The Chart Showing Meta's Ad Revenue Crossover With Google Changes The Valuation Math

The trajectory demonstrates that the Reality Labs concerns that dominated the FY2022 narrative have been fully resolved at the revenue line. Ad impressions, ad pricing, and AI-driven engagement have each independently grown. The fact that no single lever has been responsible for the acceleration makes the growth more durable than the market is modelling.

The 16 percentage point expansion in operating margin across the Year of Efficiency and the subsequent AI-driven monetisation layer produces roughly $33 billion of incremental operating profit that was not available three years ago. That incremental profit is what finances the aggressive capex programme and the ongoing Reality Labs investments simultaneously.

At $70 billion of annual capex, Meta is spending more than Alphabet on AI infrastructure and nearly as much as Microsoft. The capex absorbs roughly 34% of revenue, which is an exceptional investment rate for a software-centric business. The question for the valuation is whether this capex cycle has a four-year payback or a seven-year payback. Sell-side models that assume four years produce $1,100 fair value targets; models that assume seven years produce $600 targets.

If FY2026 free cash flow rebounds above $60 billion (consensus $54 billion), the market treats the FY2025 dip as a capex timing issue and the multiple holds. If FY2026 comes in below $40 billion, the market treats the capex programme as permanent margin erosion and the multiple compresses toward 18-20x forward earnings. The difference between those two scenarios is approximately $400 per share. Rarely does a single fiscal year carry this much weight in a mega-cap valuation.

A business maintaining 82% gross margins while tripling capex is signalling that the AI infrastructure is coming out of depreciation and power rather than direct cost of revenue. That is the specific characteristic that supports the longer-payback capex thesis. Depreciation expense is rising but the underlying monetisation of each ad impression has risen alongside it. That is a better mix than pure capex-for-capex investment.

Alphabet's Google Search advertising revenue is estimated at $235-240 billion for FY2026 consensus. Meta at $243 billion is the narrowest margin of crossover possible. The two companies have fundamentally different structures: Google monetises intent with algorithmic efficiency, while Meta monetises attention with engagement-driven AI ranking. The crossover matters because it resets the narrative around who is winning the next decade of advertising. Historically, being the largest player in a category conveys roughly two to three turns of revenue multiple premium. Meta currently trades at 8.7x forward revenue; Alphabet trades at 8.1x. If the crossover is durable, that premium should compress away from Alphabet and toward Meta over the subsequent eighteen months.

Net income growth stalled in FY2025 despite operating income growing 20%. Higher depreciation from the capex cycle plus higher R&D expense fully consumed the operating income growth at the net income line. If that dynamic continues for another year, EPS growth stalls and the forward P/E expands mechanically. This is the specific data point the Valuation Desk is watching ahead of the Q1 2026 print.

The closest historical parallel is Amazon from 2016 to 2020. AWS was consuming capex at a rate the market repeatedly modelled as unsustainable, and AWS margins were being suppressed by the investment cycle. The market spent those five years alternately pricing AWS as a $200 billion business and a $600 billion business depending on the latest quarterly capex number. Retrospectively, the actual answer was between the two extremes and closer to the upper end. Meta's AI capex cycle has similar characteristics. The payback will likely arrive, the timing is uncertain, and the market will trade back and forth on individual quarterly datapoints for at least another two years. Patience is the differentiator.

Meta's average revenue per user across its family of apps hit approximately $54 in FY2025, up from $46 in FY2024 and $39 in FY2023. That compound growth rate of 17% is above both the underlying advertising market growth and the user growth across the family. The implication is that monetisation per impression is improving faster than either of those variables. AI-driven ad ranking is the primary mechanism. Advantage+ campaigns now represent over 30% of ad spend on the platform and continue to outperform manual campaigns by roughly 22% on return on ad spend. That is a meaningful share shift in real time.

The data story here is that Meta has converted its user base from a volume lever to a monetisation lever. The question is how much more monetisation is available. Instagram Reels now monetises at approximately 75% of Feed rates, up from 55% eighteen months ago. If Reels reaches parity with Feed, that alone is worth another 8-10% to total advertising revenue. Threads is early but showing early monetisation signals. The compound monetisation lift is what supports the high-case valuation scenario.

