Morgan Stanley maintains META Overweight: layoff cost cuts offset AI capex, with upside from Neocloud backup value

Overview

This report updates the model around META's cost structure adjustments and AI infrastructure investment. Morgan Stanley incorporates the expected 10% layoffs, slower future hiring, and higher capex into forecasts, but believes cost savings can offset the rise in D&A and infrastructure costs, so it maintains 2027 EPS at about $34, a $775 target price, and an Overweight rating. The report also introduces an important but non-base backup scenario: META's large-scale compute buildout has replaceable commercialization value, and if part of that compute is leased to third parties in the future, it could provide a Neocloud safety net.

Core views

The core views have three parts. First, layoffs and slower hiring reflect a structural efficiency shift extending Meta's "Year of Efficiency" framework, which should support future opex and FCF performance. Second, product innovation in AI, recommendation algorithms, Reels monetization, ad attribution, and Meta AI remains the core basis for the Overweight rating; the platform's 3.5 billion+ daily active users and roughly 40/60 minutes of daily usage on Facebook/Instagram provide a durable monetization base. Third, although management has not made compute leasing part of the base plan, in a scarce-compute environment META's 4.4/3.9 GW of new effective capacity could be repriced by the market, creating potential 2028 EPS upside of 8%+.

Analysis framework

The analysis approach centers on earnings model revisions, cost savings estimates, capex/D&A sensitivity, DCF and long-term EBITDA multiple valuation, and risk-reward scenarios. The report first estimates one-time layoff charges and annualized opex savings, then incorporates higher AI infrastructure capex, cloud partnerships, and D&A into the cost-revenue model, and finally uses 2027 EPS, P/E multiples, DCF assumptions, and bull/base/bear scenarios to assess the reasonableness of the target price.

Methodology notes

• Valuation methodsDCF/long-term EBITDA multiple

  Target price valuation

  The $775 target price is determined by discounted cash flow and a long-term EBITDA multiple; the DCF uses about 8% WACC and about 3% perpetual growth, and the target price implies roughly 23x 2027 P/E.

• Scenario analysisRisk Reward

  Bull/base/bear price scenarios

  The report lays out a $1,000 bull case, a $775 base target, and a $450 bear case to frame the potential stock-price distribution under product innovation, efficiency gains, AI investment returns, macro, and regulatory risks.

• Cost modelHeadcount cost model

  Layoff savings and one-time charge estimate

  Assumes 10% layoffs, or about 8,000 people, with one-time charges estimated at about $100,000 per person based on historical restructuring experience, and annualized savings estimated at about $450,000 per person in opex.

• Optionality analysisNeocloud safety net

  Backup value from compute leasing

  Although not the base case, the report views META's AI infrastructure as a scarce and substitutable compute asset that could add more than 8% upside to 2028 EPS if leased to third parties for $10-15 billion a year.

Asset mapping & comparison

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

• Meta Platforms Inc (META.US)

  Covered company and primary recommended name

  Strengths

  Has more than 3.5 billion daily active users, strong distribution and data assets, and AI recommendation and advertising tools that can improve engagement, attribution, and monetization; layoffs reinforce a cost discipline culture; AI compute assets also have potential leasing optionality.

  Weaknesses

  AI/data center capex and D&A are rising, Reality Labs losses may still weigh on profits, and the company does not have a public cloud business to directly absorb all of the compute.

  Comparison

  The report notes that META's additional effective compute in 2026/2027 is close to other hyperscalers, but the market currently assigns that buildout very little or even negative value.

  Risks

  Macro pressure on advertising, regulatory limits on ad targeting, Reels monetization ramping more slowly than expected, data center execution missteps leading to higher long-term capital intensity, and wider Reality Labs losses.

• META compute capacity / Neocloud optionality

  Non-base source of backup value

  Strengths

  Compute is scarce; if leased to third parties it could generate $10-15 billion of revenue per year, and even the low-end scenario could add 2028 EPS upside of 8%+.

  Weaknesses

  This is not management's current base intent, and there is uncertainty around commercialization path, customer demand, pricing, and execution capability.

