Date: 2026-02-25 Quarter: Q4 FY25 (ended December 31, 2025) Analyst: Bert Hochfeld (TickerTarget / HHHYPERGROWTH) Prior coverage: None. First analysis of APP in the newsletter.
I haven't covered AppLovin in the newsletter before, but applying my framework to this company is an exercise I find genuinely compelling — and the conclusion is hard to escape. AppLovin delivered the strongest quarter in its history: $1.66 billion in revenue growing 66% year over year, $1.40 billion in adjusted EBITDA at an 84% margin, and $1.31 billion in free cash flow. Rule of 40 score of 150. That figure is not a typo. The company is growing at 66% while generating 84% EBITDA margins — a combination that essentially does not exist elsewhere at this revenue scale. The stock has been cut in half from its year-end 2025 highs on concerns about AI competition (specifically Microsoft's CloudX launch in February), the SEC investigation into AXON's data collection practices, and short-seller allegations. I have watched the market misprice hypergrowth companies before, but the gap between APP's operational reality and its current valuation is among the more striking I've seen. At approximately 17x EV/trailing revenue with 66% growth and 79% FCF margins, the stock is cheaper on any growth-adjusted basis than Meta, and the FCF multiple of 23x is at the low end of what I would pay for a business growing FCF at 91% per year. My one meaningful reservation is the SEC investigation — that is a binary risk I cannot quantify, and it prevents me from adding aggressively at current prices. Subject to that caveat, I am comfortable initiating a position here.
Headline KPIs:
The guidance pattern here is one of the cleaner examples I've seen of deliberately undemanding estimates. Every quarter since Q4 FY24, management has guided to the midpoint and beaten it by 4.5–8.3%:
| Quarter | Guide Midpoint | Actual | Beat |
|---|---|---|---|
| Q2 FY25 | $1,205M | $1,258.8M | +4.5% |
| Q3 FY25 | $1,330M | $1,405.0M | +5.6% |
| Q4 FY25 | $1,585M | $1,657.9M | +4.6% |
The EBITDA beat pattern is equally consistent: Q3 beat by $77.7M; Q4 beat by $93.7M. When I see four consecutive beats of this magnitude with no misses, I assume conservatism is structural, not luck. The Q1 FY26 midpoint of $1,760M should accordingly be treated as a floor estimate. Applying the 4.5–5.5% historical beat pattern implies Q1 FY26 actual revenue of 1.83–1.86 billion. At the time CFO Heaney set Q1 guidance, the company was also lapping the self-service launch of e-commerce (which went live October 1, 2025) and benefiting from a model unlock that, per Foroughi, rolled out "weeks before" the February 11 earnings call. That recent unlock is embedded in both the Q4 actuals and the Q1 guide.
The Q4 revenue print is clean — and the sequential re-acceleration to 18.0% QoQ is the most important number in the entire release. The prior quarter had decelerated to 11.6% QoQ and the year-over-year rate had compressed from 77% (Q2 FY25) to 68% (Q3 FY25) to 66% (Q4 FY25). Analysts going into the print were pricing further YoY deceleration through 2026. The 18% sequential pop — the largest dollar add in the company's history (+$252.9M QoQ) — directly contradicts that narrative.
The source of the acceleration is Foroughi's "model unlock": a material improvement in AXON's underlying prediction model, rolled out weeks before the call. Management declined to quantify the magnitude, which is characteristically unhelpful for investors trying to build a forward model. But the evidence is in the results and the guidance: a company that typically grows 10–12% sequentially in Q4 grew 18%, then guided to 5–7% sequential growth in Q1, when Q1 is historically the weakest quarter. That Q1 guidance is meaningfully above seasonal expectations.
On comparability: Q4 FY24 revenue was $999.5M (advertising-only, post-Apps divestiture). The 66% YoY comparison is therefore clean — same business. For prior quarters (Q1–Q3 FY24), the Apps gaming studio was included. The advertising-only YoY growth trend, using advertising segment revenue, shows: 91% → 75% → 66% → 73% → 71% → 77% → 68% → 66%. The deceleration is real but non-linear. The 91% figure represents a very easy comp; the 66% current rate is the more representative steady state.
