G2 just ran the numbers on AI image generation, and the results read like a plot twist nobody saw coming.
The team behind the State of AI Image Generation report analyzed 2,111 verified reviews and surveyed seven leading vendors, Recraft included. Everyone assumed this category would be won in enterprise boardrooms. Turns out it's being won in bedrooms-turned-studios and browser tabs at 11 p.m.
Seventy percent of reviewers are individual users or small businesses. Enterprise buyers make up just 14%, which means the "future of work" crowd is individual contributors and small business owners. If you're a one-person Etsy shop or a five-person studio, you don't have a design department on retainer. You have a deadline, a prompt box, and the quiet confidence of someone who has never once made a mood board in Keynote.
Eighty percent of small businesses that tracked their return on investment saw it within six months. Individual users weren't far behind at 67%. For context, that's faster than most gym memberships pay off, and a lot less guilt-inducing in month two.
Every good story needs friction, and here it's pricing. Credit limits and prompt accuracy top the list of buyer complaints, which checks out: the more you iterate, the faster the meter spins, like a taxi stuck in traffic that only you can see. Vendors feel it too. Six of seven told G2 that the real costs hide in iteration time, post-editing, and prompt optimization, not the sticker price.
And yet the same six of seven vendors say AI image generation is still significantly cheaper than traditional design. So the math holds. The bill just shows up in a different outfit than you expected.
The most interesting finding in the report isn't a chart. It's a mood shift. For years, the industry chased better pixels, as if resolution alone could win an argument with an art director. Now the conversation has moved to something a lot harder to fake: judgment.
Recraft CEO Anna Veronika Dorogush put it plainly in her interview for the report, saying the next generation of tools "will be defined by taste, aesthetics, and personalization." Technically correct images stopped being impressive around the same time everyone's uncle learned to use a chatbot. Knowing what makes an image visually strong, distinctive, and unmistakably a brand's own, that's the actual assignment now.
Honestly, that was always the assignment. A hundred passable options were never the goal. Knowing which one is right was.
Zoom out and a pattern emerges: buyers want speed and affordability, vendors are building for scale and personalization, and the real opportunity sits in the overlap, not the extremes.
Recraft has been circling that overlap since day one, and the Vector Editor is a good example, with editable paths and colors instead of a finished image you just have to accept. The category is still young enough that nobody's finished sketching the full picture, and that's the part worth sticking around for.
Full report: State of AI Image Generation in 2026, G2
According to G2's analysis of 2,111 verified reviews, 70% of AI image generation users are individual users or small businesses. Enterprise buyers represent just 14% of the market. The category is being driven by independent creators, freelancers, and small studios, not large design departments.
G2's data shows that 80% of small businesses that tracked their return on investment saw it within six months. Individual users reported similar results, with 67% reaching ROI in the same window. That's a faster payoff cycle than most software categories.
The shift is from technical accuracy toward taste, aesthetics, and personalization. Recraft CEO Anna Veronika Dorogush, quoted in the G2 report, described the next generation of tools as being defined by those three qualities. Generating technically correct images is no longer the benchmark. The competitive edge now sits in tools that help users produce work that's visually distinctive and consistent with a specific brand identity.
.svg)
