Wirestock’s $23M Raise Shows AI Labs Are Still Hungry for Human-Created Data
The funding round points to a larger shift in AI: as foundation models become more multimodal, rights-aware creative datasets are turning into strategic infrastructure.
- Wirestock raised $23 million in Series A funding to expand its AI data supply business.
- The company says it now supplies images, video, design assets, gaming content, and 3D data to AI labs.
- The announcement highlights growing demand for licensed, human-created, multimodal training data.
Wirestock, a platform that once focused mainly on helping photographers distribute and sell creative work through stock marketplaces, has raised $23 million to build out a different kind of business: supplying creative multimodal data to AI labs.
The Series A round, reported by TechCrunch, was led by Nava Ventures, with participation from SBVP, Formula VC, and I2BF Ventures. Wirestock says it now works as a data provider for foundation model companies, offering datasets across images, videos, design assets, gaming content, and 3D material.
From stock content to AI data infrastructure
The move reflects a broader market shift. Creative platforms are discovering that their contributor networks and rights-cleared content libraries can become valuable inputs for AI model development. Wirestock says it pivoted toward data supply in 2023 and now has more than 700,000 artists and designers on its platform.
According to TechCrunch, Wirestock CEO and co-founder Mikayel Khachatryan said the company was transparent with creators about the pivot and allowed artists to opt out of the data supply business. The company says “the majority” of creators moved into the new model, although it did not provide a precise number.
Why AI labs want multimodal data
Frontier AI development is no longer only about text. Image generation, video generation, creative assistants, 3D workflows, and future agentic systems need datasets that connect visual, spatial, temporal, and descriptive information. That makes multimodal data harder to source and more valuable than generic web text.
Raw content is also not enough. AI labs often need datasets that are labeled, annotated, filtered, formatted, and legally usable. Wirestock says it has retrained teams to annotate and label data in more detail, while also building enterprise sales capabilities to serve hyperscalers and foundation model companies.
| Area | Why it matters | Potential impact |
|---|---|---|
| Images and video | Core inputs for visual generation and editing models | Better creative model quality and style coverage |
| 3D and gaming assets | Useful for spatial AI, simulation, game development, and robotics-adjacent workflows | More capable multimodal and world-model systems |
| Human annotation | Transforms raw assets into task-ready training or evaluation data | Higher-quality supervised learning and model evaluation |
| Rights-cleared sourcing | Reduces legal and reputational risk for model makers | More defensible AI data pipelines |
The business signal
Wirestock’s raise also shows how data supply is becoming a competitive layer in AI. Compute remains expensive, and model architecture still matters, but data quality increasingly determines whether a model can perform well in specialized tasks. This is especially true for creative AI, where visual quality, prompt alignment, diversity, and rights status all matter.
The company told TechCrunch it currently provides multimodal data to six of the largest foundation model makers, though it did not name them. Wirestock also reported annual run-rate revenue of $40 million and said it has paid $15 million to contributors so far.
Creator upside and unresolved questions
For creators, the model could open a new revenue stream at a time when AI-generated media is disrupting traditional stock content economics. But it also raises difficult questions: how should contributors be paid, how clear is consent, can creators opt out later, and how much long-term value should flow back to people whose work helps train commercial AI systems?
Those questions are likely to become more important as lawsuits, regulation, and public scrutiny push AI companies toward cleaner data provenance. Data vendors that can demonstrate consent, quality control, and licensing discipline may become more attractive partners for AI labs.
What happens next
Wirestock plans to use the funding to hire across research, engineering, and product roles, and to build enterprise software that helps AI labs collaborate on datasets. The company is also exploring additional modalities, including audio and music.
The bigger story is that the AI data market is maturing. The next phase of model competition may depend less on scraping the open web and more on building reliable, specialized, rights-aware supply chains for human-created data.
Comments (0)