OpenAI's Enterprise Spend Controls Arrive as Microsoft Eyes a DeepSeek Escape Hatch

Frequently asked

What do OpenAI's new Enterprise spend controls actually let administrators do that they couldn't before?

The June 18 update added per-group and per-user credit caps, cross-product burn visibility covering both ChatGPT and Codex in one console, and a Cost API for programmatic budget integration. Previously, enterprise admins had aggregate usage data but no mechanism to enforce spending limits at the team or individual level before costs accumulated.

Why would Microsoft use DeepSeek when it has deep OpenAI investment?

Cost. Microsoft's public weighing of a fine-tuned DeepSeek V4 for Copilot Cowork is a signal that enterprise inference costs have reached a threshold where even a strategically committed partner will pressure-test cheaper alternatives. The investment relationship with OpenAI does not insulate Microsoft from the economics its own customers are demanding.

How does the AI-driven memory shortage affect enterprise AI procurement decisions now?

Apple's Tim Cook publicly attributed upcoming price increases to AI-driven memory scarcity — a signal that infrastructure costs are bleeding into consumer hardware pricing. For enterprise buyers, this confirms that AI total cost of ownership extends beyond model licensing into underlying hardware constraints, making spend controls and cheaper model alternatives more urgent, not optional.

More on this wire