TechCapsules

AI-summarised brief · reviewed before publication

Oracle Cloud Infrastructure (OCI) has recently entered into a partnership with xAI, making xAI's Grok models available to its enterprise customers for training and inference. This partnership adds to OCI's existing foundation models from Cohere, Meta, Mistral, and Almawave, as well as customer-supplied custom large language models (LLMs). The Grok models, including Grok 3 and Grok 3 mini, will be hosted on OCI, allowing customers to leverage Nvidia or AMD accelerators for training and inference. Interestingly, xAI also hosts its models on Microsoft Azure, an announcement that predates OCI's partnership by about a month. OCI's approach to enabling AI for its customers is distinct from that of other cloud service providers. Each provider has its own unique strategy, shaped by its roots and culture. For instance, AWS leverages its infrastructure and vast catalog of AI services, along with its custom silicon, such as Graviton and Trainium. This approach is fitting for AWS, given its origins in infrastructure-as-a-service. In contrast, Microsoft Azure's AI enablement seems to focus on productivity and developer tools for the enterprise, with an OpenAI-first approach. OCI, on the other hand, has taken a data-centric approach, which aligns with its strength in enterprise data management. As AI continues to grow in importance for enterprises, understanding the differences in approach between cloud providers will be crucial. OCI's strategy for enabling AI has been paying off, and this partnership with xAI is another step in that direction.