Is AppLovin's 76% Operating Margin Real? The Filing Says Half Is an Artifact

How durable is a 76% operating margin when the stock compensation pool behind half the improvement has already collapsed 83%? AppLovin, the AI-powered mobile advertising platform that grew revenue 70% to $5.5 billion in FY2025, posted the highest operating margin among all major tech peers. But decomposing the 10-K line by line shows that roughly half of the 34-percentage-point margin expansion since FY2023 traces to a collapsing stock compensation pool, not operational efficiency.

The other half is real. AXON, AppLovin's machine learning engine that matches advertisers with app publishers through real-time auctions, does generate genuine operating leverage — cloud infrastructure costs grew 28% against 70% revenue growth, and the agent revenue model means incremental ad spend flows through at near-zero marginal cost. But when you separate the SBC runoff from the structural improvement, the forward margin picture changes materially.

This matters because the company has already embedded the mechanism for SBC to return. In October 2025, AppLovin granted 920,526 performance stock units to key engineering employees with vesting milestones ranging from $300 billion to $1 trillion in market cap. At the current $228 billion valuation, the first trigger is 32% away. And on the revenue side, 97% of FY2025 growth came from pricing — not volume — while the filing's own risk factors warn that growth will decelerate. The margin story and the growth story are both less durable than the headline numbers suggest.

The Pricing Power Speed Trap

AppLovin's growth story underwent a fundamental inversion between FY2024 and FY2025 that the headline 70% revenue figure obscures. In FY2024, the company grew by expanding its reach — installation volume increased 50% while net revenue per installation rose 22%. In FY2025, the model flipped entirely: volume grew just 3% while pricing surged 72%.

"For the twelve months ended December 31, 2025, our revenue increased by $2.3 billion, or 70%, from the prior year period primarily due to improved Axon Ads Manager performance, where the volume of installations increased 3% and net revenue per installation increased 72% compared to the prior year period."

The math is stark: 72% pricing growth divided by the combined 75% (3% volume + 72% pricing) means approximately 96% of FY2025 revenue growth came from pricing alone. This is not a growth model — it is a price extraction model.

The bull case is that AXON 2's AI improvements justify these higher yields — advertisers see better ROI per dollar spent, so they pay more willingly. That interpretation has merit: if the engine is genuinely delivering 72% more value per installation to advertisers, the pricing increase reflects quality improvement, not rent extraction.

But the filing itself pushes back against that optimism. In the risk factors section, management explicitly warns investors not to extrapolate: the company "may not be able to sustain [its] growth rate in the future" due to "more difficult comparisons to prior periods and the saturation of the market." When the company's own lawyers write "saturation" into the risk factors alongside a 3% volume growth print, that word choice is not accidental.

Q1 2026 guidance reinforces the deceleration: the $1.745–$1.775 billion revenue midpoint implies approximately 30% year-over-year growth on a continuing-operations basis, down from 70% in FY2025. The sequential growth of 5–7% is healthy, but the year-over-year trajectory confirms that the pricing-led model cannot sustain 70% growth rates indefinitely.

AppLovin's FY2025 revenue growth of 70% was driven almost entirely by pricing — net revenue per installation rose 72% while install volume grew just 3%, a fundamental inversion from FY2024's volume-led model. For investors, the single most important leading indicator is now volume growth: if it stays below 10% for two more quarters, the pricing-only model is confirmed as structural rather than temporary.

The Margin Illusion — SBC, PSUs, and the True Cost Structure

AppLovin's operating margin trajectory from 42% in FY2023 to 59% in FY2024 to 76% in FY2025 is the headline that drives the bull thesis. But decomposing the improvement line by line reveals that approximately half traces to stock-based compensation declining as the RSU pool depleted — not to genuine operating efficiency gains.

The numbers tell the story clearly. S&M expense fell $49.2 million year-over-year, but $42.8 million of that decline (87%) came from SBC dropping from $76.8 million to $34.1 million. R&D expense fell $148.2 million, but $115.1 million (78%) came from SBC dropping from $229.6 million to $114.5 million. Combined, $157.9 million in SBC decline across these two lines accounts for 80% of their $197.4 million total decrease.

Why did SBC collapse? Not because of compensation discipline — because the RSU pool ran out. Outstanding RSUs fell 83% from 2,150,021 to 358,162 units, with a 19% forfeiture rate accelerating the depletion. Only 239,382 new RSUs were granted in FY2025. The company was simply running off legacy equity grants faster than it was issuing new ones.

