IBM has launched instances with Nvidia H100 GPUs on its cloud platform.
Customers will now be able to use the GPUs for artificial intelligence (AI) workloads, including training and inferencing.
The addition of H100s follows IBM's deployment of Nvidia A100 GPUs last year. According to Nvidia, the new GPUs enable inference performance of up to 30x faster than the A100.
As well as the A100s and H100s, IBM Cloud also offers Nvidia L40S and L4 GPUs which can be used for smaller AI workloads such as training small-scale models or deploying chatbots.
The H100 instances are available in IBM's multi-zone regions in North America, Latin America, Europe, Japan, and Australia. IBM plans to offer Intel's Gaudi 3 AI processors via its IBM Cloud in early 2025.
The launch of H100 GPUs on IBM Cloud puts the platform a long way behind competitors which have had the GPUs available for over a year in some cases.
Amazon Web Services (AWS) began offering the GPUs in July 2023, while Google, Microsoft, and Oracle all launched H100s in September of that year.
Similarly, GPU cloud start-ups have long had access, with CoreWeave and Cirrascale among those getting first dibs.
Oracle, Amazon, Microsoft, and Google all announced they would be deploying Nvidia's upcoming Blackwell line of GPUs in March 2024, with CoreWeave stating the same a few months later. The Blackwell line faced some design issues earlier this year, but was resolved by the end of August 2024. Nvidia expects to ship "several billion dollars" worth of the latest GPUs in Q4 2024.
Last month The Register reported that IBM Cloud was going through a quiet "massive layoff" program, citing a source that said it was done in secret with employees required to sign NDAs and not talk about the specifics.
IBM told The Register in a statement: "Early this year, IBM disclosed a workforce rebalancing charge that would represent a very low single-digit percentage of IBM’s global workforce, and we still expect to exit 2024 at roughly the same level of employment as we entered with."
Based on IBM's approximate 288,000 employees worldwide, one percent - the minimum meeting 'low single-digit percentage' would represent 2,880 layoffs.
Last year IBM's CEO Arvind Krishna said that the company expected to replace around 7,800 jobs with AI.