NextFin News - HSBC strategist Max Kettner is arguing that the market’s newest leadership rotation may not be a rotation at all. In his view, the hyperscalers that spent months looking expensive and vulnerable are moving back into favor just as the broader equity tape melts higher, helped by lower rate expectations, improving risk appetite and a still-intact AI capital-spending cycle.
The setup matters because the debate around mega-cap technology has shifted from whether companies can justify their spending to whether investors have already decided the answer is yes. One HSBC market note indicated the bank remains overweight on U.S. equities and expects policy rates to stay steady through 2026 and 2027, while a separate public HSBC snippet said Kettner expects further broadening in equity markets amid lower rates and renewed economic optimism. That is a friendlier backdrop for the largest cloud and AI infrastructure buyers, whose stocks had recently been treated as if the spending boom were a future problem rather than a present opportunity.
That view is arriving at a useful moment for the hyperscaler trade. The largest cloud platforms are still committing extraordinary sums to data centers, chips and networking gear, but the capital intensity is now being framed less as a margin threat and more as the price of staying relevant in a market where AI demand is still outrunning supply. A secondary market estimate cited combined hyperscale capital spending at roughly $620 billion in 2026, while Alphabet alone reported $35.7 billion of capex in the first quarter and lifted its full-year 2026 guidance to $180 billion-$190 billion.
The result is a classic late-cycle market paradox: investors can worry about the bill while still bidding up the companies writing it. That is what makes Kettner’s call worth attention. If the market is entering a melt-up phase, then the stocks with the deepest balance sheets, strongest earnings power and clearest AI monetization story tend to regain leadership first. If the market loses altitude, those same names become the quickest way for investors to de-risk. The hyperscaler trade is therefore less a simple beta call than a referendum on whether the AI capex boom is still being rewarded or is beginning to be punished.
Why hyperscalers can look expensive and still regain favor
The central point in Kettner’s argument is that valuation alone does not explain leadership when liquidity and earnings expectations are changing at the same time. In a market that is willing to pay up for growth, the biggest platforms can appear to be both the problem and the solution: their spending depresses near-term free cash flow, but it also reinforces the notion that AI infrastructure demand remains structural, not cyclical.
That distinction matters because investors have spent much of 2026 asking whether hyperscaler capital expenditure is a sign of discipline or desperation. The answer, at least for now, depends on who is spending and what the spending buys. Alphabet’s reported capex of $35.7 billion in the first quarter, along with its raised 2026 guidance of $180 billion-$190 billion, was read by the market as an affirmation that the company still sees room to monetize cloud and AI workloads at scale. The same logic applies, with different nuances, to Microsoft, Amazon and Meta: each has a different balance between infrastructure build-out, cloud revenue and advertising or software monetization, but all four are treated by investors as proxies for the state of AI demand.
That is why the hyperscaler group can recover even after a period of skepticism. When risk appetite improves, investors often move back toward the names that combine size, profitability and visibility. Those are also the companies most able to absorb a capital-spending wave without immediately jeopardizing balance sheets. In that sense, the market’s melt-up impulse can favor the very stocks that looked most overowned when sentiment was more cautious.
The flip side is that the trade is fragile. If macro conditions tighten, or if investors decide the AI build-out is overshooting demand, hyperscalers can quickly become the market’s easiest source of disappointment. Their scale helps them on the way up and exposes them on the way down. That asymmetry is exactly why Kettner’s view is useful: it frames hyperscalers not as a permanently preferred group, but as the group most likely to regain favor when the market’s tone turns indiscriminately optimistic.
The capex debate is still the core of the story
What keeps this trade alive is not simply momentum. It is the market’s willingness to believe that the AI infrastructure cycle still has years left to run. As long as that belief survives, heavy spending can be interpreted as evidence of demand rather than waste. A projected $620 billion of combined hyperscaler spending in 2026 is a staggering figure, but it is also the kind of number that can sustain an investment narrative when the underlying revenues are large enough to absorb it.
The deeper reason investors are tolerating that spending is that the largest platforms are not financing a speculative moonshot from a weak balance sheet. They are using established cash engines to fund a new layer of compute demand. That matters because the market is far more forgiving when spending is attached to an identifiable customer base and a visible strategic logic. Cloud demand, enterprise software workloads, search monetization, digital advertising and AI inference all provide plausible ways to earn back the outlay.
Still, the market is not giving hyperscalers a blank check. The debate has shifted toward timing. Investors are no longer asking only whether AI capex will pay off; they are asking when it will pay off and whether returns will show up fast enough to defend today’s multiples. That is the key reason hyperscalers can move in and out of favor so quickly. In a melt-up, timing is less important and optionality gets rewarded. In a more skeptical tape, deferred returns become a penalty.
There is also a composition issue inside the group. Alphabet and Microsoft are typically viewed as higher-quality AI franchises because their cloud and software ecosystems already provide a route to monetization. Amazon’s case is more tied to cloud utilization and infrastructure density. Meta’s spending narrative is more complicated because the company is using AI both to enhance its existing advertising machine and to expand new capabilities. Those differences matter for stock selection, but they do not change the broader point: the market is again willing to pay for the promise that hyperscaler capex can turn into durable earnings power.
Kettner has said HSBC remains overweight on U.S. equities and expects further broadening in equity markets amid lower rates and renewed economic optimism.
That is a constructive read, but it is not a call that the AI trade has become risk-free. It is a call that the market is once again rewarding scale, liquidity and credible growth stories. For hyperscalers, that is enough to bring buyers back.
What Would Break the Thesis
The simplest way to undermine the hyperscaler rebound is for earnings or guidance to disappoint on the one thing that matters most right now: the conversion of spending into revenue. If capex keeps rising faster than monetization, the market will eventually stop treating infrastructure build-out as a growth signal and start treating it as a drag on returns.
That risk is not theoretical. The whole hyperscaler trade rests on a chain of assumptions: that AI demand remains strong, that cloud capacity remains scarce enough to justify expansion, and that the incremental dollars spent on chips, data centers and power will ultimately earn a return above the company’s cost of capital. Break any part of that chain and the stock reaction can change quickly.
Macro could also matter more than the company-specific story. If rates stop drifting lower, or if investors start to worry that the economy is slowing too much to support current earnings expectations, the melt-up logic weakens. Lower rates help the long-duration growth profile of the hyperscalers; higher real yields do the opposite. That is why the HSBC framing is important: it links the resurgence in hyperscalers not only to AI spending but also to a friendlier monetary backdrop.
For now, though, the market is inclined to give the group the benefit of the doubt. The largest technology platforms still have the strongest balance sheets, the deepest liquidity and the clearest route to turning compute demand into revenue. In a tape where investors are chasing the next source of upside, that combination is hard to ignore.
The bottom line is that hyperscalers are not back because the capex question has been answered. They are back because, in a market built on optimism, the biggest sellers of AI infrastructure remain the cleanest way to express it. If the melt-up continues, they should keep finding bids. If it stalls, they will likely be the first place investors notice the difference.
Explore more exclusive insights at nextfin.ai.
