within the time window before Goldman Sachs’ report was released.

$4 trillion vs. $34 trillion: The gap Goldman Sachs sees

Of the three arguments Goldman Sachs presented this time, the first is the most striking.

Since the launch of ChatGPT at the end of 2022, the total market capitalization of global AI-related public companies has increased by $3.4 trillion. What does this mean? It’s equivalent to creating two more Apples, five more NVIDIAs, or ten more TSMCs.

What is the current total market capitalization of China's AI sector? Approximately $4 trillion.

Accounts for approximately 10% of the global AI market capitalization.

This number, if viewed in isolation, might seem "acceptable." But when compared to China's actual weight in the global AI industry—

Goldman Sachs estimates that China accounts for 16% of global AI-related revenue. This means China, with 16% of the revenue, has received only a 10% valuation.

Even more striking numbers lie ahead: as of January 2026, global mutual fund managers' allocation to China's technology sector stands at just 1.2%.

1.2%.

Not 1.2x PEG, not an annualized spread of 1.2%, but a position size of 1.2%.

10% of market capitalization contribution, 16% of revenue contribution, 1.2% of institutional allocation.

These are three severely misaligned curves. Anyone with a background in finance knows that misalignment presents an opportunity.

But more importantly, Goldman Sachs is not the first institution to make this point. Just a few days ago, Quantum Strategy explicitly stated in its client report: shift away from the U.S. Big Seven tech stocks and toward Chinese AI application stocks. Morgan Stanley’s report during the same period noted that the valuation premium of the "Big Seven" U.S. stocks has reached a ten-year low.

One end features the world's most expensive tech asset at its valuation peak, while the other end shows the world's cheapest AI asset position at its low point.

How will capital choose? Without needing Goldman Sachs to remind us, the answer is already written on the chart.

"Three Chinas": This is the section most worth reading closely.

In his analysis of the report, Goldman Sachs Asia equity strategist Timothy Moe presented a particularly insightful framework—China’s stock market is effectively being divided into "three Chinas."

First China: Offshore Software Weighted Index. Underperforming, with valuation suppression unresolved.

Second China: Onshore market. Steady recovery, with clear policy support.

Third China: AI-related hardware stocks. Soaring, with the strongest profit momentum.

This is not wordplay. This is the result of a vote by funds.

Over the past few years, global investors looked at China’s stock market and focused solely on the words “China”—either rejecting it entirely or investing in it wholesale. Either they avoided it completely or increased their positions across the board.

But after 2026, the concept of "China" is being dismantled.

Traditional internet software companies have reached growth ceilings, regulatory pressures have slightly eased, but valuations are reassessing slowly; consumer sectors are being pulled back and forth by trade-in programs and real estate cycles; while AI hardware—from computing power, storage, and optical modules to power supply, transformers, and liquid cooling—has become a completely independent β.

It corresponds not to the "Chinese economic story," but to the "Chinese AI infrastructure story."

It reflects the spillover effect of global AI capital expenditures, not retail investor sentiment.

It doesn’t even care much about the year-over-year GDP figures—it’s far more focused on the utilization rate of domestic data centers, the progress of domestic alternatives to HBM3E, and the yield ramp-up of Chiplet packaging.

With this cut, China’s AI hardware sector is no longer just a collection of Chinese概念股, but a segment of the global AI supply chain—perhaps the most undervalued one.

Goldman Sachs has launched the "Goldman Sachs China AI Value Chain" portfolio (ticker: GSXACART), covering five segments: power, semiconductors, AI infrastructure, AI models, and AI applications.

Pay attention to the order. Electricity is listed first.

This is not arbitrary layout—it’s industrial logic: the end goal of AI is electricity, a global consensus, and China is no exception.

Changxin, Changjiang, Ziguang: Three names supporting a compelling narrative

To prove that Chinese AI is not a bubble, you need more than just index charts—you need the revenue curves of specific companies.

Goldman Sachs selected three companies as key examples, all focused on the storage segment.

First, let's look at CXMT.

Goldman Sachs estimates that in 2026,鑫存储's annual revenue could exceed $50 billion. What does this mean? It implies that since 2023, the company’s revenue has more than doubled each year, and in 2026, year-over-year growth could surpass 600%.

What level of revenue is $50 billion? For comparison—Micron Technology's revenue for fiscal year 2024 was approximately $25.1 billion, and SK Hynix's full-year 2024 revenue was approximately 66 trillion Korean won (about $47.5 billion).

In other words, CXMT must reach the scale of the world's top players in the DRAM market within one year.

Now consider Yangtze Memory Technologies. Its global market share in NAND flash memory rose from 8% a year ago to 13%, placing it tied for fourth globally; revenue in Q1 2026 increased nearly 445% year-over-year.

