Markets split on AI capex
Investors are torn between signs that AI already generates contracted demand and worries that hyperscaler capital spending could be excessive. Microsoft reported stronger-than-expected results and a large computing backlog that underpins continued AI investment while some investors remain skeptical about the returns on massive capex plans ( ), and critics like Michael Burry have continued to flag concern about overstretched AI spending in related firms (economictimes.indiatimes.com).
Investors have split on artificial intelligence spending as Microsoft reports contracted demand building faster than some skeptics expected. (microsoft.com) Microsoft said on January 28, 2026 that quarterly revenue rose 17% to $81.3 billion and Microsoft Cloud revenue reached $51.5 billion. The company also said commercial remaining performance obligation, a backlog measure for signed but unrecognized revenue, jumped 110% to $625 billion. (microsoft.com) Three months earlier, Microsoft had reported commercial remaining performance obligation of nearly $400 billion, up more than 50% year over year, with a weighted average duration of two years. Chief financial officer Amy Hood said roughly half of that quarter’s spending went to short-lived assets such as graphics processing units and central processing units. (microsoft.com) A backlog is a queue of signed orders that has not yet turned into booked sales, like a contractor’s calendar filling up before the work is finished. Microsoft is pointing investors to that queue to show that data-center spending is tied to customer commitments rather than only to internal forecasts. (microsoft.com) The same argument is showing up across the sector. Alphabet said on February 4, 2026 that Google Cloud revenue was running at more than $70 billion annually and that its backlog grew 55% quarter over quarter to $240 billion, driven by demand for artificial intelligence products. (abc.xyz) The spending case still has critics. Michael Burry’s Scion Asset Management disclosed in a May 15, 2025 filing that it held put options tied to Nvidia, a major supplier to the artificial intelligence buildout, while the filing also warned that reported put values can differ from the manager’s actual economic exposure because Form 13F shows positions on a gross basis. (sec.gov) That caveat leaves room for interpretation, but the filing still showed Burry positioned for downside in a company tied closely to artificial intelligence infrastructure. His view has become a shorthand for the bearish case that hyperscalers could overbuild capacity before profits catch up. (sec.gov) Meta has also shown how expensive that buildout has become. In its January 28, 2026 results, the company reported 2025 costs and expenses rose 24% to $117.7 billion even as annual revenue climbed 22% to $201.0 billion. (investor.atmeta.com) The market divide now turns on a simple question: whether signed demand and new revenue can keep outrunning the cost of chips, power, and data centers. Microsoft and Alphabet are telling investors the order books already say yes, while skeptics are still waiting for that spending to look restrained as well as profitable. (microsoft.com) (abc.xyz)
