Hyperscaler AI Capex Under Pressure as Investors Face a Dilemma
N.R. Finch
Alphabet raised its 2026 AI capex outlook by $15 billion to $190–205 billion; the stock fell as much as 7.4% the next day — investors are asking whether massive AI infrastructure spending can actually deliver returns.
How much was spent, and why did the market punish it?
Alphabet's Q2 earnings lifted the 2026 AI capex forecast by $15 billion, pushing the total to $190–205 billion.
The stock dropped as much as 7.4% the following session. This means → the market isn't rejecting AI itself; it's asking when the money starts coming back.
Some investors draw parallels to the dot-com bubble — companies poured capital into internet infrastructure, early returns were dismal, and share prices eventually collapsed.
Same big spending — so why did Meta rally?
Meta recently signaled plans to lease idle cloud capacity, monetizing existing assets. Its stock climbed as much as 21% over several days.
In plain terms = the market doesn't oppose spending; it opposes spending with no visible payback. Meta showed a path to turn installed hardware into revenue, and sentiment flipped.
This reflects the core divide over hyperscalers right now: the question isn't whether to invest in AI, but whether the capital deployed can generate cash flow soon enough.
What are the bulls saying?
Molly Pieroni, president of Yacktman Asset Management, holds an Alphabet position and explicitly opposes calls to cut capex.
Her logic: "You have to be in the game to be well-positioned at the endgame." This means → she is betting on AI's long-term competitive landscape, not the next quarter's income statement.
She adds that Google's balance sheet is healthy, its business is well-diversified, and this quarter's earnings beat expectations — in short, the company can afford the burn.
What about those buying the beneficiary stocks?
Brandon Nelson, a fund manager at Calamos Investments, continues to hold hyperscaler-capex beneficiary names despite recent sell-offs in the group.
He compares the current volatility to the ten 10% pullbacks the Nasdaq saw between 1996 and its 2000 peak: "Those moments didn't feel good, but the bull market wasn't over."
In plain terms = he sees this as a speed bump in the AI infrastructure build-out cycle, not a trend reversal.
What comes next?
Most hyperscalers are, for now, ignoring investor pushback and sticking to their spending plans.
Microsoft, Meta, and Amazon will report earnings soon — their capex figures and management commentary will be the next key test of market patience.
This means → if those three also ramp spending sharply without showing a monetization path, selling pressure could spread further; if a Meta-style "invest and earn" signal emerges, sentiment may ease.
Content is for reference only, not financial advice.