Adobe’s second-quarter results answered one of the most immediate questions surrounding its artificial-intelligence strategy. Customers are paying. Revenue reached a record $6.62 billion in the quarter ended 29 May 2026, up 13 per cent year on year, and AI-first annualised recurring revenue more than tripled to exceed $500 million. Total annualised recurring revenue reached $27.10 billion, including about $480 million from Semrush.
For chair and chief executive Shantanu Narayen, those numbers offer evidence that generative AI is becoming a commercial product rather than remaining a costly feature programme. Adobe raised its full-year revenue and non-GAAP earnings targets. It now expects fiscal 2026 revenue of $26.50 billion to $26.60 billion and ending recurring-revenue growth of 10.2 per cent.
The more consequential question remains unresolved. Does AI strengthen Adobe’s position by making its tools more useful to more people, or does it reduce the value of specialist creative software by making competent output available through a prompt? The company is exposed to both possibilities. Its applications, customer relationships, content workflows and intellectual property give it formidable distribution. The same technology that enhances those assets lowers barriers for new competitors and allows users to substitute automated generation for some traditional work.
Narayen’s 2026 leadership challenge is to manage that contradiction without retreating into defensive bundling. Adobe must make AI good enough to change customer behaviour, price it in a way that reflects value and preserve the professional control, rights clarity and interoperability that distinguish a production platform from a novelty generator. It must also widen its audience without alienating the creators and enterprises that fund the core business.
The record quarter gives Narayen room to act. GAAP operating income was $2.24 billion, non-GAAP operating income was $2.95 billion and operating cash flow was $2.17 billion. Subscription revenue reached $6.42 billion. Those economics can support model development, infrastructure, acquisitions and aggressive product experimentation. They also raise the standard: investors will expect AI expenditure to improve growth, not merely defend existing revenue.
The moat is workflow, not generation
Text-to-image and text-to-video models have made generation widely available. Adobe cannot assume that the quality of a single model will remain a durable advantage. Models improve quickly, open alternatives spread and customers often use several providers. The stronger position lies in the workflow around generation: ideation, editing, brand control, collaboration, asset management, measurement and delivery.
This is where Adobe’s installed base matters. A designer may generate an image, but still needs precise layers, colour management, typography, rights information and export controls. A marketing team needs variations that follow brand rules, connect to approved assets and can be measured across campaigns. A document worker needs output that fits an existing process and can be reviewed, secured and signed. AI creates value when it removes friction from those tasks without taking away control.
Narayen should therefore resist evaluating AI products mainly by the volume of generated content. The more useful metrics are completion time, retention, paid conversion, expansion across applications and the proportion of output that moves into a production workflow. High generation volume can be an infrastructure cost rather than a business advantage if users do not finish or distribute what they create.
Adobe’s opportunity is to make its applications the place where machine-generated material becomes professionally usable. That requires strong links between Firefly and Creative Cloud, between Acrobat and business workflows, and between content creation and the customer-experience products that distribute and measure it. The company should allow customers to choose models where appropriate while maintaining a consistent layer of permissions, provenance and editing.
Pricing must reveal value
Generative AI has a different cost structure from conventional desktop software. Each request can consume compute, and advanced video generation is particularly intensive. Unlimited use at a fixed subscription price may stimulate adoption but weaken unit economics. Restrictive credits can frustrate users and make the product feel like a meter rather than a creative partner.
Adobe needs pricing that separates exploration from high-value production. Entry-level access can encourage new customers, while professional and enterprise tiers should reflect speed, model quality, commercial assurances, collaboration and governance. Usage charges are reasonable where infrastructure costs are material, but customers should be able to predict them. Unexpected limits would undermine the trust created by familiar subscriptions.
The emerging AI-first ARR figure is useful, but it must be interpreted carefully. Revenue may come from stand-alone products, premium usage or additions to existing contracts. Narayen should ensure that internal reporting distinguishes new demand from revenue reclassified as AI. The strategic objective is incremental value: new customers, higher retention, more seats or paid usage that would not otherwise exist.
Adobe’s updated full-year targets suggest management sees enough momentum to raise expectations. The company should still avoid using short-term demand as evidence that every AI investment will earn an attractive return. Model training, inference and acquisition costs need product-level accountability. The discipline that produced a non-GAAP operating margin near 45 per cent can coexist with investment if management understands which uses are becoming more efficient as they scale.
Trust is a product feature
Professional customers care about how models are trained, whether generated material can be used commercially and how synthetic content is identified. These concerns are not peripheral legal matters. They influence whether a global brand, film studio, publisher or public agency will put AI output into production.