APP does not report ARR, RPO, or billings — it is a performance advertising platform, not a SaaS business. The leading indicators I track here are: pixel penetration (5,306 e-commerce sites pixeled as of Q4, growing rapidly), cohort spending (existing e-commerce advertisers saw "material increase" in spend YoY plus a "pretty sizable uplift" in the weeks before the call), and the MAX auction bid density metric (increasing, with more competitors bidding into the exchange). All three are moving in the right direction.
The margin story here is exceptional by any standard I apply in this newsletter. The gross margin has expanded from 63% in Q1 FY23 to 88.9% in Q4 FY25. The operating margin from 8.5% to 76.9%. The EBITDA margin from 49% in Q3 FY23 to 84.4% now. FCF margin from 40% to 79%.
What makes this unusual is not just the levels — it is that they are still expanding at scale. At $6.6 billion in annual revenue run rate, most software businesses have already exhausted their operating leverage. APP's Q4 incremental EBITDA flow-through was approximately 95%: for every incremental dollar of revenue above the prior quarter, the company converted 95 cents to EBITDA. That is not a business with meaningful variable cost exposure. The cost structure is nearly entirely fixed: the AXON model infrastructure runs at near-zero marginal cost per additional auction processed.
The one SBC note: Q4 SBC jumped to $80.5M (4.9% of revenue) from $33.8M in Q3, due to the annual grant cycle. This is timing, not deterioration. FY2025 total SBC of $208M is down 44% year over year from $369M in FY2024. As a percentage of revenue, SBC is declining from ~10% historically to under 5%. That improvement in economic FCF quality — the gap between reported FCF and diluted FCF — is widening in shareholders' favor. Not many companies at 70% growth can say that.
The leading indicator picture is supportive, though the lack of segment disclosure makes precision modeling difficult.
The e-commerce segment is the key forward variable. Management will not break it out — Heaney's rationale is logical (if the gaming model improves 50%, e-commerce share falls mechanically even as the total business accelerates; vertical breakouts would mislead investors). I find the logic defensible, even if I'd prefer the transparency. What we know:
The 57% conversion rate is the binding constraint. Until the gen AI video model goes live and the breakage rate approaches 90%+, e-commerce GA will have more friction than the bull case assumes. But the direction is correct: the tool that directly addresses the breakage problem (automated video ad generation) is already in pilot.
The prospecting product launched in Q4 FY25 — targeting new customer acquisition rather than retargeting — received "fantastic results" and rapid adoption per management. It is too early to size the contribution, but this is the kind of product expansion that compounds: better new-customer targeting → more advertiser budget → more training data → better model → more advertiser budget.
MAX auction dynamics remain favorable. Bid density is increasing as more platforms (Meta, Unity/Vector, Liftoff Cortex, Moloco, Google) compete within APP's exchange. The economics here are asymmetric: when a competitor wins a high-value impression, APP still runs the auction and takes a 5% fee on lower-value impressions. The current conversion rate of ~1% of served impressions compares to a stated 5% achievable at full targeting confidence. That gap represents five times the revenue headroom from conversion improvement alone, before any growth in the advertiser base.
APP's competitive position in mobile gaming mediation is dominant. Pixalate's Q3 2025 data shows 85% iOS SDK reach. Per the Atlas baseline (which I concur with): MAX controls ~80% of overall mobile mediation share. Unity, the only credible near-term alternative, has seen revenue decline 10% in 2024 while APP grew 70%. Unity's "Vector" product is rebuilding but is years behind on training data.
Foroughi's response to the CloudX threat (Microsoft's February 4, 2026 launch — the proximate trigger for the stock selloff) was the most substantive part of the call. The central argument: CloudX primarily addresses search and web, not mobile gaming mediation. CloudX lacks the mobile gaming training data that makes AXON's model accurate. Getting from zero to AXON's level of predictive precision in mobile gaming would require years of data accumulation, not just a superior model architecture. This is credible. AXON has 10+ years of closed-loop gaming data — every in-app purchase, every ad engagement, every user journey across the world's largest mobile mediation network. OpenAI's general-purpose LLMs do not replicate this.
On Meta: Foroughi's argument that Meta is constrained from bidding on no-IDFA traffic by Apple's ToS is legally plausible but represents a management view, not a regulatory prohibition. I take note of this dependency — if Apple's enforcement stance softened or if Meta found a compliant way to infer identity at scale, the threat would become more immediate. For now, two-thirds of full-screen ad impressions remain outside Meta's bidding capacity. That is a meaningful structural protection.
