Adobe
Last reviewed
Jun 9, 2026
Sources
29 citations
Review status
Source-backed
Revision
v1 ยท 2,258 words
Improve this article
Add missing citations, update stale details, or suggest a clearer explanation.
Suggest edit
CtrlK
Last reviewed
Jun 9, 2026
Sources
29 citations
Review status
Source-backed
Revision
v1 ยท 2,258 words
Add missing citations, update stale details, or suggest a clearer explanation.
Adobe Inc. is an American software company best known for creative and document tools such as Photoshop, Illustrator, Premiere Pro and Acrobat. Founded in December 1982 and headquartered in San Jose, California, Adobe has become one of the most closely watched incumbents of the generative AI era, launching its Firefly family of image, vector, video and audio models in March 2023 [1]. Its AI strategy rests on three pillars: models trained on licensed and public-domain content that it markets as "commercially safe," deep integration of features such as Generative Fill into Creative Cloud applications [2], and leadership of the Content Authenticity Initiative and the C2PA provenance standard [3]. Adobe's AI period has also included a roughly $20 billion attempt to acquire Figma that collapsed under regulatory pressure in December 2023 [4], an unusual embrace of rival model providers such as OpenAI, Google and Runway inside its own products [5][6], and recurring friction with the artist communities that underpin its business [7].
Adobe was founded by John Warnock and Charles Geschke, two former Xerox PARC researchers, and is led by chair and chief executive Shantanu Narayen, who has held the CEO role since December 2007. The company reports two principal segments: Digital Media, which includes the Creative Cloud and Document Cloud subscription businesses, and Digital Experience, which sells marketing, analytics and content-supply-chain software. In fiscal 2025, which ended in late November 2025, Adobe reported record revenue of $23.77 billion, up 11 percent year over year, and said that what it calls "AI-influenced" annualized recurring revenue (ARR) represented more than one third of its overall book of business [8].
Rather than positioning AI as a standalone product line, Adobe has threaded generation, editing and agent features through nearly all of its applications, while simultaneously selling Firefly as a consumption-based service and reselling third-party frontier models inside its own interfaces [5]. This "commercially safe plus open ecosystem" posture distinguishes it from AI-native rivals such as Midjourney and from platform competitors that ship only their own models.
Adobe's PostScript page-description language helped create desktop publishing in the mid-1980s; Photoshop followed in 1990 and the PDF format in 1993. The company shifted its flagship applications from perpetual licenses to the subscription-based Creative Cloud between 2012 and 2013, and introduced Adobe Sensei, an umbrella brand for machine-learning features such as content-aware editing, in 2016.
In September 2022, Adobe agreed to acquire the collaborative design platform Figma for approximately $20 billion in cash and stock, its largest-ever deal. After more than a year of antitrust review, the companies mutually terminated the agreement on December 18, 2023, citing "no clear path" to approval from the European Commission and the United Kingdom's Competition and Markets Authority; Adobe paid Figma a $1 billion reverse termination fee [4][9]. The collapse left Adobe to pursue growth organically just as generative AI was redrawing the creative-software market, and the Firefly program, announced nine months earlier, became central to that response.
Adobe unveiled Firefly on March 21, 2023, as a public beta focused on text-to-image generation and stylized text effects [1][10]. Unlike most contemporaries, Adobe trained the first Firefly model on Adobe Stock imagery, openly licensed content and public-domain works whose copyright had expired, a dataset choice it promoted as making outputs safe for commercial use [1]. Firefly became generally available on September 13, 2023, when Adobe integrated it across Creative Cloud, Adobe Express and Adobe Experience Cloud and introduced "generative credits," a metered consumption system bundled into subscription tiers [2]. The same day, Adobe paid the first "Firefly bonus" to Adobe Stock contributors whose images had been used in training, a payment weighted by the number of images and the licenses they had generated [11].
For business customers, Adobe announced Firefly for enterprises in June 2023 and offered indemnification: it agreed to cover certain intellectual-property claims arising from Firefly-generated content, a first-of-its-kind assurance intended to ease corporate legal concerns about generative imagery [12][13]. The indemnity covers Firefly outputs themselves, not modifications users add afterward [13].
| Model | Announced | Notes |
|---|---|---|
| Firefly Image 1 | March 21, 2023 (beta) | Text-to-image and text effects; trained on Adobe Stock, openly licensed and public-domain content [1] |
| Firefly Image 2 | October 2023 | Higher fidelity; Generative Match style matching, photo settings, prompt guidance [14] |
| Firefly Image 3 | April 23, 2024 | "Photographic quality"; powered new Generate Image, Generate Background and reference-image tools in Photoshop [15] |
| Firefly Video Model | October 14, 2024 (limited beta); public beta February 2025; generally available April 24, 2025 | Marketed as the industry's first "commercially safe" video model; 1080p clips from text or image prompts; powers Generative Extend in Premiere Pro [16][17][5] |
| Firefly Image Model 4 and 4 Ultra | April 24, 2025 | Improved prompt fidelity and control; released alongside video general availability [5] |
| Firefly Image Model 5 | October 2025 | Native 4-megapixel output and conversational, prompt-based editing [18] |
At Adobe MAX in October 2025 the company also previewed generative audio capabilities, including soundtrack and speech generation in beta, and a Firefly-branded AI video editor, repositioning Firefly as an "all-in-one creative AI studio" [18]. Adobe said users had generated more than 22 billion assets with Firefly by April 2025 [5].
