Meta Shares Rise About 9-10% After Bloomberg Report That It Plans to Sell AI Computing Capacity to Outside Customers

Meta Platforms stock posted its biggest single-day gain of the year on Wednesday, July 1, 2026, after Bloomberg News reported that the company is developing a cloud business to sell outside customers access to its artificial-intelligence computing capacity and its in-house AI models ^[1]. Shares jumped roughly 9% to 10%, adding on the order of $149 billion in market value in a single session ^[2][4]. The report described a unit reportedly called "Meta Compute" that would let businesses rent Meta's surplus AI infrastructure and, potentially, tap its proprietary models ^[1][13]. Meta itself has not formally confirmed the plans, and multiple outlets note the effort remains early-stage and dependent on exactly how much spare capacity the company actually has ^[2][3].

The timing mattered. Before the news broke, Meta shares had been trailing the broader market for 2026, down roughly 5% to 7% for the year even as the S&P 500 and Nasdaq Composite posted solid gains ^[1][2]. A single, unconfirmed report was enough to erase much of that underperformance in one trading day, a sign of how hungry investors were for evidence that Meta's enormous AI spending could eventually generate its own revenue stream rather than simply drain cash ^[2].

The Numbers Nobody Disputes

Certain facts are not in question. Bloomberg broke the story on July 1, and it was quickly corroborated by CNBC, Reuters, and other outlets ^[1][2][3]. Meta's stock rose in the 9%-to-10% range that day, its largest one-day move of 2026, adding roughly $149 billion in market capitalization ^[2][4]. The rally came on the heels of a rough stretch: Meta had lagged the S&P 500 and Nasdaq for the year, down an estimated 5% to 7% before the jump ^[1][2].

The move also had losers. Shares of CoreWeave and Nebius, two specialized "neocloud" providers that rent out AI computing power, fell sharply — reports put the declines in the 13%-to-15% range — even though Meta has existing capacity agreements with both companies: roughly $21 billion with CoreWeave and up to $27 billion over five years with Nebius ^[11]. Separately, Meta's own financial disclosures are a matter of public record: in April 2026 the company raised its 2026 capital-expenditure guidance to a range of $125 billion to $145 billion, up from an earlier $115 billion to $135 billion range, after spending $72.2 billion on capex in 2025 ^[7][15]. At Meta's May 27 shareholder meeting, CEO Mark Zuckerberg said a cloud computing business was "definitely on the table" ^[9], and in May the company began cutting about 8,000 jobs, which it linked to its AI-infrastructure spending ^[8][16].

The Pressure Underneath the Headline

Strip away the day's stock-price theatrics and a structural tension remains. Meta has physically built enormous AI computing capacity and is the only one of the four major U.S. hyperscalers — alongside Amazon, Microsoft, and Google — without an existing cloud services business ^[1][2]. Once data centers and chips are already paid for, selling idle capacity is a high-margin proposition, which is precisely what makes Meta a credible threat to companies whose entire business model depends on renting out compute ^[2].

At the same time, Meta faces real pressure to justify the spending itself. The company needs a visible return path on capital expenditures that could reach $145 billion this year and a reported $600 billion cumulative buildout through 2028, or risk investor unease over a projected steep drop in free cash flow ^[7][16]. Meanwhile, incumbent cloud giants have spent years embedding themselves in enterprise software, security certifications, and long-standing sales relationships — a moat that raw GPU pricing alone does not breach quickly ^[1][2]. Meta's underlying infrastructure and its contracts with CoreWeave and Nebius remain in place regardless of how the cloud story is ultimately packaged, and the company's cash generation is under genuine strain no matter which narrative prevails ^[1][11][15].

How Each Side Sees It

Meta's own framing, as relayed by Zuckerberg and echoed in bullish coverage, casts the plan as smart monetization of infrastructure the company would have built anyway: it has already constructed world-class AI capacity for its own products, so renting out the surplus turns a sunk cost into high-margin revenue at low marginal cost — much as Amazon converted its internal computing needs into Amazon Web Services two decades ago ^[1][9]. Zuckerberg has said outside companies ask "almost every week" to buy compute or model access at a premium, suggesting the demand is real and already inbound ^[9]. Meta's incentive is straightforward: answer investor doubts about its spending, diversify beyond advertising, and reframe capex as a future asset rather than a drain ^[7][9].