Meanwhile, cash operating expenses moved in the opposite direction. COGS increased 28% ($144.5 million) from network infrastructure scaling. G&A increased 42% ($68.6 million), driven partly by $31.6 million in professional services related to the Apps divestiture and partly by the $24.5 million bad debt increase discussed below. The "asset-light AI platform" narrative requires qualification when cash costs are rising at these rates.

The forward SBC picture is what transforms this analysis from academic to investment-relevant. The $489 million in unvested RSUs at a $578 weighted average grant price will amortize at roughly $250 million per year over a 1.95-year recognition period. And then there are the PSUs.

"These PSUs vest upon the achievement of specified market capitalization milestones, including an initial milestone of $300.0 billion and, with respect to certain PSUs, additional milestones up to $1.0 trillion, based on the Company's market capitalization over any consecutive trading days during a 7-year performance period from the grant date."

These 920,526 PSUs, granted in October 2025 to key non-executive engineering employees, carry a total grant-date fair value of approximately $410 million (920,526 units at $445.89 each, valued using 70.95% expected volatility). If the $300 billion milestone is achieved — requiring just 32% appreciation from the current $228 billion market cap — PSU expense begins flowing through the income statement. The compensation architecture is designed so that success increases costs.

Modeling forward SBC at $310 million per year (RSU amortization plus partial PSU vesting) takes the SBC/revenue ratio from 3.9% to approximately 5.7%, reducing the "true" forward operating margin to roughly 72–73%. AppLovin's operating margin expanded 34 percentage points to 76% since FY2023, but approximately half of the improvement traces to $157.9 million in stock-based compensation declining as the RSU pool depleted by 83%. The remaining half — driven by AXON's genuine operating leverage — is real and impressive, but investors should price the sustainable margin, not the depleted-pool margin.

The Tax Reset Nobody Discusses

FY2024's effective tax rate of 1.4% was an anomaly, not a baseline. The $22.4 million tax provision on $1.6 billion in pre-tax income reflected massive SBC tax deductions generated by stock option exercises during a period of rapid share price appreciation. When employees exercised 2,486,033 options at a $6.85 average strike price against market prices above $400, the company captured enormous tax deductions — $164.9 million in income tax benefits from vested and exercised equity awards in FY2024 alone.

By FY2025, the tax rate normalized to 13.1% as the dynamics shifted. SBC tax benefits declined to $123.0 million as the remaining option pool shrank to 1,256,114 options at $6.10 average exercise price. The global minimum tax began impacting the company's international structure. And the tax provision jumped 23x — from $22.4 million to $519.7 million — on pre-tax income of $3.95 billion.

This matters for earnings comparability. The widely cited 111% net income growth ($3,333 million in FY2025 versus $1,579 million in FY2024) compares a year with a normalized tax rate against a year with an unsustainably subsidized one. If you apply a 13% ETR to both years, the growth rate narrows considerably — still strong, but not the 111% headline that entered analyst models.

The forward implication is that the tax tailwind is permanently gone. With the low-basis option pool approaching depletion and global minimum tax rules tightening, AppLovin's forward ETR should be modeled at 13–16%. Every percentage point of ETR above the FY2024 anomaly of 1.4% represents approximately $40 million in additional annual tax burden at current earnings levels. AppLovin's effective tax rate normalized from 1.4% in FY2024 to 13.1% in FY2025 — a 23x increase in tax provision — because SBC-driven tax deductions from prior stock price appreciation are structurally fading.

E-Commerce Opacity and the Bad Debt Signal

Management has been explicit about refusing vertical disclosure. CEO Adam Foroughi has described breaking out e-commerce versus gaming revenue as "misleading information," framing the platform as "one single auction" where the AXON engine treats all advertisers identically. The 10-K supports this framing by reporting a single operating segment with zero vertical segmentation — making AppLovin the only company among its peers without segment-level economics.

But the filing contradicts its own framing in one revealing line.

"General and administrative expenses in 2025 increased by $68.6 million, or 42% compared to 2024, due primarily to an increase of $31.6 million in professional services costs primarily associated with transaction support and an increase of $24.5 million in bad debt expense primarily related to new initiatives."

