HSBC told clients on Thursday that fears over hyperscaler overspending have emerged as the prevailing theme in equities tied to artificial intelligence, a development that is driving sector rotations through 2026.
Analyst Alastair Pinder devised a clustering model to detect which AI narrative is dominating investor behavior across global equities. The tool tracks return patterns among hyperscalers, semiconductor companies, software firms and China internet names to determine the prevailing regime.
The bank said market internals and equity returns have swung sharply this year as investors "jump between competing AI narratives." HSBC pointed to the launch of Moonshot's Kimi K3 and recent reports of lithography competition from mainland China as recent catalysts that have influenced the clustering outcomes.
HSBC's framework divides the market into five distinct regimes. The scenario with the highest assigned probability - labeled "hyperscaler overspend" - carries a 37% chance of prevailing. Under that regime, data center and semiconductor stocks outperform while so-called capex spenders lag. HSBC quantified the effect, noting annualized returns of 12.3% for tech hardware and 11.8% for semiconductors in this state, reinforcing investor shifts into Korea and Taiwan.
Another large-probability outcome is what HSBC calls "AI positioning capitulation," given a 26% probability. In this regime, crowded AI-related trades unwind and rotate into defensive sectors such as pharmaceuticals and biotech, with semiconductors falling 14.3% according to the model's outputs.
"China competition concerns" is the third scenario, assigned a 20% probability. HSBC said the Kimi K3 launch has reawakened this narrative, which tends to channel flows into mainland China media and consumer services.
The remaining two regimes were assigned smaller odds: "AI disruption fears" at 9% and "AI euphoria" at 8%. In the broad risk-on "AI euphoria" state, HSBC's model shows the AI supply chain leading performance and semiconductors gaining 19.1%.
HSBC noted the framework's utility in mapping how emerging-market equities perform and rotate as the dominant narrative shifts. By assigning probabilities and estimating sector returns across regimes, the clustering approach offers a way to assess which parts of the market stand to benefit or suffer as investor focus moves between competing AI stories.
What this means
- Investors are actively rotating between thematic narratives tied to AI, producing sharp swings in sector performance.
- Under the leading "hyperscaler overspend" scenario, data center and semiconductor equities are modeled to outperform, favoring markets such as Korea and Taiwan.
- Alternative narratives can drive flows into defensives or into mainland China consumer and media, depending on which theme gains traction.