CITIC Securities has released a research report stating that the strengthening US dollar and rising interest rate hike expectations in early May coincided with an acceleration in K-shaped divergence across global markets. The core issue is the damage that tightening expectations inflict on demand in sectors outside artificial intelligence.

As the market moves forward, the K-shaped divergence has reached a temporary extreme. Even within the overseas technology sector, the focus is narrowing, while the pricing of stocks, bonds, commodities, and currencies is already showing early signs of recessionary trading. If actual tightening follows, it could further damage demand in the traditional "carbon-based" economy. Conversely, the K-shaped divergence may see a temporary convergence.

Compared to the volatile overseas markets, A-shares are demonstrating greater resilience. There are signs that some left-side capital is beginning to explore certain non-AI sectors. Among these, a few undervalued segments already have a foundation for recovery, only awaiting a catalyst.

CITIC Securities' key views are as follows:

Stage One: The Early May Acceleration Point

If we compare US dollar liquidity expectations with the K-shaped divergence in equity markets, this year's divergence can be divided into three phases. The first phase, from the start of the year to late February, saw lingering expectations for Federal Reserve easing, a weakening US dollar index, and minimal K-shaped divergence across markets. The second phase, from late February to April, involved a hawkish market repricing, a rebounding dollar, and the beginning of pronounced divergence. The third phase, from May to the present, features further heated rate hike expectations, the dollar index hitting a yearly high, and intensified K-shaped divergence in non-US markets, creating a clear gap with US equities. This third phase also corresponds to a period of significant pullback in A-share non-AI sectors and cyclical stocks beginning to underperform their overseas counterparts.

Stage Two: Divergence at an Extreme, Even Within Tech

Market narratives and capital flows can periodically push reasonable performance divergences to extremes of overpricing. The essence is that as tightening expectations intensify, the market's threshold for growth prospects also rises. At the market level, this is reflected even in US stocks, where the technology sector itself is seeing a narrowing focus. Since May, perceptions of slowing growth at Anthropic and downstream corporate controls on AI spending have led to a consolidation in the Nasdaq and a sustained correction in the Magnificent Seven, down 12% from their peak as of June 26. Previously high-flying optical communication sectors began high volatility in mid-May, while application software sectors that had rebounded earlier came under renewed pressure in June. The US semiconductor sector performed well initially, but by June its index also entered a period of high volatility, falling 7.9% in a single week recently. Currently, only memory chip leaders continue to rise on strong earnings reports.

Stage Three: Early Recessionary Trading and Potential Demand Damage

The recent price action in overseas risk assets shows a combination of a stronger US dollar, falling US stocks, and declining commodities—including oil. Falling oil prices have eased long-term inflation expectations but have not impacted short-term inflation stickiness. The latest US PCE data for May showed core inflation at a new high since 2023, indicating highly sticky inflation not solely driven by oil. The market narrative has shifted to suggest that robust AI demand is crowding out commodity resources, forcing the "carbon-based" economy to pay higher prices, causing inflation, and compelling the Fed to hike rates. News of Apple raising prices due to memory cost increases reinforces this narrative. The market interprets that if the Fed is forced to hike to combat this supply-driven structural inflation, it could ultimately further damage demand in the traditional economy. Meanwhile, rising US real rates are seen as a result of AI's competitive advantage driving strong economic growth and dollar strength. However, the rapidly flattening US Treasury yield curve suggests this narrative is temporary. The flattening indicates the market may be reflecting that tightening could further damage traditional economic demand and, ultimately, long-term growth expectations. Overall, the current pricing of risk assets is highly contradictory, likely because the growth effects of AI innovation remain confined to a small cycle of hardware, model, and cloud companies, not yet integrated into the broader economic cycle. This implies that while K-shaped divergence has its rationale, it also possesses inherent fragility.

Stage Four: A-Shares' Resilience and Early Non-AI Interest

Recently, A-share technology sectors have shown significantly greater resilience compared to their overseas counterparts, with domestic computing power chains particularly exhibiting independent strength. While A-share companies linked to overseas markets follow the same pricing logic for cyclical, capital-intensive stocks as their foreign peers, domestic chains, under the narrative of import substitution and self-reliance, are also commanding a typical growth stock premium. These two different pricing frameworks are operating simultaneously. Beyond technology, some non-AI sectors, having undergone sufficient prior adjustment, are beginning to show characteristics of incremental left-side capital involvement. Brokerages and chemicals are typical examples. The combined trading volume share of brokerages and chemicals hit a yearly high recently, with significant volume increases. On some days, they have risen alongside AI sectors, and their stock prices have shown notable recovery. For the brokerage sector, the left-side logic is supported by low valuations, easing selling pressure from funds, and the catalyst of a tech listing boom. For chemicals, the core driver is the widening of price spreads as oil prices fall into a "sweet spot," coupled with some companies delivering strong, "as-expected" mid-year profit forecasts, leading to very strong stock performance—possibly indicating low market positioning and limited intense speculation. While it remains rare for non-AI sectors to rise against the capital draw of AI, it at least suggests considerable sidelined capital in the A-share market is watching for opportunities in non-AI sectors, though marginal changes are not yet sufficient to drive broader entry.

Stage Five: Undervalued Non-AI Sectors Poised for Recovery

Since May, A-share non-AI sectors have underperformed their overseas peers more significantly, having already priced in many negative expectations such as demand recession, monetary tightening, and Middle East peace process setbacks. They now offer relative value and have a foundation for recovery, awaiting positive changes in their own narratives. Such changes could come from an unexpected drop in oil prices post-strait reopening lowering inflation expectations, or a synchronized recovery in global non-AI industrial production and social activity. In the current environment of a strong dollar and heated rate hike expectations, stock selection needs to be more precise and patient. After all, without major marginal changes, the recovery of weaker sectors will not be smooth and may even experience synchronized adjustments if leading sectors correct.

Regarding specific allocation, maintaining a structure focused on AI and energy/chemicals is still advised. On the AI side,看好 memory, gas turbines, diesel generator sets, semiconductor equipment, and materials. Within the energy/chemicals sphere, for new energy,看好 performance delivery in electrolyte & additives and separators. For traditional chemicals, falling average oil prices and volatility are driving restocking and operational demand, while peaking macro liquidity expectations present a potential future inflection point. Currently,看好 sectors with significant cost reduction potential, relatively inelastic demand, and low valuations, such as refrigerants, phosphate chemicals, spandex, dyes, and large-scale refining. For metals,推荐 those with some AI exposure but whose valuations are temporarily suppressed by the macro/hike narrative, such as computing power metals like tin, copper, and some AI-related minor metals (e.g., tungsten). Additionally, continuing to recommend increasing allocation to undervalued brokerages, as current headwinds like liquidity pressure may gradually fade in the second half, with mid-year profit previews also serving as a potential catalyst.