Wirestock's new $23 million round points to a fast-growing AI problem: labs need better creative data, and they need to prove where it came from.
Wirestock is no longer just a creator marketplace trying to help photographers and designers sell more work. With a fresh $23 million raise, the company is positioning itself as part of the supply chain AI labs increasingly need: rights-cleared, human-made, multimodal data that can train creative models without leaving a legal mess behind.
That may sound like a narrow infrastructure story. It is not. The next stage of generative AI will not be won only by bigger models or more expensive chips. It will also depend on whether companies can get images, video, design, music, and other creative inputs that are high quality, properly licensed, and usable at scale. Scraping the open web was convenient. It was also messy. The market is now moving toward data that comes with consent, provenance, and compensation attached.
According to TechCrunch, Wirestock has raised $23 million to supply creative multimodal data to AI labs. The company says it now works with more than 700,000 creators, has access to more than 50 million assets, and has licensed more than 10 million assets for AI use. Wirestock's own site describes a model built around custom datasets, creator participation, and data that is cleared for AI training with legal protection for customers.
For years, the AI conversation focused on models. GPT, Claude, Gemini, Stable Diffusion, Midjourney, Sora, Runway and others became the visible face of the industry. But behind every model is a less glamorous question: what did it learn from? That question used to be treated as a technical detail. Now it is becoming a commercial and legal issue.
Creative data is especially sensitive because it usually belongs to someone. A photo has a photographer. A song has a composer. A video may include actors, locations, brands, styles, or personal likenesses. The more AI tools become capable of producing commercial-grade media, the more important it becomes to show that the training inputs were not just pulled from wherever they could be found.
This is where Wirestock sees its opening. The company began by helping creators distribute stock content across major marketplaces. That gave it a large creator base and an understanding of the dull but important mechanics of licensing, metadata, review, payments, and rights management. Those are not glamorous assets. In this market, they may be the whole business.
Wirestock has already shifted away from being mainly a traditional stock distribution platform. In January, The AI Journal reported that the company was sunsetting parts of its old stock content distribution service to focus on paid creative projects for AI labs, with Wirestock saying it had paid out $1.6 million to creators in the previous month. That move makes more sense now. The bigger opportunity is not selling another image into a crowded stock marketplace. It is turning creator work into structured data that AI companies can actually use.
Scale AI showed the shape of the prize
The obvious comparison is Scale AI. Scale did not become important because labeling data sounded exciting. It became important because AI companies needed dependable data operations, and the companies building the models did not always want to build that machinery themselves. If creative and multimodal AI follows the same path, a company that can reliably source, clean, license, and deliver creative inputs could become very valuable.
Wirestock is trying to fit that role for creative media. The company says it offers ready-to-use datasets and custom datasets built around specific training goals. For AI teams, that matters because general web data is often too noisy, too risky, or too poorly described. A model that needs to understand product photography, lighting, motion, human gestures, graphic layouts, or music styles needs more than a pile of files. It needs the right files, arranged in the right way, with enough documentation to survive legal and enterprise review.
There is also a creator-rights tension here that cannot be ignored. Wirestock's pitch depends on creators believing that licensing work for AI is a better deal than being scraped without permission. VentureBeat previously reported that payments can range from small bulk licensing payouts to higher fees for specific content needs. That is a practical model, but it also raises a hard question: will creators feel fairly compensated if their work helps train systems that later compete with them?
That question will follow every company in this category. The answer will not come from slogans about ethical AI. It will come from contracts, payout transparency, opt-in controls, audit trails, and whether creators can see a real income stream from participating. If Wirestock can make that work, creator trust becomes more than a marketing point. It becomes inventory.
For AI labs, the attraction is just as clear. Licensed data can reduce legal exposure, help enterprise customers feel safer using generated media tools, and give model builders access to specialized content that competitors may not have. In a market where compute is expensive and talent is scarce, exclusive or higher-quality data can become a real moat.
The $23 million round does not prove Wirestock will become the Scale AI of creative data. It does show that investors are paying attention to the less visible plumbing of generative AI. The companies that control clean, consent-based data supply may not get the flashiest demos, but they could become difficult to replace. Watch where AI labs spend their data budgets next. That will tell us whether this becomes a category or just another funding headline.
Also read: DayOne May Raise $4 Billion as AI Infrastructure Pulls in Venture Capital • HMRC is putting British AI at the center of tax enforcement • India is becoming the first real test of AI job disruption
