NetApp acquires AI data-infrastructure startup DataPelago to process data in the storage layer

NetApp said on July 16, 2026, that it acquired DataPelago, a Mountain View, California, startup whose technology pushes AI data processing directly into the storage layer. Financial terms were not disclosed.

The deal targets a specific bottleneck: moving data. DataPelago’s platform, called Nucleus, processes data where it already sits — at the storage layer, using CPUs and GPUs — rather than shuttling it to external compute clusters. The company says the approach cuts infrastructure costs by up to 80% and processes data up to 10 times faster than legacy approaches, vendor figures NetApp has not independently substantiated.

DataPelago will operate as a wholly owned NetApp subsidiary. The purchase folds a compute-in-storage architecture into NetApp’s core business of selling enterprise data infrastructure, at a moment when the cost of feeding data to AI models is becoming a line item executives scrutinize.

George Kurian, NetApp’s chief executive, said enterprises need data infrastructure as intelligent and powerful as the increasingly capable AI models and chips it feeds. Rajan Goyal, DataPelago’s founder and CEO, said joining NetApp pairs the startup’s processing technology with NetApp’s data-infrastructure portfolio.

The strategic logic is that as model training and inference scale, the expense and latency of copying data between storage and compute grow with them. Whether processing data in place delivers the claimed savings across real enterprise workloads — not benchmark conditions — is what NetApp now has to prove to customers.

The integration timeline and how Nucleus will be packaged for existing NetApp customers were not disclosed.