The ARR Engine

The ARR Engine

Adobe's case rests on a subscription base that renews and expands. That base is still large and still growing double digits, but the engine that adds to it has a specific tell: net-new Digital Media ARR has sat near $2.0 billion for three straight years — $1.91B, $2.00B, $1.98B — even as the installed base grew by a quarter and Adobe raised prices and shipped AI across the portfolio. In fiscal 2026 management is deliberately trading roughly $500 million of near-term recurring revenue for free-user reach. Whether that is offense or repair is what this chapter tests.

Net-new ARR has plateaued near $2 billion

Digital Media — Creative Cloud plus Document Cloud — is where the recurring revenue lives, and Adobe measures it as Annualized Recurring Revenue, the annual value of its subscriptions and enterprise term licences [1]. That book reached $19.20 billion at the end of fiscal 2025, up 11.5% [2]. The level is not in doubt. What matters for the forward case is the addition to it each year — the net-new dollars the engine produces.

Those dollars have been strikingly flat. Adobe added $1.91 billion of net-new Digital Media ARR in fiscal 2023 [3], crossed $2 billion "for the first time ever" in fiscal 2024 at $2.00 billion [4], then produced about $1.98 billion in fiscal 2025 — the difference between the $19.20 billion year-end book and the $17.22 billion it started with on a constant-currency basis [5].

Sources: Q4 FY2023 [6] and Q4 FY2024 [7] earnings calls; FY2025 derived from the ARR reconciliation in the FY2025 10-K ($19.20B ending less $17.22B beginning) [8].

The flat dollar figure and the falling percentage are the same fact seen two ways: on a base that grew from about $15.3 billion to $19.2 billion, holding net-new at $2 billion means the growth rate slides — from roughly 13% in constant currency in fiscal 2023 to 11.5% in fiscal 2025 [9]. A base this large decelerating in percentage terms is arithmetically normal. The sharper observation is that three years of accelerated AI product launches and higher-priced tiers did not lift the absolute output above $2 billion. Management's counter is that the trend turned inside fiscal 2025: it called Q4 net-new Digital Media ARR a record with "growth reaccelerated year over year," led by Creative Cloud Pro, Acrobat and Firefly [10]. A single strong quarter is a leading indicator, not yet a trend.

Retention is strong but only partly disclosed

The durability of the base rests on retention, and here the evidence is genuinely incomplete: Adobe does not publish a net-revenue-retention or gross-retention rate. What it discloses are proxies, and they point the same way. The enterprise cohort is expanding — customers with more than $10 million of ARR grew 25% year over year to more than 150 in fiscal 2025 [11], and that cohort was still growing more than 20% by the second quarter of fiscal 2026, with management citing "continued strength in retention across the enterprise customer base" [12].

The composition of net-new ARR is the more useful tell on pricing power. In fiscal 2025, more than 75% of net-new Digital Media ARR came from subscription growth and cross-sell and upsell, "with the remainder from value-based pricing" [13]. Fewer than a quarter of the incremental dollars came from raising prices; the bulk came from adding and expanding customers. That is a healthier mix than a business leaning on price to hit its number — but it also means Adobe's growth is more exposed to unit demand than a pure pricing-power story would suggest.

Source: Q4 FY2025 earnings call — "over 75%… from continued growth in subscriptions and cross-sell and upsell, with the remainder from value-based pricing" (management stated 75% as a floor) [14].

The upsell mechanism is visible in the product data: nearly half of Acrobat commercial enterprise agreements renewed in Q4 fiscal 2025 already upgraded to the higher-value Acrobat Studio bundle [15]. Renewal-plus-upgrade is the pattern a moat produces. The gap in the evidence — no disclosed retention rate — is a real limitation, so this reads as strong but not fully verifiable rather than proven.