"New initiatives" is the closest the 10-K comes to acknowledging that AppLovin's advertiser expansion into e-commerce and other verticals carries measurably different credit risk than its established gaming base. The $24.5 million bad debt increase is notable on its own, but the year-over-year swing is more dramatic: FY2024 saw a $6.3 million bad debt decrease, making the adverse swing $30.8 million. The timing aligns precisely with the company's public pivot toward e-commerce advertisers.

This matters because of the contractual structure. Substantially all of AppLovin's customer contracts are cancelable at any time, with 30-day payment terms after month-end.

"Substantially all of the Company's contracts with customers are cancelable at any time."

The combination creates an asymmetric information problem: investors cannot evaluate e-commerce advertiser economics independently, all contracts are cancelable at any time with zero remaining performance obligations, and the only quantitative signal the filing provides — bad debt from "new initiatives" — suggests the new advertiser mix has worse credit quality than the established base.

For context, Salesforce carries $67 billion in remaining performance obligations. ServiceNow locks customers into multi-year enterprise contracts. Even Shopify provides GMV by geography. AppLovin has zero contractual revenue lock-in, zero backlog, and zero vertical disclosure. The Tripledot relationship validates the divestiture strategy — the buyer of AppLovin's Apps business immediately became a $19 million advertising customer — but it also confirms that even known gaming partners operate on the same cancelable terms.

AppLovin disclosed a $24.5 million increase in bad debt expense from "new initiatives" in FY2025, the filing's sole acknowledgment that its e-commerce advertiser expansion carries different credit risk than its gaming base. Investors should track DSO and the bad debt-to-accounts receivable ratio as proxies for advertiser quality in the absence of segment disclosure.

Peer Context — Real Superiority, Structural Fragility

AppLovin's financial performance is unambiguously superior to its AI-driven peers across every profitability metric. The company ranks first in ROIC at 72.5%, first in operating margin at 68.5%, and first in revenue growth at 61.1%. Its Rule of 40 score of 129.5 (revenue growth plus FCF margin) is in a different stratosphere — the benchmark for healthy software is 40.

But the structural comparison reveals why AppLovin's financial superiority may command a different risk premium than the numbers alone suggest.

First, the SBC comparison with Palantir illuminates the timing artifact. PLTR's SBC runs at approximately 17% of revenue ($196 million in Q4 2025 alone, or $784 million annualized), making AppLovin's 3.9% look extraordinarily disciplined. The context matters: AppLovin's SBC is low because the RSU pool is 83% depleted, not because the company has structurally different compensation philosophy. PLTR is issuing equity aggressively; AppLovin has temporarily run out of legacy grants to amortize. When AppLovin's SBC normalizes toward 5.7%, the gap narrows — though APP remains more capital-efficient by a wide margin.

Second, AppLovin is the only levered company in the peer set. With $3.5 billion in total debt and a 0.55x net debt-to-EBITDA ratio, debt servicing risk is minimal — interest coverage stands at 20.1x. But leverage introduces asymmetry that net-cash peers don't face: if revenue deceleration coincides with SBC normalization, the fixed interest burden compresses free cash flow faster than it would for an unlevered competitor. At current rates, debt service is immaterial. The risk is that it becomes material in a scenario where everything else is also deteriorating.

Third, the revenue quality comparison is stark. CRM's $67 billion RPO backlog and NOW's multi-year enterprise contracts create contractual revenue floors that AppLovin completely lacks. When PLTR's government contracts or CRM's enterprise deals renew, they do so against switching costs and institutional procurement cycles. When an AppLovin advertiser's ROAS declines, they cancel tomorrow. The negative cash conversion cycle of -170 days (collecting from advertisers 170 days before paying publishers) generates structural float — but it also means a sudden advertiser pullback would create a cash flow air pocket before publisher payment adjustments catch up.

AppLovin ranks first among AI-driven tech peers in ROIC at 72.5%, operating margin at 68.5%, and revenue growth at 61.1%, but is the only company in the peer set carrying $3.5 billion in debt with zero contractual revenue lock-in. The financial superiority is real; the structural fragility is unique.

Stress-Testing the Margin Illusion Thesis

The analysis above argues that AppLovin's 76% operating margin is sustained by a dual illusion — SBC pool depletion and tax rate normalization — and that the revenue growth model's pricing dependence creates fragility. But any thesis worth stating is worth stating what would prove it wrong. Three specific observations in upcoming quarters would invalidate this analysis.

