Meta spent more on capex in 2025 than most industrial conglomerates. The financial engine behind that bet is more resilient than critics assume.
Meta spent $69.7 billion in capital expenditures in fiscal 2025. That number, stated with institutional calm in the earnings call, represents something historically unusual: a social media company deploying more capex than ExxonMobil, more than Boeing, more than most industrial conglomerates. The question for shareholders is not whether this is a big number. It obviously is. The question is whether the underlying business can absorb it without destroying the economics that made Meta worth owning.
The answer, based on the actual financial record, is yes, with real caveats. Meta's operating cash flow surged to $115.8B in 2025, up from $91.3B in 2024. Free cash flow fell to $46.1B from $54.1B, entirely because of capex expansion. The revenue engine is performing. The 41.4% operating margin on $201B in revenue is exceptional by any standard.
Meta is not spending recklessly. It is spending from a position of unusual financial strength, betting that AI infrastructure investment will compound into advertising efficiency, creator tools, and eventually new product surfaces. That thesis is not proven. But the financial foundation for making it is stronger than nearly any other company attempting it.
Meta's relationship with capital allocation has been through two distinct phases, and understanding the contrast matters for evaluating the current trajectory.
From 2018 through 2021, Meta ran a classic growth-cash-return machine. It generated enormous free cash flow, bought back stock aggressively, and reinvested modestly in infrastructure. In 2021 alone, Meta returned $44.5B to shareholders through buybacks on $57.7B in operating cash flow, a payout ratio that reflected confidence in the core advertising business.
Then came 2022. The metaverse pivot consumed capital, the advertising market softened, headcount ballooned, and Meta's stock fell roughly 65% from its peak. Operating income collapsed to $28.9B from $46.8B the prior year. Buybacks fell to $28B as the company defended its balance sheet. Critics called it a capital allocation disaster.
The 2023 Year of Efficiency was the correction. Zuckerberg cut 21,000 employees, rationalized the cost structure, and refocused the product roadmap. Operating income recovered to $46.8B on $134.9B in revenue. Free cash flow jumped to $43.8B. The market rewarded the pivot dramatically.
What followed in 2024 and 2025 is a new phase entirely: deliberate, structured investment in AI infrastructure at a scale the company has never attempted. This is not the chaotic metaverse spending. It is concentrated, capex-heavy, and backed by a cash generation machine that has actually continued to accelerate.
TickerXray Report
Get the complete Meta report with all 12 quantitative models, AI-generated investment thesis, and real-time data.
Revenue grew from $116.6B in 2022 to $201B in 2025, a 72% increase in three years. Gross margin held steady at 82% throughout. Operating income reached $83.3B in 2025, up from $69.4B in 2024 and $46.8B in 2023. Net income was $60.5B in 2025 after $62.4B in 2024, with the slight decline attributable to higher taxes and interest costs rather than operational deterioration.
The capex trajectory is the critical variable. Spending was $18.6B in 2021, $31.4B in 2022, $27.3B in 2023, $37.3B in 2024, and $69.7B in 2025. The acceleration from 2024 to 2025 is dramatic: capex nearly doubled in a single year. Management has guided to $60-65B for 2026, suggesting the peak may be near but the elevated level is sustained.
Free cash flow held at $46.1B in 2025 despite the capex surge, supported by $115.8B in operating cash flow. That operating cash flow number deserves emphasis. Very few companies on earth generate over $100B in annual operating cash flow. Meta is in a peer group with Apple, Microsoft, and a handful of others.
Dividends started in 2024 at $5.1B and grew slightly to $5.3B in 2025. Buybacks were $30.1B in 2024 and $26.2B in 2025. The share count has declined from 2.89B in 2020 to approximately 2.19B today, a roughly 24% reduction over five years that has meaningfully increased per-share earnings growth.
Meta's capital allocation framework in 2025 can be summarized as: fund the AI buildout first, return the remainder to shareholders, maintain a lean but adequate balance sheet, and pay a token dividend to attract income-oriented institutions.
