IBM Cloud has become the first cloud service provider to make Intel Gaudi 3 artificial intelligence accelerators available to customers, marking a significant step toward expanding access to high-performance artificial intelligence infrastructure. The integration is designed to address the high costs typically associated with deploying specialized artificial intelligence hardware and to provide enterprise customers with more choices for cost-effective generative artificial intelligence (GenAI) solutions.
This deployment represents Intel´s first major commercial rollout of Gaudi 3, signaling the introduction of competitive options to a market historically dominated by other GPU providers. The companies aim to help clients across industries—including financial services, healthcare, life sciences, and the public sector—test, innovate, and deploy artificial intelligence solutions at a lower total cost. IBM Cloud is extending Gaudi 3 availability in key regions such as Frankfurt, Washington D.C., and Dallas, and the accelerators will be accessible through IBM Virtual Private Cloud and Virtual Servers, with support for Red Hat OpenShift and IBM´s watsonx artificial intelligence platform expected soon.
Backed by recent Signal65 benchmark tests commissioned by Intel, Gaudi 3 has demonstrated marked cost efficiency and performance gains. When running Meta´s Llama-3.1-405B-Instruct-FP8 model with large context sizes, Gaudi 3 was found to be 92% more cost efficient in terms of performance per dollar compared to competitors. On smaller workloads with the IBM Granite-3.1-8B-Instruct model, it delivered 43% more tokens per second, and on large context tasks, achieved a 36% increase in throughput over alternative solutions. These metrics are especially relevant for enterprises scaling up artificial intelligence adoption as they seek lower operation costs and higher output in their workloads.
Enterprise users in banking, insurance, healthcare, and retail rely on IBM Cloud for problems such as fraud detection, personalized service, drug discovery, diagnostics, and inventory management. With Gaudi 3, these industries gain accelerated artificial intelligence model training, fine-tuning, and inferencing capabilities. This move allows customers to modernize technology stacks securely while reaping the benefits of the latest advancements in artificial intelligence hardware platforms.
