HSBC Says AI's Fear Trade Is Now Overspending-And $380B of Capex Is the Tension

HSBC's dominant AI narrative is now who pays, not whether AI matters

The AI trade has shifted. The market is less focused on whether AI is real than on who is funding it. HSBC says the prevailing regime is now hyperscaler overspend, with the firm's model assigning it a 37% probability. For investors, that makes the debate less abstract and more practical: where are relative returns more likely to come from as sentiment keeps rotating?

The market is split between payers and suppliers

This is where positioning gets complicated. Bulls still look at $380 billion in capex this year from Alphabet, Meta, Microsoft, and Amazon as evidence of conviction. Bears focus on the concern that current revenue profiles may not justify such massive spending. In simple terms, the market is dividing between the companies paying for AI infrastructure and the companies selling it.

That split has portfolio consequences. In HSBC's overspend framework, data center and semiconductor names are positioned to outperform at the expense of the capex spenders. If fear of excessive spending increases, that leadership gap could narrow around suppliers. If the spending starts to generate stronger returns, the gap could close just as quickly.

Why overspending became the leading AI narrative

The narrative matters because it gives investors a testable reason to be uneasy now. HSBC says concern around hyperscaler overspend has emerged as the dominant narrative as investors keep jumping between competing AI narratives through 2026. The immediate concern is straightforward: current revenue profiles may not justify such massive spending, while monetization and productivity gains are still several years out. That does not prove AI demand is weak. It highlights a timing tension that is easier to see in prices than a long-term bull case.

Why the market often favors the sellers

Investors naturally anchor to the clearest number available. Right now, that number is current monetization versus balance-sheet commitment. If returns are still lagging spending, the buyers can look exposed even though they are funding future capability rather than today's finished product.

That helps explain why suppliers can look more defensible in the moment. Their cash flows tend to arrive earlier in the cycle, so in a fear-driven rotation, the market often rewards the sellers of infrastructure before it rewards the payers.

What would confirm or weaken the bias

Watch three signals:

• Narrative rotation: Do fresh catalysts keep reinforcing the same relative-performance pattern, with suppliers leading and the spenders lagging?
• Supplier leadership: Does rotation stay concentrated in the visible capacity chain instead of broadening back into the original capex payers?
• Payer relative performance: Do the spenders stabilize on relative charts as investors give more weight to longer-duration returns?

The weakening case is not that AI is imaginary. It is that the capital cycle is being rewarded too early: heavy spending first, broader transformation later. If price action begins to reflect that sequence more fairly, the overspend bias would be losing control.

What HSBC's regime implies for market leadership

The practical question is where the market wants to make money next. If HSBC's hyperscaler overspend regime remains dominant, the edge likely stays with the sellers of infrastructure rather than the buyers building it.

Winners, losers, and the rotation into hardware and semis

In this regime, HSBC sees data center and semiconductor names outperform at the expense of the capex spenders. The expected returns are meaningful: tech hardware is seen at an annualized 12.3%, and semiconductors at 11.8%. That helps explain the market's willingness to favor exposures tied to Korea and Taiwan.

Treat that as regime evidence, not a permanent forecast. If the overspend narrative still has the strongest hold, capital is more likely to favor visible capacity over delayed payoff.

Why the bull and bear cases still diverge

The bull case is simple: in a market driven by narrative rotation, owning the current leader can be easier than trying to call the reversal. HSBC says investors are jumping between competing AI narratives, which suggests leadership can shift quickly-but it also means the market usually trades the theme that feels most immediate.

The bear case is more uncomfortable. Current revenue profiles may not justify such massive spending. The payers are still committing $380 billion this year, while the payoff is still years away. Bulls read that as patience. Bears read it as overreaction. For now, price is showing more patience toward suppliers than toward hyperscalers.

What would change the read

The overspend thesis is only useful if investors know what would weaken it. The key question is whether the market continues to reward the supply chain more than the spenders. If it does, the regime likely still has control. If the payers begin to regain relative strength, the narrative is probably losing influence.

Positioning risk cuts both ways. If the regime holds, staying out of hardware and semis could mean weaker relative returns for longer. If the regime breaks, waiting for confirmation could also prove costly.

Sentiment is now judging who pays, not what AI becomes

The more important tell is not whether AI will matter long term. The tell is that the market is judging the story through the balance-sheet commitment it can see now, not the outcome it still cannot.

HSBC says investors are jumping between competing AI narratives, and the current favorite is hyperscaler overspend. That is a classic sentiment shift: when fear rises, investors stop focusing on what AI could become and start auditing who funds it. In that frame, heavy capex looks less like proof of vision and more like a relative-value burden, while suppliers look safer because their returns arrive earlier in the cycle.

That is the core setup here. This is not a verdict on AI demand. It is a relative-value debate centered on payer-versus-supplier positioning, with the market currently giving more credibility to earlier cash flows than to deferred ones.