Perceptron Raises $6.5M for AI Data Network
Perceptron is building a contributor network for specialised AI datasets, though the decentralised model remains early.
AI companies do not only need bigger models. They need better data, especially specialised datasets that are hard to scrape from the open web. Perceptron is trying to build a network around that gap.
What happened
Perceptron raised a $6.5 million strategic round to build a decentralised AI data network. The company says its platform lets AI companies commission specialised datasets from a contributor network of more than 800,000 nodes.
The idea is to create a faster way for model developers and AI companies to source niche, task-specific data rather than relying only on public datasets, synthetic data or internal customer data.
Why it matters
Data sourcing is becoming one of the less visible bottlenecks in AI. As foundation models mature, companies increasingly need domain-specific datasets for evaluation, fine-tuning and specialised applications.
Perceptron’s network model is interesting because it treats data collection as infrastructure. The caveat is that decentralised data networks still have hard questions around quality control, incentives, provenance and compliance.
The bigger picture
The next AI infrastructure layer may be about trusted data supply. Companies will need more ways to collect, verify and refresh datasets for specific use cases. Perceptron is early and more speculative than the larger rounds in this batch, but it points to a real market pressure: AI systems are only as useful as the data pipelines behind them.