Menu

Fueling starving GPUs to power AI workloads

Proact IT Sweden bridges the AI infrastructure gap with help from NetApp.

Sunset view of Stockholm

Impressive numbers

1994

Founded in 1994, Proact IT offers proven experience in managing and protecting data

1200 experts

supporting clients in 12 countries across Europe and North America

+ 30 data centers

and extensive IT infrastructure capabilities

Proact

Proact IT Sweden

Country / Region

Sweden

Table of Contents

Share this page

Quick highlights

As organizations race to adopt AI, they often invest heavily in compute power while overlooking the critical storage infrastructure needed to feed AI/ML projects. Proact IT, Europe’s leading independent data center and cloud services provider, partnered with NetApp to solve this “starving GPU” problem for clients in diverse industries. By implementing a balanced stack of NetApp® solutions, Proact IT linked data scientists, developers, and IT operations, delivering high-performance, secure, and scalable environments for mission-critical workloads.

Since 1994, Proact IT has stood at the forefront of digital innovation in Europe, helping organizations manage their rapidly growing data needs. Today, the landscape is shifting. From autonomous driving software to massive medical imaging projects, the demand for high-performance AI infrastructure is exploding.

Buying hardware is relatively straightforward. The real challenge lies in implementation and communication. Charles Morrall, AI Architect at Proact IT, notes that successful AI projects require more than just powerful processors; they require a unified strategy that links the data science and IT teams. Without this connection, organizations risk building expensive silos that fail to deliver ROI.

I meet many data scientists and developers and spend time learning their language. When I put them in the same room as the IT staff, it sparks conversations. They can come together and talk things out, which connects their projects to the data in ways that wouldn't happen otherwise.

Charles Morrall, AI Architect, Proact IT

Four colleagues in a conference room

Identifying the challenge

For many Proact IT clients, the primary challenge isn’t a lack of ambition or budget; it’s a disconnect between compute and storage needs.

“Some customers like to ramp up with NVIDIA GPU machines, but they forget the storage part,” explains Morrall. This leads to the “starving GPU” phenomenon: Organizations invest in expensive compute resources, only to have them sit idle because the storage layer can’t feed data fast enough.

Organizational silos compound this issue. In large public sector entities, such as tax authorities, the AI team might be five steps removed from the storage infrastructure team. A request for storage becomes a generic IT ticket, rather than a strategic conversation about performance requirements.

Specific industries also faced unique hurdles:

  • Automotive: Companies investing in autonomous driving needed to process massive design and telemetry files with low latency, making a pure public cloud strategy impractical.
  • Healthcare: Medical providers managed petabytes of pathology and X-ray images, requiring solutions that were both massive in scale and cost-efficient.
  • Highly regulated sectors: Energy companies and defense contractors needed absolute assurances about data protection, ransomware defense, and compliance documentation.

Some want to spend most of the money on the GPUs. I encourage them to build a balanced stack. When compute, networking, and storage are harmonized, you don’t starve the expensive NVIDIA GPUs because you can’t feed them data.

Charles Morrall, AI Architect, Proact IT

Two colleagues conversing in a data center

Finding the right balance

Proact IT tackled these challenges by deploying intelligent data infrastructure to create a balanced stack. Instead of treating storage as a commodity, Morrall and his team used NetApp solutions to build architectures that cater to AI workloads. 

Bridging the human gap

Before installing a single cable, Proact IT focused on breaking down silos. “If I can get IT and AI teams in the same room,” Morrall says, “it’s not a difficult or lengthy discussion.” By facilitating these conversations, Proact IT ensured the infrastructure was designed for the actual workload, not just a generic spec sheet.

Optimizing the tech stack

To address technical requirements, Proact IT used a suite of technologies:

  • Performance storage for AI: For automotive clients, they deployed high-performance, all-flash NetApp AFF A-Series storage systems to handle the latency-sensitive autonomous driving development and large design files.
  • Scalability for healthcare: To manage petabytes of medical imaging data, Proact IT implemented NetApp StorageGRID® object storage. This allowed tiered storage, keeping critical patient data readily available, while storing colder data cost-effectively.
  • Integrated security: For the energy sector, including nuclear power management, Proact IT used NetApp Autonomous Ransomware Protection (ARP/AI) and tamper-proof snapshots. For one client, Proact IT integrated security across mixed Windows and Linux environments, creating a single source of truth for access control using Active Directory and Kerberos network authentication protocols.
  • Navigating VMware licensing changes: Proact IT helped universities and other organizations adopt affordable, open-source virtualization platforms that maintain high performance and flexibility.

Results

By shifting the focus from simple procurement to holistic architecture, Proact IT helped its clients achieve tangible operational success.

Massive scale and stability

For its healthcare client, the implementation of StorageGRID successfully supported an explosion in data volume caused by higher-resolution medical cameras. The system now manages over ten petabytes of data across three hospital sites, maintaining redundancy and high availability for critical patient diagnostics.

Unlocking AI potential 

In the automotive space, the shift to a hybrid model allowed clients to keep heavy design workloads on premises for speed, while retaining cloud flexibility where it made sense. This eliminated the latency bottlenecks that previously hampered productivity. Ultimately, drivers benefit from better safety features and more responsive in-car systems.

Cultural transformation

Perhaps the most significant result is the shift in how these organizations view IT. “It’s at this human level where real progress happens,” Morrall says. By connecting infrastructure staff to data scientists, Proact IT turned ticket-based operations into collaborative partnerships. Clients who previously didn’t know what was possible are now using data management features of NetApp ONTAP® to accelerate their development cycles.

What’s in store for Proact IT?

Proact IT customers demonstrate that successful AI adoption isn’t just about buying the fastest chips; it’s about building a data foundation that can keep up with them. By relying on NetApp to create secure, scalable, and high-performance storage environments, Proact IT ensures their clients never leave their GPUs starving.

As Proact IT looks to future endeavors, including a large-scale life science project requiring hundreds of GPUs and ultra-high-performance storage, its partnership with NetApp remains central to their strategy.

Go further

Ready to elevate your operational game? Discover how a strategic partnership with NetApp can make all the difference.

Drift chat loading