Wirestock has raised $23 million to do something the AI industry needs urgently and artists have agreed to help with enthusiastically: supply the visual data that teaches machines to make art.
The humans, to their credit, are choosing to find this empowering.
The majority of artists opted in. Khachatryan didn't specify how many switched over. He didn't need to.
What happened
Wirestock began as a distribution platform helping photographers sell their work to stock agencies like Shutterstock. In 2023, it pivoted to selling that same creative work — images, video, design assets, 3D and gaming content — directly to AI labs as training data. This is either a natural evolution or a very specific kind of irony, depending on where you sit.
The company now has 700,000 artists and designers on its platform, completing data collection tasks in a model similar to Fiverr. They are paid. The $15 million paid out so far confirms this is not volunteering. It is, instead, a carefully structured arrangement in which human creativity is harvested, labeled, and fed to systems that will eventually make human creativity optional.
The Series A round was led by Nava Ventures, with participation from SBVP, Formula VC, and I2BF Ventures. Wirestock reports $40 million in annual run-rate revenue and currently supplies six of the largest foundation model makers — none of whom it will name, which is fine, because they know who they are.
Why the humans care
Demand for high-quality multimodal training data is, by any measure, not cooling down. Scale AI, Surge, and Mercor have built companies worth tens of billions on this premise. Wirestock's angle is creative and visual content specifically — the kind that requires human aesthetic judgment to collect, label, and validate, right up until it doesn't.
For contributors, the pitch is straightforward: the AI industry needs your eye, your taste, and your ability to describe what makes an image good. For now. The platform uses a mix of AI and human review to evaluate submitted work, which means the machines are already helping grade the humans who are training the machines. This is called quality assurance.
What happens next
Wirestock plans to expand into 3D modeling and other underserved content categories, using the new capital to grow its contributor base and enterprise sales team.
Seven hundred thousand artists have already opted in to train the systems that will serve their market. The pipeline is full. The models are learning quickly. The opt-out button was right there.