NetApp announced on July 16, 2026 that it has acquired DataPelago, a California-based AI data infrastructure company that built a fundamentally different approach to getting enterprise data ready for AI workloads. The deal positions NetApp as the infrastructure layer where data processing actually happens, rather than just the place where data sits waiting to be moved somewhere else.
This is a quiet acquisition that has significant implications for anyone running AI at scale.
What DataPelago Actually Does
DataPelago’s core technology is called Nucleus. It is a universal data processing engine that uses heterogeneous accelerated computing across CPUs and GPUs to process data at the storage layer, rather than copying it into a separate compute cluster first.
That distinction matters more than it sounds. The conventional approach to AI data preparation looks like this: data lives in storage, you copy it to a processing cluster, you transform it, you copy it again to wherever your AI workload runs. Each copy adds latency, cost, and opportunity for error. Large enterprises with petabyte-scale data environments spend enormous amounts of time and money just moving data around before the actual AI work begins.
Nucleus eliminates most of those copies. Processing happens where the data already lives, which DataPelago says delivers infrastructure cost reductions of up to 80 percent and performance improvements of up to 10 times compared with conventional pipelines.
NetApp is calling this capability “zero-copy activation of enterprise data for AI.” Following the acquisition, DataPelago will operate as a wholly owned subsidiary of NetApp.
Why This Acquisition Matters Now
The AI model quality problem is largely solved. You can get frontier-level reasoning from half a dozen providers at declining costs. The problem that remains for most enterprises is data readiness.
Ask any data team running AI in production and they will tell you the same thing: the models are fine, the pipelines are the hard part. Cleaning, transforming, and delivering data to AI workloads at the speed and scale those workloads demand is where projects actually stall.
DataPelago addresses this directly by collapsing the distance between where data lives and where processing happens. For organisations running high-volume AI workflows on structured data — financial reports, transaction logs, operational databases, manufacturing sensor feeds — that kind of infrastructure efficiency improvement is not incremental. It is a redesign of how the cost structure works.
The timing also reflects where the market is heading. As AI agents move from assistants to autonomous workflows, the volume of data they need to access and process scales dramatically. Infrastructure that worked for occasional human-triggered queries does not work when agents are running thousands of continuous cycles.
What This Means for Business
If you are a data professional managing infrastructure for AI workloads, this acquisition signals that the market is moving toward compute-at-storage architectures rather than compute-alongside-storage. The old model of centralising data before processing it is becoming a bottleneck, and vendors are starting to design around that reality.
For business leaders evaluating AI infrastructure investments, the key question is whether your current data pipeline design is limiting the performance or economics of your AI deployments. If your team is spending significant time on data movement and transformation before AI workloads can run, the architectural approach DataPelago represents is worth understanding.
For NetApp customers specifically, expect DataPelago’s Nucleus technology to be integrated into NetApp’s broader storage and data management portfolio over the coming quarters.
The broader pattern is one we have tracked throughout 2026: AI capability is commoditising at the model layer, so the durable competitive advantages are shifting to infrastructure, data quality, and the operational systems that keep AI workloads running reliably. NetApp is making a clear bet on owning the infrastructure layer where that competition will play out.
Enterprise DNA’s data education platform exists precisely for moments like this. The data professionals who understand accelerated compute architectures, storage-layer processing, and AI data pipeline design are the ones who will evaluate, implement, and optimise these systems. If your team needs to build that understanding, our courses in data engineering and AI infrastructure provide the foundation.
Source
NetApp Newsroom