The third company, Unigroup Guoxin, is submitting its prospectus to the Beijing Stock Exchange. The sponsor is CITIC Construction Securities.

What are these three companies doing together?

Together, they transform China's storage industry narrative from one of "catching up" to one of "competing."

In the past, the three major players in storage were Samsung, SK Hynix, and Micron. In the future, when discussing storage, you cannot overlook Yangtze Memory and CXMT.

It's not about sentiment; it's about share.

Adding to this, China's chip sales in May rose 111% year-over-year, and overall exports increased by 19.4%—this is no longer just "AI hype," but solid customs data.

Goldman Sachs used a term in its report: "structural rather than temporary." In other words, this cycle is not about thematic speculation—it's about fundamentals.

50%–100% undervaluation: Why is Goldman Sachs making such a bold claim this time?

Goldman Sachs provided a very bold figure in its report—China’s AI-driven potential economic benefits from efficiency gains and new profit creation could be 50%–100% higher than what is currently implied by stock prices.

50% to 100% undervaluation potential.

If this were said by a self-media outlet, I might just laugh it off. But coming from the precise wording of Goldman Sachs’ Global Investment Research division, it’s something to take seriously.

The reasoning behind it is actually quite simple:

First, the ceiling for global AI capital expenditures continues to rise. Goldman Sachs previously estimated that AI capital spending could reach $7.6 trillion over the next six years. Regardless of whether this $7.6 trillion is ultimately spent in the United States, South Korea, or China, China’s supply chain cannot be entirely absent.

Second, China's narrative of "autonomy and controllability" in AI has given it a second valuation curve independent of NVIDIA's ecosystem. This curve was previously suppressed by geopolitical factors, but precisely because of this suppression, the certainty and urgency of domestic alternatives have been significantly heightened.

Third, the pace of commercialization of AI applications in China is far faster than outsiders realize. Names like DeepSeek, Qwen, Kimi, and Doubao are no longer just "Chinese versions of OpenAI"—they are models actively used by developers worldwide. Metrics such as API pricing, token usage volume, and enterprise customers are steadily rising.

The combination of three factors results in a 50%–100% upside potential.

Is it high? Not low.

Unreasonable? Given the current market-to-revenue ratio of China’s AI hardware industry, it’s not unreasonable at all.

However, note that Goldman Sachs provided a "potential space," not a "target price."

The potential space implies two key factors that determine whether this potential can be realized: first, AI capital expenditures cannot experience a sharp decline; second, U.S.-China technological competition must not escalate to extreme scenarios.

If either of these two variables deviates, the discount will be reduced by 50% to 100%.

Will today's Korea be tomorrow's China?

At this point, we must bring things back to reality.

Goldman Sachs recommends going long on China’s AI value chain—the logic is sound, the data is solid, and the timing is precise. But should you blindly jump in? The answer is no.

Why did Korea halt trading today? Because over the past year, global investors used leveraged ETFs, margin financing, and every available form of leverage to push SK Hynix and Samsung Electronics to their valuation limits. When even a slight shift in fundamentals occurred—such as Micron’s inventory guidance or rumors of reduced capital spending by Microsoft—a cascade of forced selling ensued.

Nomura's report stated that market panic over cooling storage is "clearly overblown." That's true. But on the flip side, once sentiment shifts, a rebound will still take time.

Does China's AI value chain face similar risks?

Yes.

First, valuation dispersion is widening. The P/E ratios of top-tier assets have already reached historical highs, while lower-tier assets are still catching up. Once foreign capital bundles these assets together for allocation, significant beta risk will emerge in the middle.

Second, the alignment between production capacity and demand has not yet been proven. The capacity ramp-up of XMC and Yangtze Memory is aggressive. If global AI capital expenditure growth slows by 2027, could the newly added domestic capacity become stuck in inventory? This is a question that must be asked.

Third, geopolitical factors can never be ignored. The marginal direction of U.S. sanctions on China’s semiconductor industry will directly impact the pace of domestic substitution. If sanctions ease, domestic substitution slows; if sanctions tighten, short-term earnings come under pressure. Either scenario will amplify market volatility.

Therefore, for ordinary investors, my assessment is:

The direction is correct, but timing is harder than direction.

Goldman Sachs recommends a "value chain portfolio," not a single stock. What this means is—if you're going to invest, invest in the entire chain. Betting on a single point amplifies both risk and reward.

From "NVIDIA Faith" to "China Value Chain Faith"

Finally, let’s raise the perspective and talk about one thing.

