Adobe reports fiscal third-quarter earnings tonight, in its first release since naming Anil Chakravarthy as the company’s next CEO. This Adobe earnings preview comes as Wall Street expects roughly $6.69 billion of revenue and $6.08 of adjusted EPS, figures sitting almost exactly in the middle of Adobe’s prior guidance. Yet judging this quarter on a conventional beat-or-miss framework risks missing what investors actually care about.
OpenAI’s new GPT-6 Astra has revived the debate over whether increasingly capable frontier models can absorb functionality that once required specialized software. Salesforce, ServiceNow and Intuit were among the software names hit as that concern resurfaced this week. Adobe sits directly in the middle of that argument.
The obvious question is whether AI threatens Photoshop, Premiere and Creative Cloud. The more consequential question may be different: can Adobe become the workflow layer through which competing AI models are actually used and monetized?
This quarter can provide evidence, but the real verdict probably arrives over the next two quarters.
The Base Case Is Almost Too Obvious
Consensus is not demanding a dramatic financial surprise. Analysts expect approximately $6.69 billion of revenue and $6.08 of adjusted EPS, compared with Adobe’s own Q3 targets of $6.67 billion to $6.72 billion and $6.05 to $6.10, respectively. Management had also guided to roughly a 44% non-GAAP operating margin.
That makes the traditional earnings hurdle unusually straightforward. Adobe has spent years producing solid revenue, earnings and cash-flow execution, yet the stock has repeatedly failed to reward those results. Shares have fallen after 15 of the company’s past 20 earnings reports and are down roughly 27% in 2026.
The market is therefore already assuming that another clean quarter will not resolve the structural argument. Investors know Creative Cloud remains highly profitable. They know Acrobat has enormous distribution. They know Adobe can beat quarterly EPS through strong recurring revenue conversion, disciplined spending and buybacks.
What they do not know is how much those advantages are worth in an AI-native world.
Last quarter, Adobe actually complicated its near-term earnings story deliberately. Management chose to push more users toward free Firefly, Express and Acrobat experiences while deferring planned Creative Cloud “line optimizations.” Adobe acknowledged that the decision would reduce second-half individual subscriber ARR relative to its previous plan because management wants a much larger acquisition funnel before applying more aggressive monetization.
That means a softer ARR trajectory alone cannot automatically be read as competitive deterioration. Some of it is self-inflicted by design.
The market is effectively being asked to tolerate weaker near-term monetization in exchange for a much larger AI-era funnel.
That is why the conventional narrative — stable Creative Cloud, decent margins, another EPS beat — is incomplete. Adobe must increasingly prove that the users entering that funnel eventually become economically valuable users.
The Metric That Actually Matters: AI-First ARR
Adobe has no shortage of AI engagement statistics. What matters now is whether that engagement is becoming revenue.
The cleanest scoreboard is AI-first ARR growth. For this Adobe earnings preview, that metric matters more than another routine headline beat because it directly tests whether adoption is becoming monetization.
Adobe exited Q2 with AI-first ARR above $500 million, more than triple the prior-year level. Firefly ending ARR alone was approaching $300 million across the Firefly application, credit packs and enterprise offerings, while management said Firefly ARR had increased roughly 50% sequentially. Creative freemium monthly active users had expanded from more than 50 million to above 90 million year over year.
Those are strong adoption signals. They are not yet proof that Adobe has solved the monetization problem.
This distinction matters because Adobe is willingly changing its economic model. Historically, Adobe largely monetized access to applications. In the emerging architecture, a user may enter through an LLM, ask an agent to execute a task, use a third-party model to generate an asset and only then employ Adobe technology for editing, production, governance and delivery.
Adobe’s response is surprisingly open rather than defensive.
Firefly now integrates models from Google, OpenAI, Runway, Luma, Kling and others. Adobe explicitly markets Firefly as a multi-model environment where users choose the model that best fits a particular creative job without leaving the workflow.
That strategy moved deeper into professional workflows this week. Adobe introduced generative capabilities directly inside Premiere’s timeline with model choice spanning Adobe Firefly and partner models including Google Veo, Kling, Runway and Luma.
The economics are the crucial part. Premiere users can spend Adobe generative credits regardless of whether the underlying generation comes from an Adobe model or a partner model. They do not need separate accounts with every provider.
In other words, Adobe is testing whether it can become the creative AI tollbooth even when somebody else builds the engine.
If AI-first ARR continues accelerating as Adobe expands access to third-party models, the market gets evidence that model competition does not necessarily commoditize Adobe. Every new frontier model could instead make Adobe’s orchestration layer more useful by giving creators more capabilities without forcing them to rebuild their workflows elsewhere.
If AI usage explodes while ARR barely moves, the conclusion becomes much less comfortable.