Adobe has positioned Firefly around commercially safer training sources and has supported content credentials that record information about creation and editing. Narayen should continue treating provenance as infrastructure. Credentials need to survive export and distribution, integrate with other tools and remain understandable to users. The company should be transparent about what a credential proves and what it does not.
Rights clarity is equally important. Enterprise customers require contractual assurances, controls over the use of their data and options to train or customise models within approved boundaries. Independent creators want confidence that their work is not being absorbed without consent and that AI does not reduce the market for original skill to anonymous input. Adobe cannot resolve every conflict in the creative economy, but it can set product rules that respect ownership and make attribution practical.
The company also has to govern the misuse of its tools. Synthetic media can enable fraud, impersonation and deceptive political content. Overly broad restrictions may suppress legitimate expression, while weak safeguards can damage the brand and invite regulation. A mature approach combines model controls, user accountability, provenance and cooperation with platforms where content is distributed.
Semrush broadens the strategic canvas
Adobe’s inclusion of Semrush in second-quarter results signals a wider ambition in marketing. Semrush brings search, traffic and digital-presence data that can connect content creation to discovery and commercial performance. The combination can help business customers decide what to create, produce it efficiently, distribute it and measure whether it works.
The integration opportunity is real, but so is the risk of portfolio complexity. Adobe already spans creative software, documents and enterprise customer experience. Adding another substantial platform can produce cross-selling, or it can create overlapping interfaces and sales motions. Narayen should define a small number of workflows where Semrush data materially improves Adobe products rather than trying to place every capability into every suite.
One promising direction is the changing nature of search. Consumers increasingly receive answers from AI systems rather than lists of links. Brands need to understand how their information appears in those answers, how authority is assessed and how content should be structured for discovery. Semrush can give Adobe analytical depth in that transition, while Adobe can provide the creation and governance tools.
Integration discipline will also protect margins. Adobe used about $1.56 billion of cash for acquisitions in the second quarter, and Semrush contributed about $480 million to total ARR at the end of the period. The value of the transaction will depend on retention, cross-selling and product leverage, not the speed with which logos and account systems are combined.
Leadership depth matters in a transition
The departure of chief financial officer Dan Durn in June 2026, with long-serving Adobe finance executive Steve Day becoming interim CFO, adds an organisational test. A strategic shift involving new cost structures, acquisitions and changing revenue models requires strong financial leadership. Narayen must ensure that the search for a permanent CFO is not treated as routine succession.
The next finance chief needs to understand both subscription economics and usage-based AI services. Capital allocation will involve choices among model development, infrastructure agreements, acquisitions and share repurchases. During the second quarter Adobe repurchased about 8.5 million shares, while cash and cash equivalents ended the period at $4.92 billion. Shareholder returns remain appropriate, but they should not limit investments that can establish defensible AI workflows.
Narayen also needs leadership below the executive suite that can resolve conflicts between established products and new ones. Firefly should not be required to protect every legacy interface, and Creative Cloud teams should not be able to treat AI as an external feature supplier. Product groups need common objectives around customer outcomes, even when those outcomes change where revenue is recorded.
Expanding the market without eroding the profession
Adobe’s largest AI opportunity may come from people who never considered themselves creators. Business professionals can now make a credible presentation, edit a short video or prepare campaign assets without mastering specialist tools. This expansion is central to the 16 per cent growth in subscription revenue from business professionals and consumers in the second quarter.
Professional users may interpret the same trend as commoditisation. Narayen should not dismiss that concern. Adobe’s long-term franchise depends on skilled creators who push the tools, establish standards and produce the work that others value. The company needs a product ladder in which AI helps occasional users begin while giving professionals deeper control, automation and the ability to build reusable systems.
Done well, democratisation can increase demand for expertise. More organisations producing content need stronger art direction, brand governance and differentiation. Done poorly, it can flood channels with undistinguished material and reduce willingness to pay. Adobe cannot determine the cultural outcome, but its tools can favour intentional creation over volume alone.
The second-quarter results show that Narayen has avoided the immediate danger of AI paralysis. Adobe is shipping products, customers are adopting them and revenue is growing. The next stage is more exacting. Growth must come with credible economics, trusted rights practices and workflows that become harder to replace as models become easier to access.
Adobe’s moat was never simply the ability to manipulate pixels or documents. It was the role its formats and applications played in turning ideas into professional output. Narayen’s task is to make AI deepen that role. If he succeeds, generation will be an entry point into Adobe’s system rather than an alternative to it. Record revenue will then represent more than a successful quarter; it will mark the renewal of a platform for a different way of working.
