At Intel Vision 2025 in Las Vegas, held from March 31 to April 1, IBM announced the availability of the Intel Gaudi 3 AI accelerator on IBM Cloud. This milestone marks the public cloud debut of Gaudi 3 for production workloads, enabling enterprises to deploy and scale AI workloads more efficiently and cost-effectively.
The Intel Gaudi 3 AI accelerator is currently accessible in the IBM Cloud Frankfurt (eu-de) and Washington DC (us-east) regions, with planned availability in the Dallas (us-south) region during Q2 2025.
According to IBM’s AI in Action 2024 report, 67% of executives reported a revenue increase of 25% or more through AI adoption. While AI’s business potential is evident, managing infrastructure costs remains a challenge. The integration of Gaudi 3 on IBM Cloud aims to address this, empowering organizations to test, innovate, and scale generative AI with improved cost efficiency.
Also Read: OpenAI Introduced GPT-4.1 with Enhanced Performance
The longstanding collaboration between IBM and Intel continues to focus on providing scalable, flexible solutions that allow clients to dynamically adjust resources while controlling costs and boosting efficiency. Intel Gaudi 3 supports a range of deployment options on IBM Cloud, including:
- Standalone Servers on IBM Cloud VPC
Intel Gaudi 3 is deployable as standalone servers within IBM Cloud Virtual Private Cloud (VPC), a secure and robust environment allowing customers to build isolated private clouds with the benefits of public cloud infrastructure. Clients can customize compute, storage, and networking, with support for Red Hat Enterprise Linux AI image options for greater control over software stacks and workloads. - Container Worker Node Availability
Starting in Q2 2025, Intel Gaudi 3 will be available as a worker node within Red Hat OpenShift AI clusters and Red Hat OpenShift on IBM Cloud, catering to organizations using managed container infrastructure. - Bring Your Own watsonx License
Enterprises seeking full-stack control can deploy IBM watsonx.ai software on Intel Gaudi 3-based virtual servers within IBM Cloud VPC, also expected in Q2 2025. Watsonx.ai offers an end-to-end AI development studio equipped to manage the entire AI lifecycle with flexible deployment options. - Deployable Architectures (DAs)
IBM Cloud’s Deployable Architectures will enable rapid utilization of Gaudi 3’s capabilities across various deployment models. These include DAs for watsonx, IBM Cloud Virtual Server for VPC, and Red Hat OpenShift on IBM Cloud, all anticipated to launch in the second half of 2025.
IBM and Intel remain committed to enhancing secure deployment environments. In a recent collaboration, Intel® Trust Domain Extensions (TDX) became available on IBM Cloud Virtual Server for VPC. This advancement strengthens the Confidential Computing portfolio, offering joint customers enhanced data isolation, confidentiality, and integrity at the virtual server level.
Through this deepened partnership, IBM and Intel aim to accelerate the adoption of high-performance AI workloads with a secure, scalable, and cost-effective cloud infrastructure.
