CoreWeave: Accelerating AI Infrastructure with NetBox Cloud
Background
CoreWeave has established itself as a leading specialized GPU cloud provider in the AI infrastructure market, experiencing explosive growth by deploying GPU-powered compute at cutting-edge scale and speed.
As of December 31, 2024, their environment included:
- 32 data centers across the United States with recent expansion into the UK and Sweden
- Over 250,000 GPUs deployed across these facilities
- High-density racks consuming over 360 MW of active power
- Complex networking configurations including InfiniBand fabrics
- Custom liquid cooling and power distribution solutions
The Challenge: Scaling at Speed
As demand for its specialized data centers surged with the AI boom in 2023, CoreWeave experienced impressive growth. The team was faced with spinning up multiple data centers concurrently, presenting significant infrastructure management challenges. At the time, CoreWeave was manually spinning up new infrastructure and relying on multiple spreadsheets to document and manage its infrastructure, which was not going to be scalable. CoreWeave’s ability to rapidly deploy new infrastructure would directly impact their bottom line and customer satisfaction.
The Solution: NetBox Cloud as Infrastructure Source-of-Truth
In 2023, CoreWeave’s team adopted NetBox as its scalable, centrally managed infrastructure database. This decision helped CoreWeave establish a reliable foundation for documentation, provisioning, and asset management across its rapidly expanding data center footprint.
After seeing value in NetBox, CoreWeave transitioned to NetBox Cloud, a hosted version of open source NetBox that includes enterprise-grade features for security and scale. The company’s rapid expansion—from 3 data centers in 2022 to 32 by the end of 2024—has been supported by NetBox Cloud as its infrastructure management platform, enabling rapid deployment, extensive automation, and operational efficiency at a massive scale.
Business Impact
NetBox Expertise
Adopting NetBox Cloud eliminated the administrative overhead of self-hosting while maintaining the automation capabilities CoreWeave needed to support their growth. The decision allowed CoreWeave’s engineers to focus on building infrastructure rather than managing NetBox. It also ensured the team always had access to the latest NetBox innovations via fully managed plugins and upgrades.
“The pain of maintaining NetBox, administrating it, upgrading scripts and plugins, that all took away from the important work of spinning up new infrastructure for our customers,” says Marcus Butler, engineering manager for network development at CoreWeave. “Now we can focus on building and managing infrastructure and leave NetBox to the experts.”
Access to the team and resources with NetBox Labs, the commercial stewards of NetBox, has been invaluable to CoreWeave as well. NetBox Labs’ customer success team members frequently share NetBox best practices and answer questions as they arise.
Automation Brings Standardization and Efficiency at Scale
For CoreWeave, each new data center deployment involved thousands of GPUs and intricate networking requirements. NetBox Cloud’s robust platform enabled CoreWeave to establish standardization across their data centers, which ensures consistency in deployments and accelerates the process of bringing new infrastructure online. Leveraging NetBox Cloud’s powerful APIs, CoreWeave fully automated provisioning workflows. This “build once, replicate many times” approach creates a repeatable formula for success.
Jim Julson, director of network at CoreWeave, notes that this repeatability has had implications for the company’s bottom line. “We have dozens of data center builds this year alone. NetBox is crucial for accelerating our timelines, deploying a data center even a month sooner as a result of these efficiencies can directly impact our revenue,” Julson said.
By shortening infrastructure deployment cycles, CoreWeave rapidly brought new capacity online and grew revenue significantly.
Looking Ahead
CoreWeave continues to expand its use of NetBox Cloud as they grow their infrastructure footprint, including expanding internal team adoption of the platform. For example, the compute team is working to leverage NetBox Cloud’s capabilities for their own infrastructure management needs. The team is also exploring new ways to visualize and document their complex data center environments and building out more advanced network automation.
CoreWeave is looking forward to taking advantage of new NetBox Labs offerings, including enhanced discovery and assurance capabilities. “We’re excited about new features that improve our ability to reconcile infrastructure drift and expand automation further,” said one of the engineers. “NetBox continues to evolve alongside our growing needs.”
As CoreWeave continues its rapid expansion, NetBox Cloud will remain a cornerstone of their infrastructure management strategy, supporting their mission to deliver cutting-edge GPU cloud services with cutting-edge speed and efficiency.