The Israeli cookware company example on the call — $4M revenue to $16M to projected $80M, with 65% of UA spend on APP — is the SMB compounding flywheel in miniature. This is the same flywheel that drove gaming indie developers to scale: APP helps small advertisers outperform expectations → they allocate more budget → APP's model improves on their data → performance improves further → they grow → they become larger advertisers. It worked in gaming; there is no structural reason it cannot work in e-commerce.
This is where the investment case is most compelling. I am applying my GARP framework here at the profitable end of the spectrum — P/E, FCF multiple, and EV/EBITDA rather than pure EV/S, because APP is genuinely profitable at GAAP earnings levels.
Assuming stock at approximately $270 at time of analysis, with 338M diluted shares outstanding:
| Metric | Value | Context |
|---|---|---|
| Market cap | ~$91B | 338M shares × $270 |
| Net debt | $1,026M | Dec-2025 |
| Enterprise value | ~$92B | — |
| EV / TTM Revenue ($5,481M) | 16.8x | 66% YoY growth |
| EV / Forward Revenue ($7,040M ann.) | 13.1x | Q1 FY26 guide × 4 |
| EV / TTM EBITDA ($4,512M) | 20.4x | 82% margin |
| EV / TTM FCF ($3,952M) | 23.3x | 72% margin |
| P/E GAAP (TTM, $3,334M net) | 27.4x | 111% net income growth YoY |
PEG analysis: EV/S of 16.8x ÷ revenue growth of 66% = PEG of 0.25x. At the Q1-guided forward rate of ~52% growth: 13.1x ÷ 52% = 0.25x forward PEG. I have covered a lot of hypergrowth companies in this newsletter, and a sub-0.30 PEG for a company at this profitability level and revenue scale is genuinely unusual.
Cohort comparison: For companies growing at 50–70% YoY, typical EV/S in the current market runs 20–30x, depending on FCF margin profile. APP at 17x is 30–45% below where it would trade on cohort mechanics alone, if the SEC and competition narratives were not present. The closest large-cap peer for comparison is Meta: trading at roughly 12x EV/TTM revenue with 15–20% growth. Meta's PEG on EV/S basis is therefore ~0.70–0.80x. APP's PEG at 0.25x is less than one-third of Meta's on the same metric, despite growing four times faster.
The bear case on valuation: If YoY growth decelerates to 30–35% by H2 FY26 without e-commerce compensating, and if the SEC investigation materially impairs AXON's data collection mechanism, fair value on a 10x forward EV/S multiple at $7–8B forward revenue would be roughly $70–80B market cap — a modest downside from current levels. That is not the catastrophic outcome the current narrative implies. The downside is bounded by the cash generation: the company is producing $3.95B of annual FCF and buying back $2.58B per year. Even in a 30% growth scenario, the business compounds its intrinsic value substantially.
The upside case: If e-commerce scales to 20–25% of revenue by Q4 FY26 (from ~10% in Q1 FY25), the YoY growth rate inflects upward rather than decelerating. A 60–70% growth rate on a $7–9B revenue base with 82%+ EBITDA margins would support 20x+ EV/forward revenue — roughly double today's multiple. That is a 2x stock from here. Asymmetry favors holding.
I rate AppLovin's management as reliable with one yellow flag.
The guidance pattern is a textbook example of what I look for: every quarter underpromised, every quarter overdelivered, four consecutive beats averaging 5.3% above midpoint. SBC is declining in absolute dollars and percentage terms simultaneously. The balance sheet has been actively improved: net debt fell from 1.17x EBITDA (December 2024) to 0.23x EBITDA (December 2025) funded entirely by FCF. Foroughi runs the company with 1,500–1,600 employees generating $5.6M of quarterly EBITDA per head. That operational discipline is real and sustained.
The yellow flag: the CTV initiative. In the Q4 FY24 call, Foroughi explicitly committed to building a CTV performance advertising product for 2025. There was zero mention of this in Q3 or Q4 FY25 — no disclosure, no update, no acknowledgment that the priority had changed. A management team that abandons public commitments without explanation will do so again. I will watch the e-commerce GA (H1 FY26 commitment) and the gen AI video model ("shortly") for the same pattern. If either slips without explanation, credibility scoring moves down.
