Morgan Stanley Report: Meta’s Cloud Plan Focuses on Leasing Idle Compute Power; Price Target Raised to $775

Written by: Rita

Tide Guide

Bloomberg recently cited reports that Meta is planning a cloud computing business encompassing model hosting APIs and bare-metal computing rental. Morgan Stanley’s assessment on July 1 was straightforward: the portion of this plan focused on renting out idle computing power is far more viable than a full-fledged cloud service competing with AWS, as the former does not require large-scale hiring or an enterprise sales team, while the latter carries significantly higher execution risk.

More importantly, the numbers: leasing 250 MW of computing power at $40 per watt for one year could boost 2028 earnings per share by approximately 8%; scaling up to 1,000 MW would increase this boost to 33%. However, Morgan Stanley made it clear that its overweight rating on Meta is not based on optimism about its cloud business; the core rationale supporting its $775 price target remains structural improvements in efficiency and user engagement.

The rumored Meta cloud initiative falls under the newly established Meta Compute department, set up in January this year, and consists of two components: one is a model hosting API service for developers, similar to AWS Bedrock, covering models such as Muse and Spark; the other is closer to bare-metal compute rental services. Meta has not yet commented officially on this.

The report notes that the model hosting API business demands higher standards in technology, hiring, and execution. Meta’s Muse series models performed modestly on key benchmarks such as TerminalBench and SWE Bench Verified, which measure programming and third-party API invocation capabilities, indicating a significant gap behind frontier models like Gemini. Additionally, Meta lacks the mature enterprise sales teams that AWS, Azure, and GCP possess. Morgan Stanley believes that a comprehensive API service combining models and applications is more of a self-validation effort, carrying significantly higher risks than the well-established offerings of mature cloud providers. In contrast, briefly renting out idle computing resources requires no large-scale hiring or new team formation, representing a much lower-barrier path.

Morgan Stanley estimates that Meta’s owned computing power will increase from 1.4 gigawatts in 2025 to 1.9 gigawatts in 2026 and 3.4 gigawatts in 2027. For comparison, Amazon and Google are each expected to add 5 gigawatts and 9 gigawatts of computing power respectively in 2027 alone—this is precisely why Meta theoretically has room to lease out capacity. The report estimates that Meta currently leases approximately 2.5 gigawatts of computing power from third parties such as Coreweave, Nebius, GCP, and Oracle; by 2026, this third-party capacity will bring Meta’s total available capacity to 4.5 gigawatts, while by 2027, the third-party portion will shrink to 0.6 gigawatts, resulting in a total capacity of approximately 4.0 gigawatts. Morgan Stanley believes that Meta cannot re-lease the third-party computing power it rents, but this also indicates the company retains greater flexibility in its computing resource allocation, creating the potential to temporarily lease out its owned capacity.

Morgan Stanley’s sensitivity analysis shows that leasing 250 MW of computing power at $40 per watt for one year could boost 2028 earnings per share by approximately $3, or nearly 8%. If the leased capacity increases to 1,000 MW, the upside could reach $11.88, equivalent to a 33% upside potential. The higher the price and the larger the scale, the more pronounced the elasticity—even at a lower rate of $20 per watt, leasing 250 MW could still add approximately $1.49, or 4%, to earnings.

Morgan Stanley’s current model assumes Meta’s capital expenditures will rise from $145 billion in 2026 to $175 billion in 2027 and $205 billion in 2028, corresponding to approximately 3.5 gigawatts of additional computing power in 2027, assuming this capacity is primarily used for Meta’s own operations rather than building a full-scale cloud services platform. The report notes that if Meta truly scales its business of renting out computing power, capital expenditures could be further revised upward.

Viewed within the broader industry context, Morgan Stanley estimates that combined capital expenditures by cloud providers and emerging cloud companies will grow from $246 billion in 2024 and $433 billion in 2025 to $834 billion in 2026 and $1.2 trillion in 2027. For 2027, the breakdown is approximately: Amazon at $225 billion, Google at $350 billion, Meta at $175 billion, Microsoft at $276 billion, Oracle at $108 billion, Coreweave at $41 billion, and Nebius at $31 billion.

The report notes that Morgan Stanley has assigned Meta a "Buy" rating, primarily based on the company's structural shift toward long-term user engagement and efficiency improvements, with cloud computing itself not being the core rationale for the rating. Leasing computing power serves more as a transitional earnings buffer; the true valuation support hinges on whether new products such as Meta AI, business agents, private messaging, and diffusion models can sustainably scale, and whether subscription revenue can unlock a new growth trajectory. The target price of $775 corresponds to a 23.1x multiple on the expected 2027 EPS of $34.16, representing approximately a 37.6% upside from the June 30 closing price of $563.29. The bull-case target is $1,000 (28x multiple), while the bear-case target is $450 (14x multiple). The report also highlights that Meta currently trades at a roughly 35% discount to Google’s P/E ratio, approximately two standard deviations below its long-term average and near multi-year lows.

The weakest link in Morgan Stanley’s calculation is assuming that renting out computing power is a temporary business that Meta will consistently and stably execute over the long term. Moreover, Bloomberg’s report has not yet been confirmed by Meta’s official channels, and key variables such as rental pricing, lease duration, and specific customers remain hypothetical. No matter how detailed the sensitivity analysis, the underlying assumptions have not been substantiated. For investors, a more significant consideration is the hidden thread of capital expenditures: if Meta truly transforms cloud computing into a legitimate business—rather than merely renting out idle computing power—Morgan Stanley’s current estimate of $175 billion to $205 billion in capital expenditures will likely be exceeded. This would simultaneously impact expectations for free cash flow and the market’s assessment of whether Meta’s valuation discount can narrow.

Disclaimer

This article is a compilation and interpretation by Chaoxiang Research of third-party brokerage research reports. The ratings, price targets, earnings forecasts, and related judgments cited herein are the views of the brokerage's analysts and represent only the position of their respective institutions; they do not reflect the views of Chaoxiang Research nor constitute any investment advice.

The market carries risks; make decisions independently. This article should not be used as a basis for buying or selling any securities.