Coreweave Gets First Nvidia GB300 AI Servers From Dell

Key Concepts:

• Nvidia's latest generation chips (Blackwell Ultra, Rubin series)
• CoreWeave's advantage in receiving Nvidia chips first
• Backlog conversion to revenue
• Training vs. Inferencing
• Dell's role in server manufacturing
• Infrastructure supercycle

Nvidia Chip Distribution and CoreWeave's Advantage

The discussion centers around the strategic rollout of Nvidia's latest generation server designs, specifically focusing on the Blackwell Ultra chip and the upcoming Rubin series. CoreWeave is highlighted as the initial recipient of these advanced chips. This early access provides CoreWeave with a significant competitive advantage over other hyperscalers. The Blackwell Ultra is expected to deliver a 50% performance upgrade in token overpriced testing.

CoreWeave's Backlog and Revenue Conversion

CoreWeave possesses a substantial backlog of approximately $26 billion. Access to the most advanced Nvidia chips, particularly the Blackwell Ultra, is crucial for accelerating the conversion of this backlog into revenue. The latest Blackwell Ultra chips are expected to be primarily utilized for training purposes, while older series chips will be used for inferencing over time.

Dell's Role in the Ecosystem

Dell is identified as the manufacturer responsible for assembling the servers that house the Nvidia chips. Nvidia provides the chips, and CoreWeave brings the racks and servers online for cloud consumption. Dell's involvement in the supply chain is considered a positive development, as it allows them to secure this role before other server manufacturers, such as Quanta or Chinese OEMs, can enter the market.

Infrastructure Supercycle

Oracle and OpenAI's $30 billion contract announcement is cited as evidence of a broader "infrastructure supercycle" that is currently unfolding. This supercycle is characterized by significant investments in infrastructure to support the growing demands of AI and other compute-intensive applications.