The $69.7B capex is concentrated in data centers, servers, and networking infrastructure for AI training and inference. Meta has been explicit that it is building compute capacity ahead of demand, accepting near-term FCF compression in exchange for positioning advantage. This is the same logic that Amazon used when building AWS, and the same logic Nvidia is using when investing in new fab capacity.
The buyback program remains material. At $26.2B in 2025, Meta is still retiring roughly 2% of its float annually at current prices. The share count reduction since 2020 is one of the more underappreciated elements of the Meta investment thesis. Per-share earnings growth has consistently outpaced total earnings growth by 2-3 percentage points annually.
The balance sheet is clean. Total debt was $58.7B at year-end 2025 against $35.9B in cash and $217.2B in equity. The net debt position is manageable relative to the cash generation profile. Meta is not using leverage to fund the AI buildout; it is using its own operating cash flow. That distinction matters for assessing risk.
The dividend, at $5.3B annually, is more symbolic than functional. The yield is under 0.5% at current market cap. But the initiation in 2024 was a signal: Meta's management believes the core business is mature enough to sustain regular cash returns while still funding aggressive growth.
The AI investment thesis at Meta operates on three distinct layers, and they are not equally credible.
The first layer is advertising performance. AI improves ad targeting, creative generation, and auction efficiency. This is not a future benefit; Meta has been quantifying it in earnings calls. Reels monetization improved significantly as AI optimized the feed. Advantage+ AI campaigns have shown measurable ROAS improvements for advertisers. This layer is real, it is generating revenue today, and the capex funds the infrastructure to keep improving it.
The second layer is AI-assisted content creation and discovery. Meta has rolled out AI tools for content creators on Instagram and Facebook, automated content moderation improvements, and conversational AI products integrated across its apps. These tools increase engagement, reduce creator friction, and improve the quality of content that flows through the platform. The monetization timeline is less clear than layer one, but the strategic logic is sound.
The third layer is genuinely speculative: AI as a new product category, autonomous agents, mixed reality assistants, and eventually monetization of Meta AI as a standalone service. Management talks about this layer with genuine conviction. Investors should treat it with appropriate skepticism. New product categories are hard to predict, and Meta's track record in new categories outside of acquisitions is mixed.
The risk in the current capex cycle is mispricing the return timeline. If layers one and two generate the returns management expects, the $69.7B in 2025 capex will look cheap in retrospect. If the AI advertising efficiency gains plateau or competition erodes Meta's targeting advantage, the capex cycle represents a significant misallocation. The financial history provides no clear answer on which outcome is more likely.
Context matters when evaluating AI infrastructure spending. Not every company claiming to invest in AI has the structural justification to do so at scale.
Meta's core business is algorithmic content delivery and auction-based advertising. Both are directly and quantifiably improved by better AI. The company runs some of the largest recommendation systems in the world, serving billions of users across Instagram, Facebook, WhatsApp, and Threads. The infrastructure that serves those recommendations and the AI that optimizes them are the same infrastructure that enables new AI products. The capex is not a separate AI business grafted onto the existing model. It is an upgrade of the existing machine.
This contrasts favorably with companies spending on AI in the hope that it will eventually produce a new revenue stream. Meta's AI investment has a clear, near-term feedback loop: better recommendations lead to more engagement, more engagement leads to higher ad CPMs, higher CPMs lead to more revenue. The 2025 revenue of $201B on a user base that has not grown dramatically reflects the monetization intensity that AI is enabling.
The competitive implications are also meaningful. Alphabet's ad business faces similar dynamics. Amazon's advertising segment is growing. But neither has the same pure social feed optimization advantage. TikTok, which arguably taught Meta how to use AI for content delivery through competitive pressure, represents the most direct competitive threat. The irony is that TikTok's potential US restrictions could be a tailwind for Meta's advertising business.
Meta trades at 22.37x trailing earnings and 14.23x EV/EBITDA as of late March 2026. For a business growing revenue at 22% annually, generating $83B in operating income, and sitting on net debt that is trivial relative to its cash flow, those multiples look optically modest.
