Cloud compute infrastructure provider CoreWeave Inc. today announced that it has secured $2.3 billion in debt financing, which the company said it intends to commit entirely to purchasing more hardware used for artificial intelligence training.

Magnetar Capital led the funding round Coatue, DigitalBridge, BlackRock, PIMCO, Carlyle and Great Elm also participated in the funding round. Magnetar also led the company’s Series B and its extension, which raised $421 million between them, bringing the total raised by the company before this debt financing to $571 million.

With the explosion of interest in generative artificial intelligence, the drive to secure bigger and better AI models has led to the need for more cloud infrastructure to train them, which requires enormous amounts of compute and graphics processing units to train the models.

CoreWeave began as an Ethereum cryptocurrency mine in 2017, which used Nvidia Corp. GPUs to process Ethereum blockchain transactions at large scale in order to earn crypto tokens. The company used racks of GPUs in data centers to earn the currency. In the past few years, the company changed its strategy to focus on general-purpose computing and generative AI technologies.

With the power of Nvidia GPUs under the hood, CoreWeave can outperform other cloud providers when it comes to specialized use cases, especially when it comes to AI-specific needs, the company said.

Since its release in late 2022, OpenAI LP’s generative AI chatbot ChatGPT has shown the world that generative AI models are capable of amazing feats, including understanding human speech and replying with humanlike expressiveness. Image generating models such as DALL-E 2, Stable Diffusion and Midjourney can create photorealistic and vivid artwork based on text prompts.

“It’s undeniable that AI has the potential to solve massive global problems – from enhanced productivity and increased access to education, to accelerated development of cancer treatments and beyond,” said Michael Intrator, co-founder and chief executive of CoreWeave. “This explosive growth of AI and machine learning has triggered an equally explosive demand for the infrastructure needed to support it.”

CoreWeave delivers access to Nvidia’s most powerful in high-performance datacenter GPU technology including the Nvidia H100 Tensor Core, which part of the Nvidia HGX H100 platform, via cloud clusters. Additionally, CoreWeave customers can also access other Nvidia cards such as A100, A40s and RTX A6000s as well.

Customers can make use of these Nvidia GPUs for use cases such as AI and machine learning, visual effects and rendering, transcoding and pixel streaming. CoreWeave boasts that it has one of the largest HGX clusters in the world and supports more than 3,500 Nvidia H100 GPUs in its supercomputer infrastructure.

Last month, CoreWeave announced an upcoming data center in Plano, Texas, that will become fully operational by the end of the year. The 450,000-square-foot facility is valued at $1.6 billion and will become the company’s first facility in Texas.

Image: CoreWeave