First, SBC/revenue stays below 4.5% through FY2026. If the RSU pool depletion represents a genuine structural shift toward lower equity compensation — rather than a timing gap between old grants expiring and new grants ramping — the SBC decline is permanent, not temporary. This would mean the PSU milestones either aren't activating (the stock isn't reaching $300 billion) or the company has deliberately shifted to cash compensation. Either outcome makes the "margin illusion" label incorrect.

Second, volume growth exceeds 15% in any quarter. The pricing-dependence concern rests on the assumption that 3% volume growth represents addressable market saturation. If AXON's expansion into e-commerce meaningfully expands the installation base — rather than simply repricing existing gaming volume — the growth model has two engines instead of one. Volume above 15% would demonstrate that the total addressable market is expanding, making the pricing-only growth model a complement to volume rather than a fragile substitute for it.

Third, operating margin sustains above 72% for four consecutive quarters despite SBC normalization. The margin decomposition shows approximately half the improvement came from SBC decline. If margins hold above 72% even as SBC expense rises toward $310 million per year, the genuine operating leverage is stronger than the decomposition suggests — meaning AXON's marginal economics are powerful enough to absorb rising compensation costs while maintaining profitability.

The data points to watch in Q1 2026 earnings: volume growth versus pricing growth in the revenue decomposition, SBC expense relative to the $53 million per quarter run rate, and whether management provides any vertical revenue commentary despite its stated opposition to segment disclosure. The market is pricing permanent margin superiority. The filing says the margin is partly borrowed time. The next two quarters will show which narrative the data supports.

Frequently Asked Questions

What was AppLovin's operating margin in FY2025?

AppLovin reported a 76% operating margin in FY2025, up from 59% in FY2024 and 42% in FY2023. However, approximately half of the improvement came from a $157.9 million decline in stock-based compensation across S&M and R&D lines, not from operational efficiency gains. Cash operating expenses actually rose: COGS increased 28% and G&A increased 42%. The forward operating margin, adjusted for SBC normalization to approximately $310 million per year, is closer to 72–73%.

How much of AppLovin's margin improvement is from stock-based compensation?

Roughly 50%. S&M SBC declined $42.8 million (from $76.8 million to $34.1 million) and R&D SBC declined $115.1 million (from $229.6 million to $114.5 million), totaling $157.9 million. This accounts for 80% of the combined S&M and R&D expense decrease of $197.4 million. The RSU pool collapsed 83% from 2,150,021 to 358,162 outstanding units, creating a structural depletion event that will reverse as new grants at higher prices and PSU vesting flow through.

What are AppLovin's PSU milestones?

In October 2025, AppLovin granted 920,526 performance stock units to key non-executive engineering employees with a 7-year performance period. Vesting requires achieving market cap milestones starting at $300 billion (32% above the current $228 billion market cap) up to $1 trillion. The total grant value was approximately $410 million ($445.89 per unit at 70.95% expected volatility). If milestones are achieved, this creates a significant deferred SBC expense not reflected in the current 3.9% SBC/revenue ratio.

Why is AppLovin's revenue growth pricing-dependent?

In FY2025, installation volume grew only 3% while net revenue per installation grew 72%, meaning approximately 97% of the 70% revenue growth came from pricing. This represents a fundamental inversion from FY2024, when volume grew 50% and pricing grew 22%. Management acknowledges this risk in the 10-K, warning that the growth rate may decline in the future due to "more difficult comparisons to prior periods and the saturation of the market."

What is AppLovin's infrastructure cost floor?

AppLovin has $702.8 million remaining on a $1.3 billion cloud commitment (having already paid $579.3 million) plus $173.9 million in lease obligations for server and network equipment. Combined annual infrastructure obligations are approximately $525 million, or 9.6% of FY2025 revenue. This creates a cost floor that constrains margin flexibility and qualifies the "asset-light" characterization of the business model.

How does AppLovin compare to Palantir?

Both are AI-driven growth companies with 50%+ revenue growth. AppLovin's operating margin (68.5%) is 2.2x Palantir's (31.6%), and AppLovin's ROIC (72.5%) is 3.2x Palantir's (22.7%). However, Palantir's SBC runs at approximately 17% of revenue versus AppLovin's 3.9% — a gap that partly reflects AppLovin's RSU pool depletion rather than structural compensation discipline. Palantir has zero debt while AppLovin carries $3.5 billion, and Palantir provides Government/Commercial segment disclosure while AppLovin reports a single segment with no vertical breakout.