From 2023 to 2025, the global AI capital market had only one belief—NVIDIA. Around it, an entire ecosystem of supporting narratives emerged, including HBM, CoWoS, optical modules, liquid cooling, and power infrastructure. All capital flowed into this belief.

By 2026, this belief first showed signs of cracking.

It’s not because NVIDIA is failing. The GB300 is still selling like crazy, and Rubin is on the way.

But because the market has finally realized that the AI value chain cannot be monopolized by a single country.

The United States holds an absolute lead in chip design. TSMC possesses exclusive production capacity in advanced processes. South Korea has a stockpile advantage in HBM. Japan boasts hidden champions in materials and equipment. China, meanwhile, has abundant electricity, manufacturing capabilities, application scenarios, accelerated domestic substitution, and the world’s most comprehensive supply chain depth.

Only when these five segments are combined do they form the global AI value chain.

Any segment being artificially hyped to unsustainable heights is unhealthy. When one segment (Korean storage) is sold off and another (Chinese hardware) is lifted up, this isn't a "zero-sum game"—it’s capital reallocating its weight toward a "complete value chain."

The deeper implication of Goldman Sachs' report is actually an announcement of a shift in belief:

Shift from betting on single-point technological monopolies to betting on the revaluation of the entire value chain.

Shift from "NVIDIA eats all the alpha" to "Each segment of the supply chain has its own alpha."

China's AI was the largest alpha discovered in this repricing.

In conclusion

Goldman Sachs is not a philanthropist. It’s calling for a long position on China’s AI value chain today not because it loves China, but because its models indicate that’s the outcome.

Don’t idolize this report. Wall Street might be bullish today and bearish tomorrow. What truly determines where money flows is never a single analyst’s comment, but the real progress of the industry.

But this report does mark a moment—

A Korean AI story begins to crack at its highest note, while a Chinese AI story transitions from "expectation" to "data."

A moment when a global mutual fund's allocation to Chinese technology stands at just 1.2%, nearly as low as it can possibly go.

A moment when ChangXin Memory is preparing for an IPO, Yangtze Memory is ramping up production, and Unigroup Guoxin is submitting its listing application—multiple catalysts aligning simultaneously.

A moment when the "Three Chinas" are clearly segmented, and AI hardware is independently established as β.

Time waits for no one.

Goldman Sachs has pulled the trigger. The bullet is already on its way.

Who gets hit and how far it reaches depends on every upcoming earnings report, every production capacity announcement, and every product launch.

Seeing through everything won’t help. In this market, the first ones to spot the direction are the first to get on board. Whether you get motion sick or get off midway—that’s a story for later.

The direction has changed. The rest is up to you.

Reference materials

1. Caixin. "Goldman Sachs Issues Major Statement: Go Long on China's AI Value Chain." July 9, 2026. https://www.cls.cn/detail/2421215
2. Everyday Financial News / NetEase Finance. "Goldman Sachs: Chinese AI Companies' Market Valuations Are Severely Misaligned with Their Market Potential, Leaving Substantial Upside Room." July 9, 2026. https://www.163.com/dy/article/L1D6JPDR05198CJN.html
3. TMGM Market Insights. "Abandoning Korea, Turning to China! Goldman Sachs Releases Report: Long China's AI Value Chain." July 9, 2026. https://www.tmgm.com/zh-hans/analysis/market-insight/goldman-sachs-china-ai-investment-report
4. Jiemian News. "Wall Street Institutions Quantum Strategy Bets on Chinese AI Amid Exit from U.S. Tech Giants." July 7, 2026. https://www.jiemian.com/article/14714391.html
5. 21st Century Business Herald. "Afternoon Surge in A50! Sci-Tech Innovation Board Explodes! What Happened?" July 9, 2026. https://m.21jingji.com/article/20260709/herald/cba6afc04817df9ba57e7a9254b360d2.html
6. Cnyes. "AI Ignites 13% Global Capital Expenditure Growth! Goldman Sachs Reiterates HALO Trade." July 8, 2026. https://news.cnyes.com/news/id/6526740
7. Yahoo Finance Hong Kong. "AI Investment Focus Shifts to Real Industries! Goldman Sachs: Capital Expenditure Could Reach $7.6 Trillion Over the Next Six Years." July 7, 2026. https://hk.finance.yahoo.com/news/ai投资重心迈向实体产业-高盛-未来6年资本支出或达7-6兆美元-050904807.html
8. Chun Yi Futures. "Goldman Sachs: China's AI Industry Chain Possesses Unique Competitive Advantages Underestimated by the Market." July 9, 2026. https://www.capitalfutures.com.tw/zh-tw/Financial/BreakingNewsArticle?ContentId=C26070900581