The freemium pivot spends the engine's output

The fiscal 2026 plan is where the tension surfaces. Entering the year, Adobe guided total-company ARR growth of 10.2%, about $2.6 billion of net-new — "the highest beginning-of-year guide for total net-new ARR" it had ever given [16]. By June it held that same 10.2% target — but the mix underneath it had changed materially [17].

Two things moved in opposite directions. Adobe closed the SEMrush acquisition, adding roughly $480 million of ARR to the book [18]. And it chose to route more of its surging web traffic into free "friction-free" onboarding instead of paywalls — a shift management said "lowers our second-half ARR growth expectations from individual subscribers" and "will come at the cost of short-term ARR" [19]. On the call, management accepted an analyst's framing of roughly a half-billion-dollar downward adjustment to organic ARR, and split it: "maybe half… is a result of deferring those creative price line optimizations, and the other half is about going full steam on… the freemium experience" [20].

Source: SEMrush ~$480M ARR added [21]; the ~$500M organic reduction, split roughly half deferred pricing / half freemium, from management's Q&A [22]. Split shown 50/50 per management's characterization.

Read together, the arithmetic is pointed: an acquisition worth about $480 million of ARR roughly offsets a self-imposed organic cut of about $500 million, and the headline 10.2% growth target survives largely intact. The organic engine was throttled; a bought book of business filled the hole. That is what this chapter establishes.

Half of that cut is a pricing decision. Adobe had "previously planned Creative Cloud second-half line optimizations" — further price increases — and deferred them [23]. Management frames the deferral as focus, not weakness: "it is not going away," and the creative business is "extremely stable" [24]. A skeptic reads the same fact as reluctance to test price elasticity while free, capable AI alternatives multiply. The evidence does not yet separate the two.

The bet the pivot is making

The other half is a reach bet, and its scale is real. Adobe's free funnel is now enormous: monthly active users of Acrobat and Express surpassed 850 million, up about 20% year over year, and creative freemium users — web and mobile versions of Firefly, Express, Photoshop and others — grew from 50 million to 90 million in a year [25]. The wager is that converting a sliver of roughly 940 million free users to paid, over time, beats collecting a smaller number of direct-to-paid subscribers today, because freemium converts tend to engage more and carry higher lifetime value [26].

The monetization that has to validate that bet is still small against a $19.2 billion base. Firefly's ending ARR was approaching $300 million exiting Q2 fiscal 2026 [27], and the broader set of AI-first products crossed $500 million, roughly triple a year earlier [28]. Fast-growing, and directionally the proof management points to — but together only a low-single-digit share of the book, and not yet large enough to prove the funnel pays back the ARR it is giving up.

Sources: MAU figures — Q2 FY2026 call [29]; Firefly ARR "approaching $300M" [30] and AI-first ARR "greater than $500M" [31]. MAU in millions; ARR shown at the disclosed thresholds.

The read

The recurring-revenue engine is healthy in level and intact in retention, but it has not been accelerating: net-new Digital Media ARR held near $2 billion for three years, and management is now spending part of that output on free-user reach while a bought book of business keeps the growth rate whole. The most defensible reading is that the freemium pivot is an offensive reallocation — funded, deliberate, and consistent with strong enterprise retention — rather than an admission that direct-to-paid conversion is breaking. The strongest fact against that reading is that Adobe simultaneously deferred planned price increases: pausing both new subscribers and price during the same window is what a company does when it is unsure how much the market will bear.

What would settle it runs in one of two directions. Total net-new ARR climbing durably above the roughly $2.6 billion fiscal 2026 guide, with Firefly and AI-first ARR compounding fast enough to cover the freemium give-up, would confirm the reach bet is paying back. Total net-new ARR slipping below that guide — with organic ARR decelerating as free tiers absorb users who would otherwise have paid, and the deferred price increases never returning — would confirm the opposite. The line items to watch are total Adobe net-new ARR, the Firefly and AI-first ARR run-rates, and whether the Creative Cloud price increases reappear in fiscal 2027.