What is AppLovin's effective tax rate and why did it change?

AppLovin's effective tax rate increased from 1.4% in FY2024 to 13.1% in FY2025, a 23x increase in tax provision from $22.4 million to $519.7 million. The ultra-low FY2024 rate was driven by $164.9 million in SBC tax benefits from stock option exercises during a period of rapid share price appreciation. As the remaining option pool depletes (1,256,114 options at $6.10 average), these benefits diminish. Forward ETR is expected at 13–16%, eliminating the tax tailwind entirely.

Why does AppLovin refuse to disclose e-commerce revenue?

CEO Adam Foroughi has stated that breaking out e-commerce versus gaming revenue would be "misleading information," framing the platform as "one single auction." The 10-K reports a single operating segment with no vertical segmentation. This is unique among peers — Palantir, Shopify, Salesforce, and ServiceNow all provide segment-level economics. The lack of disclosure creates an asymmetric information gap for investors trying to evaluate the credit quality and economics of new advertiser verticals.

What are the securities class action lawsuits against AppLovin?

Multiple putative class actions were filed under Exchange Act Sections 10(b) and 20(a) covering November 7, 2024 through March 27, 2025. The consolidated Brownback Action added CTO Basil Shikin as a defendant via an Amended Complaint in September 2025, with a motion to dismiss hearing scheduled for March 2026. AppLovin has accrued zero loss contingencies, stating it cannot estimate potential exposure.

What does AppLovin's negative cash conversion cycle mean?

AppLovin has a cash conversion cycle of -170 days: DSO of 100 days, DIO of 0, and DPO of 270 days. The company collects from advertisers approximately 170 days before paying publishers, creating a structural float where growth generates cash rather than consuming it. This is a genuine competitive advantage — but it also means a sudden advertiser pullback would create a cash flow air pocket before publisher payment adjustments normalize.

What is AppLovin's Rule of 40 score?

AppLovin's Rule of 40 score is approximately 129.5 (61.1% revenue growth plus 68.4% FCF margin). The benchmark for healthy software is 40. Among peers, Palantir scores approximately 107, Shopify 48, ServiceNow 43, and Salesforce 27. However, AppLovin's score is partly boosted by the SBC artifact depressing reported costs — adjusting for forward SBC normalization would reduce the score by approximately 2 points, still extraordinary by any standard.

What would disprove the margin illusion thesis?

Three observations would invalidate it: (1) SBC stays below 4.5% of revenue through FY2026, showing the decline is structural discipline rather than pool depletion; (2) volume growth exceeds 15% in any quarter, proving the addressable market is still expanding beyond the current 3% rate; (3) operating margin sustains above 72% for four consecutive quarters even as PSU vesting activates and new RSU grants ramp.

Methodology

Data Sources

This analysis uses AppLovin's FY2025 10-K filed with the SEC on February 19, 2026, accessed via the MetricDuck filing viewer. Financial metrics are computed by the MetricDuck metrics processor from raw XBRL data extracted from SEC EDGAR. Peer comparison data for ServiceNow, Palantir, Shopify, and Salesforce is sourced from each company's most recent annual filing through the same pipeline. Guidance figures reference Q4 2025 earnings call transcripts.

Limitations

• SBC forward estimates are approximations. The $310M annual forward SBC figure assumes partial PSU vesting and straight-line RSU amortization. Actual SBC will depend on stock price trajectory, forfeiture rates, and the timing of new equity grants not yet announced.
• Volume and pricing decomposition is management's categorization. The 3% volume / 72% pricing split comes from AppLovin's own MD&A disclosure. We cannot independently verify the methodology used to allocate growth between volume and pricing.
• Peer comparisons use trailing metrics. Each peer's data reflects its most recent annual filing, which may not align to the same fiscal period. PLTR SBC figures are annualized from quarterly data.
• E-commerce revenue contribution is unknown. Without segment disclosure, all revenue quality analysis relies on indirect signals (bad debt, management commentary) rather than direct measurement.
• Tax rate projections assume current law. Forward ETR estimates of 13–16% assume no material changes to US corporate tax rates or global minimum tax implementation.

Disclaimer

This analysis is for informational purposes only and does not constitute investment advice. The author does not hold positions in APP, NOW, PLTR, SHOP, or CRM. Past performance and current metrics do not guarantee future results. All data is derived from public SEC filings and may contain errors or omissions from the automated extraction process.