Trending Articles
Trending Topics
Trending
IBM has announced the private technical preview of GPU-accelerated query processing for watsonx.data, designed to help enterprises tackle the rising cost of analytics and AI.
The announcement comes as organisations continue to face increasing pressure from large-scale analytics and AI workloads, with IBM noting that traditional CPU-based systems are struggling to keep pace with growing data demands. According to the company, this has led to slower insights, rising infrastructure costs and mounting pressure on data teams.
In a joint announcement, IBM watsonx.data Product Manager Margo Harrell and Program Director Hemant Suri explained the offering works by moving compute-intensive query operations, such as joins, aggregations and filtering, from CPUs to GPUs.
“By leveraging massively parallel processing analytical workloads can execute drastically faster than traditional approaches, while significantly reducing infrastructure usage and cost,” they said.
The pair added that faster queries could help organisations make quicker decisions, while lower compute requirements would reduce the cost per query and allow analytics and AI workloads to scale without increasing infrastructure costs.
According to IBM, the capability is fully transparent to users, meaning existing queries, data formats and connectors can continue to operate without modification.
“One of the biggest barriers to modernisation has always been complexity. Many performance solutions require rewriting SQL, migrating data or rearchitecting pipelines. With GPU acceleration that trade-off disappears,” Harrell and Suri said.
IBM added that the solution is designed for hybrid environments, allowing integration across cloud and on-premises deployments while maintaining open-source Presto compatibility and enterprise-grade governance.
The announcement follows a showcase at Nvidia GTC, where IBM, Nvidia and Nestle demonstrated up to five-times faster analytics workloads while reducing costs by as much as 83% through GPU acceleration and open data architecture.
IBM also revealed it had internally deployed the technology through its CIO organisation, acting as a “Client Zero” for GPU-accelerated watsonx.data. IBM boasted that in early deployments, teams achieved up to 25 times faster query performance compared to CPU-only execution, while reducing workload costs by approximately 80% due to reduced runtime.
The company said it is now expanding access to the capability through its private technical preview program, allowing select clients to work directly with product and engineering teams and gain early access to the technology.
