CoreWeave gets first high-end Nvidia AI chips from Dell

CoreWeave has become the first company to receive an AI server system powered by Nvidia’s GB300 NVL72 chip, built by Dell Technologies and set to be deployed in the US.

This development underscores CoreWeave’s commitment to providing cloud computing power for companies that require Nvidia chips for AI software.

The new system may enhance the ability to manage larger and more complex AI models at faster speeds.

Dell, which also supplies servers to xAI, has been expanding its AI server business, contributing to small stock gains for Dell, Nvidia, and CoreWeave.

The new chips, introduced in March 2025, aim to improve upon Nvidia’s existing Grace Blackwell systems.

However, their release coincides with global economic and geopolitical uncertainties that may affect the demand for AI infrastructure, according to analysts.

🔗 Source: Bloomberg

🧠 Food for thought

1️⃣ Cloud providers’ race for AI hardware reflects shifting competitive dynamics

CoreWeave’s position as the first recipient of NVIDIA’s GB300 NVL72 systems highlights the evolving competitive landscape among specialized AI cloud providers.

Unlike general-purpose cloud giants, CoreWeave has strategically focused on rapid GPU deployment specifically for AI workloads, allowing it to gain significant market share despite being a relatively young company1.

This deployment comes amid intensifying competition in the AI infrastructure space, where providers are differentiating through various approaches. CoreWeave emphasizes bare metal access and performance optimization for high-end AI labs, while competitors like Nebius offer more integrated software experiences1.

The announcement reinforces CoreWeave’s close relationship with NVIDIA (which is also an investor), demonstrating how strategic partnerships can provide competitive advantages in the rapidly evolving AI infrastructure market.

CoreWeave’s early access to these chips will likely strengthen its position with key customers like OpenAI, as performance benchmarks show the new platform offers a 10x boost in user responsiveness and 5x improvement in throughput per watt compared to previous generations2.

2️⃣ Advanced AI chips create widening performance gaps in a geopolitically constrained market

The GB300 NVL72 deployment illustrates the growing technological divide in AI infrastructure, where access to cutting-edge chips creates significant competitive advantages.