The stock market is witnessing a stark divergence: technology sectors are repeatedly hitting new highs while consumption sectors remain persistently weak. For some investors, it's a bull market; for others, it's a bear market. Is this extreme polarization a short-term fluctuation or a long-term norm?
CITIC SEC's Chief A-share Strategist Qiu Xiang recently shared his insights on the market outlook.
According to Qiu, A-share investors must completely abandon traditional bull-bear thinking and replace it with a "K-shaped" mindset. The market will exhibit a long-term "barbell structure," with Artificial Intelligence (AI) on one end and supply-constrained assets on the other. For the second half of this year, the other end of the barbell is energy and chemicals.
"The ultimate goal of monetary policy is to serve the real economy, not financial market traders," Qiu stated.
"There will always be two types of capital in the market, and there will always be good and bad investments. We rarely think in terms of 'what bull market is next.' Instead, we allocate around the most prosperous directions that best match investment preferences," he explained.
"In school, we learned about Markowitz's portfolio theory from textbooks, discussing the efficient frontier and balanced allocation. However, balanced allocation might be the least needed model in this era," Qiu remarked.
"The AI theme spends most of its time consolidating or correcting, with only brief periods of rapid ascent. The most important factor is not the direction, but the timing," he noted.
"When AI capital perceives valuations as high, it reduces positions but does not switch to other sectors. For non-AI sectors to rally, new buyers need to enter the market," Qiu emphasized.
"Any business that possesses scarcity has value, even in the silicon-based era. Investors need not be overly anxious," he concluded.
Interest Rate Hike Expectations Likely to Fade as Market Prices in Reality
"Viewing A-shares from a global perspective" is a point Qiu repeatedly stresses.
He pointed out that for the 30 largest manufacturing companies in the A-share market (excluding finance), the proportion of overseas revenue has risen from 5% two decades ago and 20% a decade ago to 46% by the end of last year. "More importantly, over 80% of these companies' profits come from overseas markets," Qiu added.
This means the previous investment logic for A-shares based on the "China economic story"—foreign trade, urbanization, and consumption upgrades—is no longer applicable. "Investors now buy Chinese lithium battery companies because they are the best globally, not simply to invest in China," Qiu said, explaining that many A-share manufacturing firms are essentially globalized operating entities whose fundamentals are directly tied to global real demand.
Consequently, external variables like U.S.-China relations have an unprecedented impact on A-share fundamentals.
Addressing market concerns about Federal Reserve rate hikes, Qiu stated bluntly, "The ultimate goal of monetary policy is to serve the real economy, not financial market traders." He noted the current significant structural divergence in the U.S. economy, with consumer confidence deteriorating while AI-driven industrial investment remains exceptionally strong. "AI requires low financing rates. Overseas cloud providers can no longer support capital expenditures solely with free cash flow and must resort to debt financing. Guiding interest rates higher now would only add extra debt costs for these companies," he explained.
Even if rates were hiked, the damage wouldn't be to the AI industry itself, as leading companies are not sensitive to financing rate increases of 5-6 percentage points, but to households. North American student loan delinquency rates are already rising, and consumer credit indicators continue to worsen. "Hiking rates in this environment does more harm than good. You might suppress headline inflation, but you'd also cripple the economy, because the only prosperous direction is AI," Qiu argued.
He judges that the market's aggressive pricing of rate hikes will likely be corrected. "Especially if the Strait of Hormuz opens soon and gasoline prices fall, by July or August, people will realize the probability of a September or October hike is very low."
K-Shaped Divergence is the New Normal; Use a Barbell Strategy to Navigate Capital Structure Shifts
"People can feel it intuitively: a small number of stocks are rising while most are falling; silicon-based things are going up, carbon-based things are going down," Qiu observed.
He pointed out this divergence has deep macroeconomic roots: over half of U.S. economic growth is driven by AI, and the same is true for South Korea and China. At the A-share level, the combined net profit of the communications, electronics, and non-ferrous metals sectors has grown 64% since the start of 2024, while all other sectors have declined 25% over the same period. "Which stocks are easier to make money on? Those that easily generate profits," Qiu said.
More importantly, capital is also undergoing a K-shaped divergence. Qiu's team has modeled that in the AI era, the pricing gap between physical and mental labor is being flattened, and wealth disparity between individuals will increasingly depend on existing savings. This will foster two extreme investment preferences: aggressive capital chasing thematic tools like AI, and conservative capital flooding into stable products like fixed-income plus, wealth management, and insurance, with the "moderate camp" shrinking.
This determines the market's long-term barbell structure: one end is offensive AI, the other is supply-constrained assets offering stable returns. "From 'AI + dividends' in 2023, to 'AI + resources' in 2024, to this year's 'AI + energy/chemicals,' the barbell structure has persisted. It's just that the other end changes based on the specific supply constraints each year," Qiu explained.
