Highlights:

  • Founded in 2017 and based in San Francisco, Weights and Biases has secured over USD 250 million in funding from investors.
  • CoreWeave assured that existing Weights and Biases customers will retain the flexibility to deploy their AI models on any infrastructure, though they may be encouraged to explore CoreWeave’s cloud services.

Cloud-backed AI infrastructure startup CoreWeave Inc. planned to buy Weight and Biases Inc., an AI model development company for an undisclosed amount.

The acquisition was confirmed just minutes after it was reported that CoreWeave was in talks to purchase Weights and Biases in a deal valued at USD 1.7 billion.

CoreWeave has become one of the most sought-after companies in the AI industry by providing enterprises with access to a critical resource: the graphics processing units (GPUs) that power modern AI services. Its cloud infrastructure enables enterprise customers, including Meta Platforms Inc., to access Nvidia’s high-performance GPUs, such as the H100 and H200 chips, which are optimized for large language models and other AI workloads.

Founded in 2017 and based in San Francisco, Weights and Biases has secured over USD 250 million in funding from investors. Its platform is designed to accelerate AI model development, offering tools for managing training datasets, analyzing model performance, and troubleshooting technical issues.

Weights and Biases also provides advanced features, including tools designed to address “hallucinations”—errors occasionally generated by large language models. The platform has gained significant traction, with the company reporting more than one million AI engineers using its tools, including professionals from OpenAI, Meta, Nvidia Corp., Snowflake Inc., and Toyota Motor Co.

CoreWeave plans to integrate Weights and Biases’ capabilities into its cloud infrastructure, creating a more seamless, end-to-end AI experience. Customers will not only be able to run their models on CoreWeave’s cloud but also build and test them within the same ecosystem. Together, the two companies aim to help enterprises accelerate AI development and bring new applications to market more quickly.

CoreWeave assured that existing Weights and Biases customers will retain the flexibility to deploy their AI models on any infrastructure, though they may be encouraged to explore CoreWeave’s cloud services.

Holger Mueller of Constellation Research said that CoreWeave’s leadership understands many enterprises prefer an all-in-one AI solution, streamlining both model development and deployment within a single platform.

“They want a turnkey cloud platform that lets them build and operate their next-generation AI applications and do everything associated with them in one place, and that’s what CoreWeave aims to give them,” he added.

CoreWeave Co-founder and CEO Michael Intrator praised Weights and Biases as an exceptional platform that enables companies of all sizes to navigate the complexities of AI model development, deployment, and performance monitoring.

“Together with CoreWeave, we will bring this grit and passion for innovation to customers at an even greater scale, with the goal of rapidly accelerating adoption across the world’s leading AI labs and enterprises,” promised Weights and Biases CEO Lukas Biewald.

The acquisition is contingent on “customary closing conditions” and is expected to be completed in the first half of the year. If finalized as planned, it will add momentum to CoreWeave’s efforts to go public on the Nasdaq stock exchange later this year.

Recently, the company announced that it had filed the necessary paperwork for its initial public offering, with plans to trade under the ticker symbol “CRWV.”

CoreWeave’s IPO filing highlighted its rapid growth, reporting a remarkable 700% increase in revenue to USD 1.92 billion in fiscal 2024. The company anticipates an even faster expansion, with more than USD 15 billion in signed contracts yet to be fulfilled. However, despite its revenue surge, CoreWeave remains unprofitable, posting a loss of USD 863.4 million for the year.