  Comparison

  Rather than viewing it purely as a cost center, the report treats AI infrastructure as a substitutable, multi-use asset.

  Risks

  Compute supply or leasing prices could fall short of expectations, contract execution and utilization could be insufficient, or infrastructure buildout could overrun budgets or be delayed.

Key data

• RatingOverweight / Top PickMorgan Stanley maintains a positive rating.
• Target price$775.00Implies about 25% upside, or roughly 23x 2027 P/E.
• Current price$618.43From the current price point in the risk-reward chart.
• 2027 EPSabout $34After layoff savings largely offset higher investment, the 2027 EPS estimate is broadly unchanged.
• Expected layoff scale10%, or about 8,000 peopleThe report builds this assumption into the model, with effects expected in May 2026.
• One-time chargeabout $800mnBased on the historical experience of roughly $100,000 per person during the 2022-2023 Year of Efficiency.
• opex savingsabout $2bn in 2H26, about $3.5bn in 2027Estimated using average opex of about $450,000 per affected employee.
• Larger layoff rumorup to about 16,000 peopleIf true, 2027 could see about $3.5bn more savings and about $1.20 of EPS; the company has not commented.
• Hiring growth adjustment2027 net growth reduced from about 5% to about 3.5%Corresponds to new hires falling from about 4,200 to 2,600, and reflects reports of about 6,000 positions being closed.
• capex adjustment2026/2027 raised by 7%/6%, or about $10bn eachReflects higher guidance and bottom-up GW capacity build estimates.
• D&A impact2027 D&A increases by about 9%Higher AI infrastructure capex drives depreciation and amortization higher.
• Effective new computeabout 4.4/3.9 GW in 2026/2027Including hyperscale deals, at a scale close to other hyperscalers.
• Neocloud backup revenue potentialabout $10-15bn annuallyA non-base backup scenario of leasing compute to third parties.
• Neocloud EPS upside8%+ EPS upside in 2028Even the low-end leasing scenario could contribute upside of this magnitude.
• User base3.5 billion+ daily active usersSupports ad distribution, data, and GenAI product rollout.
• Usage timeabout 40/60 minutes daily on Facebook/InstagramThe report believes algorithm and product improvements can still lift engagement and monetization.

Impact & implications

The investment implication is that the market may be underestimating META's ability to simultaneously drive cost reduction, product innovation, and AI infrastructure assetization. In the near term, layoffs and slower hiring should help cushion EPS from the pressure of capex and D&A growth; in the medium term, Reels, ad measurement, Meta AI, agentic tools, and click-to-message can continue to support revenue growth; and in the long term, large-scale compute buildout may still reflect asset value through a Neocloud backup solution even if it is not fully consumed by internal business use. However, if data center execution falters, capital intensity keeps rising, or ad growth slows, the target price and FCF outlook would come under pressure.

Risks

• Macro pressure or weaker consumer spending could make ad demand come in below expectations.
• Regulatory restrictions could limit META's ad targeting, data usage, or attribution capabilities.
• Reels engagement and monetization could ramp more slowly than expected, reducing growth and valuation certainty.
• Reality Labs losses could widen further, weighing on operating profit and free cash flow.
• AI/data center execution missteps could make long-term capital intensity higher than expected.
• Additional opex or capex could pressure operating profit growth and FCF.
• The layoff scale reported in the media has not been confirmed by the company; if the actual scale, costs, or savings are below model assumptions, earnings improvement may fall short of expectations.

What to watch

• New product updates at Meta Conversations (June 3) and Meta Connect (September 23-24), especially Meta AI, Muse, and agentic products.
• The actual layoff scale, one-time charges, savings realization pace, and management's response to the rumors.
• Whether hiring slowdowns continue, especially whether 2027 net headcount growth falls below 3.5% after about 6,000 positions are closed.
• Reels engagement, ad monetization, improvements in ad measurement/attribution, and progress on click-to-message.
• Changes in 2026/2027 capex, D&A, hyperscaler deals, and the cost-revenue model.
• Whether AI infrastructure shows signs of third-party leasing, partnerships, or other Neocloud commercialization signals.
• The impact of macro conditions, regulation, and Reality Labs losses on profit and FCF.