He stated that balanced allocation might be the least needed model in this era. "We rarely think about what bull market is next or whether the market will correct. Instead, we consider that there will always be two types of capital in the market, always good and bad investments, and we only allocate around the most prosperous directions that best match investment preferences."
AI's Three Major Rallies: Timing Trumps Direction
Reviewing the AI theme since 2023, Qiu divides it into three major rallies. The first wave was the imagination of AGI when GPT first emerged, centered on the training-side Scaling Law. The second wave was in early 2025, driven by a surge in Google Token usage, establishing the inference-side logic. The third wave began this February with breakthroughs in Coding Agents and a sharp rise in Anthropic's revenue, opening up the imagination for commercialization.
"The AI theme spends most of its time consolidating or correcting, with only a small portion of the time in rapid ascent," he reminded. Many companies saw tenfold gains during these three waves, but not without 40% to 50% corrections in between. "The most important thing is not the direction, but the timing."
The current market focus is on when Anthropic's Annual Recurring Revenue (ARR) growth will slow, which will determine the rhythm of the current AI hardware cycle playing out like a "cyclical stock" theme. "We estimate the data might remain decent until before its IPO, which could be a阶段性高峰 (stage peak)."
Energy & Chemicals Await a "Tipping Point"; Non-AI Sectors Need New Buyers
For the other end of the barbell, Qiu clearly identifies this year's "stable end" as energy and chemicals, covering new energy, traditional energy, chemicals, and power equipment.
He explained that the supply side faces strict constraints (resource nationalism, dual-carbon and energy consumption controls, anti-internal competition), while demand is relatively stable (electrification, basic necessities). Leading companies enjoy stable competitive landscapes and can provide steady cash returns. "Even without the Middle East conflict, there would be opportunities. China's emphasis last year on 'anti-internal competition' makes chemicals and new energy two of the most important industries this year," Qiu noted.
However, the Middle East crisis led to strait closures and the "physical elimination" of some capacity, creating additional supply-demand gaps. Precisely because oil prices are too high, downstream manufacturers are not restocking, making demand appear weak. "This is actually a temporary disturbance," Qiu judged. Once the strait reopens for navigation, restocking and production will resume, and energy/chemical categories should perform. This sector has already seen significant declines over the past three months, offering attractive valuations, and is only waiting for a "tipping point."
Regarding the启动 (start-up) of non-AI sectors, Qiu believes it's unrealistic to expect capital withdrawn from AI to switch over. "When AI capital perceives valuations as high, it reduces positions but does not switch to other sectors. For non-AI sectors to启动, new buyers need to enter the market." When will these "new buyers" arrive? He provides three signals: the reopening of the Strait of Hormuz for navigation, the Fed ceasing to send hawkish signals, and the completion of the减持 (reduction) phase for broad-based ETFs.
Brokers and Insurance Offer Value; Consumption Still Needs to Wait
Within non-AI sectors, Qiu provided further analysis. For cyclical manufacturing sectors like new energy, chemicals, and non-ferrous metals, their performance isn't poor, but A-share companies trade at a significant discount compared to their overseas peers.
"I think for A-shares, in non-hot sectors, there's always a sense of urgency, a preparedness for danger in times of peace. They preemptively price all potential negative risks into their valuations," Qiu observed. Once the feared tightening and recession are disproven, these stocks should see upward revisions.
Qiu believes that major financial sectors like brokers and insurance are primarily suppressed by capital flows, specifically the ongoing large-scale redemptions of ETFs. "The PB-ROE positioning for brokers is at a historically low point, exceeding the level before September 24, 2024, and approaching that of October 2022." As the减持 phase concludes, the non-bank financial sector might perform during the mid-year reporting season.
The consumption sector currently has relatively weak fundamentals and needs to wait for a clear and strong turnaround in consumption data.
From Building AI to Adapting to AI
At the end of the interview, Qiu shared a story about Kodak and Fujifilm. Two decades ago, digital cameras颠覆 (disrupted) the film industry. Kodak转型 (transformed) to make digital cameras and suffered heavy losses. Fujifilm重新审视 (re-examined) its core competencies, realizing its foundation was chemical technology, and eventually衍生出 (developed) cosmetics, medical imaging, and semiconductor photoresists. Ten years later, Kodak went bankrupt, while Fujifilm's revenue and profits hit record highs.
"This is called转型向内求 (transformation by looking inward)," Qiu said. "Building AI has little to do with most investors and companies, but adapting to AI, and极致放大 (maximizing) one's own inherent稀缺性能力 (scarce capabilities), is the key."
He listed current investment implications: companies making fiberglass and ceramics, unrelated to AI last year, are now connected to the theme; data service providers' 15-20 year trust relationships with clients are hard for Agents to replace; the real bottleneck in North American AI data center construction isn't money, but electricians and construction workers, giving infrastructure service providers extremely high pricing power.
"Any business that possesses scarcity has value, even in the silicon-based era," Qiu concluded. "Investors need not be overly anxious. Moving from building AI to adapting to AI is both an individual choice and an investment mindset."
Comments