OpenAI, CoreWeave Sign US$12 Billion Deal to Secure GPU Resources

OpenAI has signed a five-year, US$11.9 billion contract with CoreWeave, a company specializing in GPU-based cloud computing services. The deal reflects OpenAI's goal to reduce its technology dependence on Microsoft, its main cloud infrastructure provider.

OpenAI is looking to diversify its infrastructure providers to avoid vulnerabilities associated with dependence on a single player, such as Microsoft, sources tell Reuters. This move aligns with the company's efforts to ensure a stable supply of computing resources, essential for training its AI models.

The relationship between OpenAI and Microsoft has been marked by strategic collaboration, but also by competitive tensions. In January 2023, Microsoft invested over US$10 billion in OpenAI, obtaining exclusive rights to integrate its technologies into services such as Microsoft Azure and Microsoft 365 Copilot. However, this dependence has limited OpenAI's autonomy, exposing it to operational risks in the event of changes in Microsoft's policies or delivery capabilities.

CoreWeave, founded in 2017, has positioned itself as a key player in the cloud computing sector, specializing in Nvidia GPU-based services essential for AI development. With 32 data centers and more than 250,000 GPUs, CoreWeave offers a robust and scalable infrastructure, attracting customers such as Microsoft, which represents 62% of its revenue.

The agreement between OpenAI and CoreWeave includes an US$11.9 billion investment and the acquisition of US$350 million worth of stock as part of CoreWeave's future IPO. This agreement not only grants OpenAI priority access to one of the world's largest GPU pools, but also gives OpenAI strategic influence over CoreWeave's decisions.

This move is part of a broader trend within the AI industry, where GPU shortages have become a critical bottleneck. Sam Altman, CEO, OpenAI, has publicly expressed concern about the lack of hardware resources, noting that "GPU availability is a key determinant in the advancement of AI."

Microsoft has not been idle. The company has accelerated the development of its own advanced language models, such as the "Mai" series, and has bolstered its AI division with the hiring of Mustafa Suleyman, former Co-Founder, DeepMind. These initiatives seek to reduce Microsoft's dependence on OpenAI and compete directly in the market for large language models.

In the long term, according to Wired, this deal could reshape the AI landscape. OpenAI is not only strengthening its position vis-à-vis Microsoft, but also establishes a strategic alliance with a key infrastructure provider. CoreWeave, meanwhile, consolidates its position in the market and prepares for its IPO, with revenue growth from US$228.9 million in 2023 to US$1.9 billion in 2024.