Adobe does not need to prove that Firefly is the world’s best model; it needs to prove that owning the workflow is economically more important than owning every model inside it.
Adobe enters tonight’s report with strong AI adoption, a heavily compressed valuation and a market debate centered on structural disruption. The key question is not whether Firefly wins every model comparison, but whether Adobe can convert its expanding multi-model workflow into durable economics. AI-first ARR and paid consumption are the clearest evidence investors need next.
Rising AI-first ARR and generative-credit consumption could show Adobe monetizes third-party models through its workflow, reducing fears of structural disintermediation.
AI engagement could keep rising without comparable paid monetization, leaving Adobe to absorb higher AI costs while Creative Cloud economics weaken.
Watch AI-first ARR growth alongside generative-credit consumption and Creative Cloud retention to determine whether Adobe’s larger AI funnel is becoming economically valuable.
Adobe’s valuation already reflects substantial disruption risk, but the rerating case depends on proof that AI usage expands monetization rather than merely defending subscriptions. The decisive variable is whether workflow ownership can convert model proliferation into durable ARR and consumption growth.
The Upside Surprise: Adobe Becomes The AI Aggregator
The upside pre-mortem does not require revenue to dramatically exceed $6.69 billion. It requires management to demonstrate that the AI funnel is beginning to convert faster than investors expect.
The clearest version would combine continued AI-first ARR acceleration with rising generative-credit consumption, strong Firefly paid conversion and stable Creative Cloud retention. In this Adobe earnings preview, that combination is the cleanest route to a genuine narrative change. Investors would also want confirmation that customers using partner models inside Adobe are generating incremental consumption rather than merely substituting one Adobe feature for another.
That would fundamentally change the interpretation of Adobe’s multi-model strategy.
Instead of asking whether Google’s Veo, OpenAI’s image models, Runway or Kling will replace Firefly, investors could begin asking why professional creators would want separate subscriptions, interfaces and production environments for every model when Adobe can aggregate them inside tools they already use.
This is particularly powerful in video.
A generative model can create a clip. A production workflow still involves footage organization, timeline editing, color, sound, revisions, collaboration, brand requirements and final delivery. Adobe’s September Premiere release is explicitly designed around placing generation inside that broader process rather than forcing editors to move their project into a standalone AI generator.
Adobe is pursuing the same strategy at the interface level. Its creative agent is being extended across Premiere, Photoshop, Illustrator, InDesign and Frame.io, while Adobe has also taken creative capabilities into ChatGPT, Claude and other AI platforms. That means Adobe is increasingly willing to meet users wherever their intent originates rather than insisting the Adobe application itself must always be the starting point.
That matters after Astra.
OpenAI describes GPT-6 Astra as a major advance in professional work, software engineering and computer use, capabilities that contributed to renewed investor anxiety about traditional software this week. But a world where conversational AI increasingly becomes the interface does not necessarily mean Adobe disappears. Adobe can try to become one of the execution systems underneath that interface.
The psychology around the stock could change quickly if investors see evidence that this is happening.
Adobe’s valuation already reflects substantial skepticism. Based on the figures supplied ahead of the quarter, its NTM normalized P/E has compressed from 19.88x in May 2025 to 9.82x, while NTM EV/revenue has fallen from 7.34x to 3.74x.
At those levels, Adobe does not necessarily need to convince investors that AI is harmless. It needs to reduce the probability investors assign to outright structural disruption.
If third-party AI usage increases Adobe’s monetization rather than bypassing it, the moat debate changes from model ownership to workflow ownership.
That would be a much more consequential upside surprise than another two-cent EPS beat.
The Downside Surprise: Usage Grows, Economics Do Not
The bearish pre-mortem is almost the mirror image.
Adobe could report perfectly respectable revenue and EPS while giving investors fresh reasons to worry about the next several years. The warning sign would be continued explosive AI engagement without comparable progress in paid AI monetization.
Management already told investors that pushing traffic toward freemium products will hurt near-term ARR. That is acceptable only if those users ultimately convert, consume credits, move into higher-value workflows or expand into Creative Cloud.
The dangerous outcome would be Adobe repeatedly reporting larger MAU numbers while asking investors to wait longer for the economics. That is the key downside tension in this Adobe earnings preview.
Firefly users who convert are reportedly highly engaged and show meaningful credit consumption, but management said on the previous earnings call that the broader financial payoff from its more aggressive freemium strategy would play out through fiscal 2027. That creates a timing problem. Every quarter between now and then becomes another test of whether management can show leading indicators strong enough to justify the ARR being sacrificed today.
There is also an uncomfortable second-order risk in Adobe’s model-agnostic strategy.
Giving customers access to Google, OpenAI, Runway, Kling and Luma makes Adobe’s workflow more attractive. But it also implicitly acknowledges that Adobe may not own the model customers prefer. If third-party models capture more of the underlying economics while Adobe cannot charge enough for orchestration, the company could end up hosting increasingly valuable AI functionality without capturing a proportionate share of its value.
