Meta's potential cloud and compute monetization could increase revenue, but J.P. Morgan favors internalizing compute in Meta's own AI products

Overview

This report discusses Meta's potential strategy to commercialize cloud infrastructure, model access, and raw compute for third-party customers. The context is Bloomberg reporting that Meta is preparing a cloud infrastructure business plan, including charging developers for managed model access and renting raw compute in a neocloud-like model. J.P. Morgan believes this path could deliver incremental revenue and attractive AI infrastructure returns for Meta, but the firm's preference remains for Meta to consume compute through its own products and large user base rather than primarily selling infrastructure externally.

Core views

The core views are: first, external monetization of AI infrastructure has meaningful financial upside for Meta, with the report estimating that 1GW of capacity could generate about $20 billion in annual revenue and contribute several dollars of EPS; second, this revenue would complement Meta's existing AI product efforts around business agents, subscriptions, and Meta Glasses; third, compared with selling compute, J.P. Morgan prefers Meta to develop core AI products and scale inference demand across its roughly 4 billion users; fourth, if Meta begins selling compute more suitable for inference, it may indicate that AI product pull outside advertising remains limited; fifth, the future capex path is still unclear, and Meta is expected to continue investing in frontier models and AI products within Family of Apps.

Analysis framework

The report uses an event-driven and scenario-based modeling approach: based on media coverage and management comments from the annual shareholder meeting and 1Q26 earnings call, it combines assumptions on 1GW revenue and EPS contribution to assess the commercialization value of external cloud and compute sales, capex implications, and the strategic signal for Meta's AI product roadmap.

Methodology notes

• Scenario modelingAI Infrastructure Monetization Scenarios

  1GW compute revenue potential

  The report uses the assumption that 1GW of capacity could generate about $20 billion in annual revenue and add several dollars of EPS to size the potential financial return of selling compute and model access externally.

• Strategic comparisonComparing External Compute Sales vs Internal Product Utilization

  Compute utilization priority

  The report compares selling infrastructure to third parties versus using compute within Meta's own AI products, concluding the latter better aligns with long-term product strategy and monetization through its user base.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• META.US

  Covered security

  Strengths

  Meta has an estimated 4 billion-user base, a Family of Apps ecosystem, frontier model investments, and large-scale AI infrastructure; if it sells compute, 1GW of capacity could produce about $20 billion in annual revenue.

  Weaknesses

  The report describes AI product traction outside advertising as still limited, and high capital spending imposes stricter requirements for return realization.

  Comparison

  Like other large infrastructure players, Meta can recover part of its investment through external infrastructure sales, but J.P. Morgan still prefers Meta to use compute for its own core AI products rather than adopting a neocloud-style rental model.

  Risks

  Ongoing increases in capex, insufficient internal AI product absorption, uncertainty around external compute demand and pricing, and a potential mismatch between model and inference demand structures.

Key data

• Monday price reaction+9%Bloomberg reported that META shares rose on Monday after media coverage that Meta was considering cloud infrastructure and compute monetization plans.
• Current price$612.91Price date is 2026-07-01.
• RatingNeutralThe report cover page shows that J.P. Morgan rates META as Neutral.
• Latest historical target price$725The rating history table shows an N rating and a target price of $725 for 2026-04-30.
• Potential revenue estimateabout $20B/yearThe report estimates 1GW of capacity could produce around $20 billion in annual revenue.
• 2027 capex forecast$202BJ.P. Morgan currently forecasts Meta's 2027 capex at $202 billion.
• User baseabout 4B usersThe report emphasizes that Meta can deploy core AI products across a user base of roughly 4 billion.
• Completion time2026-07-02 01:08 AM EDTBoth completion and publication time of the report are stated as this time.

Impact & implications

If Meta advances cloud and compute monetization, it could strengthen near-term market confidence in AI infrastructure returns and provide a partial recovery path for large capital outlays; over the medium to long term, however, the implications are more complex because external compute leasing may suggest that internal AI product demand is not yet sufficient to fully absorb capacity. For investors, the key is not merely whether compute can be sold, but whether Meta can convert AI capabilities into high-quality proprietary product revenue, improved ad efficiency, and durable monetization within its user ecosystem.

Risks

• The future capex path is unclear, and Meta may continue to allocate substantial capital to frontier models and AI products.
• If external compute sales become a major focus, it may reflect limited AI product demand or traction outside advertising.
• Revenue, margins, customer demand, and competitive dynamics for external cloud and compute offerings remain uncertain.
• Training and inference compute structures may differ, and leasing some inference compute may not reduce high spending related to training.
• The research house discloses potential conflicts related to market making, client relationships, investment banking, and compensation involving Meta or related entities.

What to watch

• Whether Meta formally launches model access, APIs, or raw compute rental services for third parties.
• The release cadence and commercialization model of Muse Spark and subsequent frontier model versions.
• User adoption and revenue contribution of AI products in business agents, subscriptions, Meta Glasses, and Family of Apps.
• Whether 2027 and beyond capex guidance converges toward or diverges from J.P. Morgan's $202B forecast.
• Whether management continues to emphasize internal compute demand first, or shifts toward more aggressive external monetization.