The comparison to peer multiples is instructive. Alphabet trades in the mid-20s on earnings despite similar revenue growth. Apple trades at 30x despite slower growth. Microsoft trades at 35x with comparable growth. On a pure earnings multiple basis, Meta appears to be the cheapest of the mega-cap technology compounders.
The caveat is the capex overhang. The 22x trailing multiple is computed against $60.5B in net income that was earned while spending $69.7B in capex. If capex normalizes to a maintenance level of, say, $30B annually, free cash flow would be closer to $85B, implying a FCF yield around 6.4% on the current $1.33T market cap. That is an attractive starting point for a business with Meta's growth characteristics.
The analyst community agrees. Of 68 analysts covering Meta, 47 rate it a strong buy and 13 a buy. The consensus price target of $862.60 implies roughly 38% upside from recent trading levels. The hold and sell ratings are a small minority. Consensus is not always right, but near-unanimous institutional conviction is data worth acknowledging.
The most direct risk is that AI advertising efficiency gains are a one-time step change rather than a continuous improvement curve. If Meta has captured most of the available performance improvement through AI targeting, incremental capex will deliver diminishing returns. The revenue growth in 2025 suggests this is not yet the case, but it is the most plausible scenario in which the current capex level looks like a mistake in retrospect.
Regulatory risk is persistent and underappreciated. Meta operates under consent decrees from the FTC, faces ongoing antitrust scrutiny in the EU, and is a recurring target for privacy enforcement actions. A forced divestiture of Instagram or WhatsApp would materially change the capital allocation calculus. This is a low-probability but high-impact scenario that is not adequately reflected in the current multiple.
The share count reduction story could stall if management prioritizes capex over buybacks. If the $60-65B capex guidance for 2026 holds and operating cash flow does not continue to grow, buybacks could fall further from the 2025 level of $26.2B. Per-share earnings growth would slow as a result, and some of the multiple support would erode.
Stock-based compensation at $20.4B in 2025 is meaningful. It represents roughly 30% of net income and is a real cost to shareholders, not an accounting abstraction. The share count reduction story is partially offset by SBC issuance, and investors should look at net reduction after SBC rather than gross buyback numbers when assessing shareholder value creation.
Meta's $70 billion capex year is either a historic misallocation or the foundation of the next decade of growth. The financial record supports the optimistic interpretation: operating margins held above 41%, operating cash flow hit $115.8B, and the advertising business continued to grow at 22% annually. Companies spending recklessly typically do not produce those numbers.
The investment thesis requires believing that AI-driven advertising performance improvements are durable and continuing, that Meta's platform advantages over emerging social competitors remain intact, and that the infrastructure build does not overshoot actual compute needs by a factor that would be embarrassing in retrospect.
At 22x trailing earnings with $46B in annual free cash flow and a net debt position that is manageable by any measure, Meta is not priced for perfection. It is priced as a mature advertising business, which is almost certainly the wrong frame. The capital allocation decisions of 2024 and 2025 suggest management believes it is building something more durable than a mature advertising business. The question is whether the market will agree before the capex cycle completes.
Full forensic analysis of Meta
+ 6 more models included
150,000+ stocks covered
Global coverage across 60+ exchanges. Every report includes all 12 quantitative models and AI analysis.
View plansEvery report runs 12 quantitative models and generates an AI investment thesis. From Piotroski scores to manipulation detection -- get the full picture in seconds.
12 forensic models
Piotroski, Altman, Beneish, DuPont & more
AI investment thesis
Synthesized outlook on every stock
Manipulation detection
Spot red flags before they hit the news
150,000+ tickers
Global coverage across 60+ exchanges
Expected return
Forward return projections for every stock
Real-time data
Live prices, insider trades, news sentiment
Free accounts get 1 report per month. Pro gets unlimited.
With $83.3 billion in operating income and a PEG ratio of 0.83, Meta has the firepower and the valuation cushion to make its VR bet without risking shareholder value.
Seven consecutive earnings beats, 41% operating margins, and the largest AI infrastructure buildout in big tech. The valuation does not reflect any of it.
Strip out Reality Labs losses and Meta's advertising engine trades at 18x forward earnings — a 20% discount to Alphabet. The AI capex panic created an opportunity.