Meta cloud plan divides analysts as AI spending scrutiny rises

Meta Platforms’ consideration of a cloud business has sharpened investor debate over whether its AI infrastructure spending can produce adequate returns. The shares rose nearly 9% last Wednesday after Meta confirmed to CNBC’s Jim Cramer that a long-discussed cloud service was being developed, though CNBC reported the stock was a few dollars below that closing level a week later.

Bloomberg, which first reported the plan, said Meta is discussing whether to sell access to artificial intelligence models running on its own systems or to offer raw computing capacity. Either approach would move the Facebook and Instagram owner closer to the business model used by Amazon Web Services, Microsoft Azure and Google Cloud, which rent computing resources to outside customers while also supporting their own operations.

The plan comes as investors scrutinize Meta’s capital expenditure budget. CNBC reported that the company now expects fiscal 2026 capex of $125 billion to $145 billion, up from a previous range of $115 billion to $135 billion. FactSet data cited by CNBC showed analysts had expected $122.64 billion, below even the low end of the revised range.

Public cloud economics could give Meta another way to monetize data-center capacity. Large cloud providers typically reserve computing power for internal needs and sell surplus capacity to external customers. In AI data centers, capacity is often discussed in gigawatts because power availability constrains how much computing equipment can run continuously.

JPMorgan estimated, according to CNBC, that each gigawatt of Meta compute capacity offered through a cloud business could generate $20 billion in annual revenue and add several dollars to earnings per share. Meta’s fiscal 2025 revenue rose 22% to almost $201 billion, while earnings per share slipped 1.6% to $23.49, CNBC reported.

The bearish view is that a cloud push may indicate Meta has built more AI infrastructure than it needs. Needham analyst Laura Martin wrote Monday that Meta is entering the cloud market because it overbuilt capacity and will not require all the compute implied by its 2026 capex plan, according to CNBC. Martin’s team also said the company would face entrenched, well-financed competitors, including AWS, Google Cloud and Microsoft Azure.

Needham also raised a margin concern. The firm said Meta could be shifting some emphasis from its core advertising business, which it described as carrying a 70% margin, toward cloud services, which it described as a 35% margin business.

The bullish case is that a cloud unit could give Meta a financial hedge if internal AI demand falls short of available capacity. CNBC reported that Meta has said it currently needs all the compute it can obtain. A cloud offering could also broaden demand across outside companies and industries if Meta opens part of its infrastructure to customers.

Canaccord Genuity analysts said in a Monday note cited by CNBC that the negative case on Meta has “gone too far.” They pointed to an accelerating advertising business, emerging subscription tiers and market prices for capacity as reasons Meta’s valuation discount to other large technology stocks looked harder to justify.

Meta has continued to announce AI-related products and partnerships. CNBC cited Tuesday’s release of Muse Image, an AI image-generation model aimed at creators and advertisers, along with lower-cost smart glasses, an enterprise business tool, plans for a prediction-markets app and a partnership with Qualcomm.

Meta’s stock performance has lagged several major cloud and AI peers this year, according to CNBC. The shares were down more than 8% year to date, compared with a nearly 5% gain for Amazon and a gain of more than 15% for Alphabet. Microsoft was down about 20%. CNBC also reported that Meta traded at 17.7 times next-12-month earnings estimates, below Amazon at 25.5 times, Microsoft at 19.6 times and Alphabet at 24.9 times.

This story draws on original reporting from CNBC.