OpenAI announced plans to acquire Ona, a move designed to expand its AI agent capabilities by adding secure, persistent cloud environments that can support long-running workflows. The acquisition addresses a fundamental limitation in current AI deployment: the inability for agents to maintain state and context across extended operational periods. By integrating Ona's infrastructure, OpenAI aims to enable AI agents to execute complex, multi-step enterprise tasks—such as data processing pipelines, customer service workflows, and operational automation—without losing contextual memory or requiring constant human intervention. This represents a strategic evolution beyond stateless API calls toward genuinely autonomous systems capable of sustained work.
The timing of the Ona acquisition gains weight when contextualized against OpenAI's concurrent enterprise momentum. BBVA, Spain's largest bank, recently scaled ChatGPT Enterprise across 100,000 employees, demonstrating the velocity at which large organizations are embedding OpenAI's models into core operations. That deployment signals demand for production-grade AI tooling, but also exposes limitations: banking workflows require persistent state, audit trails, and the ability to handle long-running processes like loan applications or transaction reconciliation. Ona's cloud infrastructure—built for secure, persistent execution—directly enables the next tier of enterprise automation that customers like BBVA will inevitably demand.
Complementing this infrastructure push, OpenAI simultaneously launched its Academy courses and deepened support for EU AI transparency standards, signaling a company simultaneously building bottom-up (skills training) and top-down (enterprise infrastructure and regulatory credibility) positioning. The Ona acquisition, however, represents the most consequential move: it transforms OpenAI from a model provider into an agent platform provider, closing the gap between prototype and production deployment. For enterprises evaluating AI investments, the ability to run persistent, stateful agents fundamentally changes cost-benefit calculations. OpenAI is not just releasing better models—it's building the operational infrastructure that makes those models actually deployable at enterprise scale.