The market would then view generative credits less as a new growth engine and more as a mechanism for defending the existing subscription base.
Leadership uncertainty could amplify that interpretation. Chakravarthy does not become CEO until December 1, while Shantanu Narayen moves to executive chair. Adobe also disclosed that Creativity and Productivity President David Wadhwani will step down on September 27. That transition arrives precisely when Adobe must decide how aggressively to reshape pricing, product distribution and investment around AI.
The downside therefore is not necessarily a disastrous quarter.
It is a respectable quarter followed by language suggesting that monetization remains several quarters away, Creative Cloud optimization remains deferred and AI investment must rise faster than revenue.
The real bear case is not that creators stop using Adobe tomorrow; it is that AI makes Adobe’s workflow more expensive to defend and harder to monetize incrementally.
At a time when investors have become extremely sensitive to software disintermediation, that could keep valuation multiples compressed even if reported earnings remain solid.
Beyond Tonight: The Next Two Calls Matter More
The next phase of the Adobe story is about proving that its AI architecture produces a durable economic flywheel.
The first signal is conversion. Adobe’s creative freemium MAU has already surged above 90 million. Acrobat and Express MAU exceeded 850 million last quarter. The question over the next six to twelve months is whether those increasingly AI-driven audiences translate into faster Firefly ARR, Acrobat AI monetization and eventually broader Creative Cloud relationships.
The second is consumption economics.
Adobe is moving from a world dominated by subscription seats toward a mixture of subscription, credits and potentially outcome-based monetization. Agent usage is already designed to consume Adobe credits. If users increasingly delegate multi-step workflows to agents, credit consumption could become an important expansion mechanism rather than simply another feature bundled into existing subscriptions.
The third is whether Adobe’s model-agnostic architecture actually improves its competitive position. That question sits at the heart of this Adobe earnings preview because it determines whether outside AI innovation becomes a threat to Adobe or an input into its own platform economics.
Management does not appear to be trying to wall off the ecosystem. Adobe is putting rival models inside Firefly and Premiere while simultaneously putting Adobe capabilities inside ChatGPT and other conversational interfaces. It is a deliberate bet that distribution works both ways: Adobe can import the best intelligence while exporting its professional tools.
The fourth is enterprise adoption. Adobe GenStudio ARR grew more than 25% year over year last quarter, while AI-first ARR in Customer Experience Orchestration grew fourfold. Chakravarthy built much of that enterprise business before being selected as CEO, making the interaction between creative content, marketing workflows, customer data and agentic orchestration particularly important to watch under his leadership.
And finally, investors need to watch what happens to Creative Cloud itself.
Stable retention alongside accelerating Firefly and agent usage would suggest Adobe is expanding the market without damaging its premium franchise. Weakening Creative Cloud economics accompanied by strong free usage would point toward cannibalization instead.
The next two quarters need to show that Adobe’s bigger funnel eventually produces a bigger wallet, not simply a bigger audience.
That is the bridge management now has to build between AI adoption and fiscal 2027 growth.
Adobe need not win the model race; it must monetize the workflow.
The Interpretation Matters More Than The Beat
Adobe enters this earnings report in an unusual position. Its financial fundamentals remain healthy, but its valuation increasingly resembles a business facing structural uncertainty.
At approximately 9.8x NTM normalized earnings, 3.7x NTM revenue and a roughly 10.5% NTM levered free-cash-flow yield based on September 9 data, the market has already compressed Adobe’s multiples dramatically from where they stood little more than a year ago. That means tonight’s reaction may have less to do with whether revenue is $30 million above or below consensus and more to do with whether investors alter the terminal assumptions behind those multiples.
OpenAI’s Astra makes the question more urgent, but not necessarily simpler.
Frontier models could absorb more creative functionality and weaken the value of standalone applications. They could also proliferate so rapidly that professional users increasingly need a trusted platform capable of bringing those models together, applying precision tools around them and carrying work from generation through final production.
Adobe is clearly betting on the second outcome.
Its decision to integrate competing models, expose Adobe functionality through external AI assistants and monetize generative activity through credits suggests management understands that the moat cannot simply be “we built the model.” The emerging moat has to be we own the workflow in which the models become useful.
Tonight can tell investors whether that strategy is beginning to translate from product architecture into financial evidence. That is ultimately what this Adobe earnings preview is designed to test.
The headline revenue and EPS figures will matter. But over the next several quarters, the metric worth following is much simpler: whether AI-first ARR begins growing fast enough to prove that Adobe can monetize the new creative stack as effectively as it monetized the old one.
Disclaimer: We do not hold any positions in the above stock(s). Read our full disclaimer here.
