Meta’s push into cloud computing means Wall Street has to prepare for lower margins

Meta CEO Mark Zuckerberg appears poised to make a big bet on a potentially big market, but one that commands much slimmer margins than his company's dominant online ad business. Cloud infrastructure has proven to be highly lucrative for hyperscaler peers Amazon, Microsoft and Google, and Zuckerberg has hinted of late that Meta could be headed in that direction. On Wednesday, CNBC's Jim Cramer confirmed that Meta will sell excess computing power to outside customers. The company is debating whether to offer access to AI models hosted on its infrastructure or to sell access to raw computing power, according to Bloomberg. Wall Street welcomed the news. After slumping for the past year, Meta's stock started the third quarter with a bang, jumping 9% on Wednesday for its sharpest rally in more than five months. Investors have been looking for Meta to diversify its business and monetize its multi-hundred-billion-dollar investment in advanced data centers and artificial intelligence infrastructure. "Making this as a revenue stream has been part of their road map," said Karan Ramchandani, managing director at advisory firm Post Oak Group. "It seems like a no-brainer to compete in the market, to sell compute power to other B2B players." At Meta's annual shareholder meeting in May, Zuckerberg said a potential cloud computing business is "definitely on the table." And seven months earlier, on an earnings call, Zuckerberg said companies are regularly "asking if we have compute that they could buy from us at some premium to what we've bought it at." Just before Wednesday's rally, Meta's stock closed out its fourth straight quarterly drop. In April, Meta boosted the high end of its 2026 capital expenditures guidance by $10 billion to $145 billion. Some of that is getting funded through debt, with the company raising $25 billion from a bond sale just as it was reporting first-quarter earnings. "I think that this is a response to complaints that the company may be overspending and skepticism that Meta will ever earn a commensurate return on its capex," said Paul Meeks, head of technology research at Freedom Capital Markets, regarding Meta's push into cloud. "The problem with this company is that it only builds, or only thus far, capacity for itself, and it's not really monetizing any AI apps yet." Almost all the financial benefits of Meta's AI spending to date have been recognized in the company's core advertising business, which has seen dramatically improved targeting capabilities and has offered a wider suite of creative tools to marketers. Meta still gets 98% of its revenue from digital ads. Zuckerberg has been trying to change the narrative, with cloud being perhaps the most ambitious new effort. Meta shares jumped almost 4% in May, when the company announced various paid subscription plans for Instagram, Facebook and WhatsApp in addition to two subscription services for its Meta AI app and website. Meta declined to comment for this story. Not trying to be AWS Cloud infrastructure is a particularly valuable commodity as the generative AI boom nears its fourth anniversary, and is something that very few companies can afford to provide at scale. The U.S. leaders are Amazon Web Services, Microsoft Azure and Google Cloud, which have all built large businesses by allowing companies to offload their computing requirements. Mark Mahaney, an analyst at Evercore, said it's unlikely that Meta will try to challenge those hyperscalers. Rather, Mahaney sees Meta following in the footsteps of so-called neoclouds such as CoreWeave and Nebius, which offer access to AI-specific computing products like Nvidia chips and systems. Shares of CoreWeave and Nebius both suffered double-digit drops on Wednesday following the Meta report. Mahaney said Meta may have been motivated in part by Elon Musk's SpaceX. The company, which owns xAI, has recently signed deals to offer capacity to Google and Anthropic amounting to more than $2 billion in combined revenue a month as well as to startup Reflection AI. Brian Schechter, a partner at Primary Venture Partners, also made the comparison to SpaceX. He said the companies are similar in that they've spent billions of dollars training big AI models on top of their own infrastructure. Both companies "failed to bring to market an AI model that drove huge customer traction," Schechter said. "Being able to monetize their compute after a missed training run shows how compute can function more like a commodity." One area of concern for some investors will be the potential hit to Meta's profitability. Selling cloud services typically requires building a big enterprise sales and support team, and the margins can't match what Meta generates from ads. Meta's gross margin of 82% is among the highest in the tech industry, and the company recorded an operating margin of 41% in the latest quarter. Google provides a glimpse of what's to come. Google's services business, which mostly comes from ads, notched an operating margin of 42% in the first quarter, while for cloud the margin was 18%. It took many years just to get there. The company launched its cloud infrastructure business in 2008 and made it generally available in 2011. In 2020, Google started disclosing financials, and didn't record a profit until the first quarter of 2023. Meeks said that while Meta "probably has one of the most glorious business models in tech," anything it enters outside of online ads "would be dilutive to their business and would lower their margins from their glory days." "As a Meta shareholder, I'd rather see them continue with open models and monetize AI through products and services with much higher margins than get into the brutal battle of building data centers in places like North Dakota," Meeks said. WATCH: Meta building out cloud business is 'a really smart pivot,' says Evercore ISI's Mark Mahaney.