The incumbent hyperscalers — Amazon Web Services, Microsoft Azure, and Google Cloud — see a more limited threat. Their case is that selling raw GPU capacity is the easy part; what enterprise customers actually buy is a deeply integrated software stack, security compliance, and sales relationships built over years, none of which Meta has in the enterprise market, where its brand is tied to social media and advertising rather than business infrastructure ^[1][2]. Their incentive is to protect high-margin cloud franchises and reassure their own shareholders that a new entrant will not quickly erode pricing, and most analysts view the near-term threat to them as modest ^[3][11].

The neocloud specialists — CoreWeave, Nebius, and IREN — have the most direct stake. Their argument is that their business is purpose-built for exactly this market, and that they already hold large, multiyear contracts, including sizable ones with Meta itself ^[11]. Analyst opinion here is split and explicitly contested: Roth Capital has called the sharp selloff in these stocks an overreaction to a plan that remains unconfirmed and capacity-gated, while Bernstein analysts counter that hyperscaler competition was always inevitable and is genuinely "problematic" for the pure-play firms ^[12]. Their incentive is to defend valuations built on the assumption that Meta would remain a customer, not become a rival, and the roughly 13%-to-15% single-day decline in CoreWeave and Nebius shares shows how much is riding on that question ^[11][19].

A fourth camp — spending skeptics and Meta's own workforce — reads the same announcement as evidence of a problem, not a solution. Their argument is that Meta calling its compute "excess" implies the company may have overbuilt, and that a projected sharp drop in free cash flow, combined with roughly 8,000 layoffs the company itself tied to AI spending, suggests the cloud pivot is closer to damage control than bold strategy ^[7][8][16]. Their incentive is financial discipline and scrutiny of job cuts justified by spending whose payoff remains unproven ^[8][16]. For now, the stock's jump shows markets siding with the monetization story, but if returns disappoint, this camp's comparisons to the dot-com bubble era stand to gain force ^[7].

How the Coverage Split

Bloomberg, as the outlet that broke the story, struck the most measured tone overall, though even it led with the ambition of the plan — "planning a cloud business" — before noting the effort remains early and unconfirmed ^[1]. CNBC's market-desk coverage centered the stock's dramatic "pops 9%" move, generally balanced but foregrounding the bullish price reaction over the plan's unconfirmed status ^[2].

Right-leaning U.S. business coverage, exemplified by Fox Business, tied the cloud push to Meta's earlier layoffs and Zuckerberg's refusal to rule out further cuts, emphasizing costs borne by workers and pressure on management to justify the spending rather than the technology's promise ^[8]. Center-left U.S. outlets such as Fortune and NPR leaned more skeptical of return on investment, highlighting the projected collapse in free cash flow and invoking dot-com bubble comparisons, often foregrounding Zuckerberg's vaguer answers on returns, such as his description of an ROI question as "very technical" ^[7][16]. TechCrunch's SpaceX analogy — "turn excess AI compute into cash" — carried a similarly skeptical undertone, subtly implying the buildout may have been oversized [and reported elsewhere].

Non-Western outlets based in India and Hong Kong, including Eastern Herald and the South China Morning Post, generally treated the story as a market and competitive-strategy development, naming the Meta Compute unit and its leadership and focusing on economic and business angles rather than U.S. domestic politics or job losses ^[13]. Some of that coverage, however, stated unconfirmed details — such as the "Meta Compute" name and a reported "Muse Spark" model — with more certainty than Meta itself has offered, presenting the plans as more settled than they currently are ^[13]. Across the spectrum, one thread runs through nearly all of it: a substantial share of the day's roughly $149 billion market-value swing rested on a report Meta had not confirmed.