The SEC response posture (hiring Alex Spiro at Quinn Emanuel, an aggressive white-collar firm) is consistent with a company that believes the allegations are without merit. It could also be consistent with a company managing a serious legal problem with maximum aggression. I cannot distinguish between the two from the outside. That is the honest answer.
Primary: SEC investigation. This is a binary risk. If the SEC determines that AXON's ML model was trained on user data collected via fingerprinting in violation of Apple/Google ToS, APP could be required to retrain the model from scratch using only compliant data — a multi-year degradation of AXON's predictive accuracy. The three pending federal lawsuits add litigation cost and distraction. No resolution timeline visible. The Bloomberg February 20 confirmation that the probe is "still active and ongoing" means this overhang remains present. I am not adding aggressively until there is clarity.
Secondary: YoY deceleration. Growth is decelerating from 77% (Q2 FY25) toward a guided ~52% in Q1 FY26. If e-commerce does not scale fast enough to compensate, the YoY rate could compress to 30–35% range by H2 FY26. Below 30% growth, APP exits my hyper-growth threshold and the valuation multiple must compress toward GARP P/E territory. The Q1 re-acceleration case (model unlock + self-service launch) mitigates this risk, but it is the key variable to watch.
Tertiary: E-commerce advertiser retention. Ridge Wallet and Jones Road (per scuttlebutt data) both pulled back after initial strong ROAS — the "arbitrage window" closed as the model learned the competition. AXON must prove durable incrementality, not just early-mover ROAS. The measurement gap vs. Meta is real and slows enterprise adoption. The 43% lead breakage rate is higher than I would like going into GA.
Initiate long. Target weight: 6–8% in the high growth portfolio.
The combination of 66% revenue growth, 84% EBITDA margin, 79% FCF margin, and an EV/S of 17x with a PEG of 0.25x is the most attractive risk/reward profile I have reviewed in recent quarters. The stock has been punished for two narrative risks — AI disruption (credibly refuted by management; CloudX does not have mobile gaming training data) and the SEC investigation (real, binary, unresolvable on current information). The operational business hit all-time highs on every metric in Q4 FY25. The sequential re-acceleration signals that the model unlock embedded in Q4 actuals continues into Q1 FY26.
My concentration limit prevents a full position at 10%+ until the SEC investigation resolves. At 6–8%, APP becomes the second-largest position in the portfolio behind NVDA. I would add to 10%+ on SEC resolution or on evidence that e-commerce is scaling faster than the guided ~10% revenue contribution.
The Q1 FY26 result will be the first meaningful test: if APP beats the $1,760M midpoint by 4.5–5% (consistent with the historical beat pattern), it will report approximately 1.83–1.86B, and YoY growth will be in the 56–61% range — an acceleration, not a deceleration, versus the 52% implied by the guide. That print, combined with clarity on the gen AI video model rollout and e-commerce GA timing, will be the catalyst for a re-rating.
This is my first analysis of APP. No prior corpus coverage. Beliefs established from first principles:
| Metric | Prior Assumption (entering) | Actual | Delta |
|---|---|---|---|
| Revenue beat | 4–6% above guide midpoint (based on track record) | 4.6% | Confirmed |
| Sequential acceleration | Unlikely given Q4 seasonal | +18.0% QoQ | Large positive surprise |
| EBITDA margin | ~82–83% (Q3 exit rate) | 84.4% | Positive surprise |
| FCF conversion | 75–77% (trend line) | 79.0% | Positive surprise |
| E-commerce update | Referral-only confirmed; GA timeline reaffirm | H1 FY26 confirmed; 30-day LTV:CAC breakeven disclosed | Positive; conversion gap (43% breakage) is new data |
| Management tone | Defensive, post-selloff | Proactively confrontational on competition; data-backed | Consistent with high-quality CEO |
| SEC | Unresolved | Bloomberg confirmed still active | No change |
| Model unlock | None expected mid-quarter | Material; baked into Q4 and Q1 guide | Key new finding — explains re-acceleration |
The model unlock is the single largest new data point relative to expectations. It explains why Q4 beat consensus by more than history suggested, and why Q1 guidance is above seasonal norms. If AXON's model improvement cadence is quarterly (as Foroughi implied), the bears' deceleration narrative underestimates the endogenous growth driver embedded in the system.
AppLovin FY2025: Revenue $5.5B; EBITDA $4.5B; FCF $3.95B; Rule of 40: 150. First coverage. Initiating long at 6–8% of the high growth portfolio.