In February 2025 Adobe shipped a standalone Firefly web application built around the new video model [17]. At MAX London on April 24, 2025, it went further and began offering competing third-party models inside Firefly, starting with OpenAI's GPT Image, Google's Imagen 3 and Veo 2, and Black Forest Labs' Flux 1.1 Pro, all selectable beside Adobe's own models [5]. Firefly mobile apps for iOS and Android followed on June 17, 2025, alongside Firefly Boards, a collaborative moodboarding canvas [19].
The partner roster expanded through 2025 and 2026 to include ElevenLabs and Topaz Labs (added at MAX in October 2025) [18], plus models such as Runway's Gen-4 and Aleph, Luma AI's Ray3, Ideogram 3.0, Moonvalley's Marey, Pika, Google's Veo 3.1, Imagen 4 and Gemini 2.5 Flash Image, and Kuaishou's Kling 3.0 [20][21][22]. In December 2025, Adobe added prompt-based video editing to Firefly [20] and signed a multi-year strategic partnership making it Runway's "preferred API creativity partner," with early access to models including Gen-4.5 [6]. In April 2026, Adobe announced Firefly AI Assistant, an agentic interface that plans and executes multi-step creative workflows across Firefly and Creative Cloud apps from conversational prompts, and said its catalog had grown past 30 Adobe and partner models [22]. In early 2026 Adobe also moved paid Firefly plans to unlimited generations with Adobe's own models, retaining credits mainly for premium partner models [23].
Firefly's most visible deployment is inside Adobe's flagship tools. Generative Fill, which inserts, extends or removes image content from a text prompt directly in Photoshop, entered beta on May 23, 2023, and shipped in the general release that September alongside Generative Expand [24][2]. Illustrator gained Generative Recolor, Adobe Express gained text-to-image features, and Premiere Pro gained Generative Extend, which lengthens video clips using the Firefly Video Model [2][16].
On the document side, Acrobat AI Assistant, a conversational tool for querying and summarizing PDFs, became generally available in April 2024 as an add-on subscription starting at $4.99 per month [25]. For marketers, Adobe sells GenStudio, a generative content-production suite in Experience Cloud. The April 2026 Firefly AI Assistant extended this integration push toward agents that orchestrate work across Photoshop, Premiere, Lightroom and Express from a single conversational surface [22].
Adobe founded the Content Authenticity Initiative (CAI) at its MAX conference on November 4, 2019, together with The New York Times Company and Twitter, to develop provenance metadata for digital content [26]. In February 2021 it co-founded the Coalition for Content Provenance and Authenticity (C2PA) with Arm, BBC, Intel, Microsoft and Truepic, merging the CAI's work with the Microsoft- and BBC-led Project Origin to produce a single open technical standard [3]. The resulting Content Credentials system attaches cryptographically signed metadata recording how a piece of content was made and edited, including whether AI was involved.
Adobe applies Content Credentials automatically to Firefly outputs, marking them as AI-generated [2], and has built credential inspection and attachment tools into Creative Cloud apps. The effort has become the de facto industry framework for AI-content labeling, with camera makers, newsrooms, platforms and other AI vendors adopting C2PA metadata, although coverage across the open web remains partial [26].
Adobe monetizes AI through generative credits, standalone Firefly subscriptions, enterprise Firefly deals and AI add-ons. It told investors that ARR from "AI-first" standalone products, principally the Firefly app, Acrobat AI Assistant and GenStudio, exceeded its $250 million target for fiscal 2025 [27], while broader AI-influenced ARR surpassed one third of total business exiting that year [8]; one analysis put the figure near $8 billion, up from roughly $3.5 billion a year earlier [28]. Generative-credit consumption tripled in the fourth quarter of fiscal 2025, a period when Adobe temporarily offered unlimited generations on select plans [8]. In its first quarter of fiscal 2026, reported in March 2026, total ARR reached $26.06 billion, up 10.9 percent year over year, though net-new Digital Media ARR of about $400 million missed analyst expectations, feeding a persistent investor debate over whether generative AI ultimately strengthens Adobe or exposes it to AI-native competitors such as Midjourney, OpenAI's Sora and Google's media models [28].
Relations with creators have been more turbulent. Some Adobe Stock contributors objected that their images trained Firefly without an explicit opt-in, and critics called the Firefly bonus payments modest [11]. A larger controversy erupted in June 2024, when a mandatory re-acceptance of terms of use, quietly amended that February, surfaced language allowing Adobe to access user content "through both automated and manual methods"; many artists read it as permission to train AI on their private work, and the inability to uninstall apps without accepting the terms amplified the backlash [7]. Adobe published clarifications and then rewrote the terms later that month, stating explicitly that customers own their content and that Adobe does not train generative AI on customer content, with the exception of submissions to the Adobe Stock marketplace [29]. The episode, alongside the commercially-safe positioning of Firefly, illustrates the balance Adobe attempts to strike: courting professional creators wary of generative AI while racing to ship it across every product it sells.