Enterprises are shifting their focus from raw GPU capacity to distributed, secure cloud infrastructure for AI-native applications. According to Vultr CMO Kevin Cochrane, the initial phase of AI infrastructure prioritized massive centralized clusters for a small group of buyers, neglecting the broader needs of enterprise developers.
Vultr is positioning itself as a full-stack cloud provider by utilizing AMD silicon and open software stacks. The company aims to address the complexity of global-scale workloads by emphasizing decentralized systems and open standards, moving away from the centralized models that dominated the early AI computing cycle.
To improve inference efficiency, Vultr is collaborating with Vast Data Inc. on a global namespace that enables KV cache sharing across nodes. This approach addresses the need for private, isolated networks between data centers, which Cochrane identifies as a primary barrier for enterprises adopting new cloud infrastructure.
Vultr has joined the Cloud Native Computing Foundation as a platinum member to support the open-source ecosystem. The company also launched an Open Source Credits Program, which provides infrastructure resources to maintainers of CNCF projects to facilitate the development of new AI technologies.
