AWS and IBM Considering Major GPU Cloud Partnership

IBM is seeking additional Nvidia GPUs, and Amazon's cloud division, AWS, is in discussions to meet this need through a significant cloud deal.

AWS is reportedly negotiating a substantial contract estimated at $475 million over five years, aimed at providing IBM with Nvidia GPUs via its cloud infrastructure, according to internal Amazon information shared with Business Insider.

Details of the Proposed Agreement

Under the envisioned agreement, IBM would make use of AWS's EC2 servers featuring Nvidia’s AI processors. IBM's research division has already begun training several AI models on these servers, utilizing AWS's SageMaker, an advanced machine learning platform.

According to a source familiar with the negotiations, the discussions focus on AI model training, and while still in progress, there is no guarantee they will culminate in a finalized deal. Currently, both AWS and IBM have not publicly commented on the negotiations.

Implications for AWS and IBM

This potential deal could signify a substantial enhancement of the AI collaboration between AWS and IBM. Earlier in the year, IBM disclosed plans to expand its use of AWS services for its Watson AI initiative.

These negotiations also underscore the continuing strong demand for Nvidia GPUs. AWS has created its proprietary AI chips, Trainium and Inferentia, and is promoting them to cloud consumers as cost-effective and energy-efficient options. However, it remains uncertain if these proprietary chips would feature in any deal with IBM.

AWS’s Growing AI Market

If successful, this cloud agreement could further propel AWS’s AI segment. As per conversation with analysts last month, AWS expects to achieve 'multi-billion dollar' revenue in AI this year, evidencing growth at 'triple-digit rates' annually. Notably, Amazon reports that its AI division is expanding thrice as fast as Amazon Web Services grew initially.

AWS's impressive financial performance last quarter included revenue totaling $27.5 billion, an increase of 19% compared to the previous year. The company is poised to invest upwards of $75 billion in capital expenses this year, focusing largely on enhancing cloud infrastructure like GPUs, networks, and data centers.