Reality Labs lost approximately $18-19 billion in FY2025, broadly stable to FY2024's $17 billion loss. Management has not guided to a specific break-even timeline. The segment produces roughly $2.5 billion in revenue against $21 billion in operating expense. The gap between those numbers is the clearest negative lever in the Meta valuation model.

Strip Reality Labs losses entirely out of Meta's FY2025 operating income and the company would have produced $102 billion in operating profit rather than $83 billion. That is a 23% haircut to the bull-case earnings power. The valuation question is whether the Reality Labs investment produces a franchise worth $18-19 billion annually or zero. The spread between those outcomes is wide.

Historically, the market has largely ignored Reality Labs losses when the core advertising business is growing above 18%. The market has priced them punitively when growth slows below 10%. The current growth rate is 22%. As long as that holds, Reality Labs is a tolerable investment overhang. If it breaks below 15%, Reality Labs becomes the lightning rod.

Meta repurchased roughly $34 billion of stock in FY2025 at an average price near $580. Dividends totalled approximately $5 billion, introduced in FY2024 and now growing. Combined capital return of nearly $40 billion represents roughly 87% of free cash flow. That capital return cadence is meaningful but leaves limited slack for balance sheet growth or M&A. Cash and equivalents sit at roughly $68 billion against $29 billion of debt, giving net cash of $39 billion.

The share count has compressed by approximately 8% over the past three years, which has supported per-share earnings growth. If the buyback cadence slows to finance capex, per-share growth compresses accordingly. That is a quiet lever in the valuation that most models do not stress-test. Meta's capital return durability depends on free cash flow durability. Circular, but true.

The Valuation Desk has modelled three distinct paths for the forward multiple. Path one: capex payback materialises cleanly by FY2027, free cash flow rebuilds to $70-80 billion annually, and the forward multiple holds at 23-25x earnings. Fair value in this path is $740-820 per share. Path two: capex plateaus rather than compresses, free cash flow sits in the $50-55 billion range, and the forward multiple compresses to 20-22x earnings. Fair value in this path is $560-640 per share. Path three: capex continues growing into FY2027, free cash flow compresses further, and the market begins treating Meta like a cyclical capex story rather than a software growth story. Fair value in this path is $440-510 per share.

The probability distribution across these three paths is approximately 40%, 45%, and 15%. The expected fair value, weighted across paths, lands at $635-680 per share. That is tight around the current share price. Meta is at fair value, not clearly cheap or expensive, and the next twelve months of data will tell us which path the business is actually on. Positioning should reflect the bounded upside rather than the long-tail opportunity.

Relative to Alphabet, Microsoft, and Amazon, Meta has the highest capex-to-revenue ratio of the group at 35%. Amazon is at 12%, Microsoft at 22%, Alphabet at 20%. The intensity of Meta's investment cycle is a multiple of peers. That intensity either demonstrates conviction or demonstrates panic depending on how the numbers resolve.

The counterweight is that Meta has the highest operating margin of the group at 41%. Alphabet is at 32%, Microsoft at 44%, Amazon at 11%. On a normalised basis, Meta generates more operating profit per dollar of revenue than Alphabet does, which means its capex intensity can be sustained longer than a lower-margin business could sustain. The combined picture (high intensity plus high margin) is unique in the mega-cap space.

By comparison, Microsoft's AI capex intensity is also high but the Azure cloud business produces near-immediate revenue conversion from each dollar of capex. Meta's capex is primarily training infrastructure for AI models that are consumer-facing rather than enterprise-revenue-generating. The commercial payback path is less linear. That is the risk the Valuation Desk weighs most heavily.

The single most valuable data point over the next three quarters will be the ratio of capex growth to free cash flow growth. In FY2025, capex grew 87% year over year while free cash flow declined 15%. That ratio is unsustainable. If FY2026 brings capex growth of 20-30% combined with free cash flow growth of 15-20%, the payback thesis holds. If capex grows another 50%-plus while free cash flow stays flat, the payback thesis is dead and the multiple compresses.

The ratio is a cleaner read than any single metric in isolation. It normalises for the dollar magnitude of each variable and captures the directional relationship that matters for the valuation. We will be tracking this ratio quarterly and updating the fair value range accordingly. Today, the Valuation Desk is in the middle of the expected range at a neutral stance. A clean Q1 print that demonstrates capex deceleration combined with free cash flow growth would shift us to constructive.