Enterprise AI Cost Controls
OpenAI Introduces Global Admin Console for ChatGPT Enterprise to Track AI Spending
Key Points
Zipline has been using Codex since January 2026, and its co-founder Ryan Oksenhorn said the company recently expanded Codex adoption “to the whole company,” asking OpenAI to build usage analytics to help them “find and train-u…” (as quoted in OpenAI’s post).
OpenAI said it previously introduced per-role credit limits for custom roles earlier in 2026, and Thursday’s update extends that approach with three tiers: workspace-wide defaults, group-level limits, and individual overrides.
Employees requesting more credits are expected to provide context—OpenAI’s update includes a way to submit a request with a note explaining what additional capacity is needed.
The rollout was tied to OpenAI’s blog post dated June 18, 2026, with Reuters reporting that enterprise customers could begin using the new analytics and spend controls starting Thursday.
OpenAI on Thursday rolled out new credit tracking and spending controls for ChatGPT Enterprise, giving company administrators a single dashboard to monitor and cap AI costs across their entire workforce. The update, confirmed by Reuters, makes the tools available to all ChatGPT Enterprise and Edu customers starting June 19, 2026.
The change comes as companies face growing "bill shock" from AI tools. Internal reports cited by analysts at Intellectia.AI suggest some large enterprises, including Uber, burned through their annual AI budget in just a few months due to uncontrolled usage of agentic AI features.
One Dashboard Now Shows All ChatGPT and Codex Spending
The centerpiece of the update is an upgraded Global Admin Console. For the first time, spending on ChatGPT — for general tasks — and Codex — for coding and automation — appears in a single view, according to Let's Data Science. Admins can break down usage by individual user, product, and AI model. They can also track trends over time to spot the heaviest spenders before costs spiral.
OpenAI also launched a unified Cost API alongside the dashboard. This lets finance and IT teams pull spending data automatically into corporate software systems like SAP or Oracle, according to Cryptopolitan. The goal is to let companies build automated "chargeback" pipelines — a system where AI costs get billed back to the specific team or department that ran them.
Three-Tier System Lets Bosses Set Limits by Team or Person
Admins now have three levels of control. First, they can set a default spending cap for the whole company. Second, they can assign different limits to specific groups — for example, giving the engineering team a higher cap than the marketing team. Third, they can create individual overrides for power users who need more capacity, according to Headtopics.
Employees who hit their limit are not simply cut off. They can submit a request for more credits — but they must include a note explaining why they need them. OpenAI's design forces workers to justify the extra spend before an admin approves it. Employees can also view their own usage at any time, so they can manage their pace before hitting a wall.
Zipline Asked OpenAI to Build This After Going Company-Wide on Codex
Drone logistics firm Zipline was one of the companies that pushed OpenAI to build these tools. Zipline began using Codex in January 2026, and co-founder Ryan Oksenhorn said the company soon expanded it "to the whole company." He said his team asked OpenAI to create usage analytics "to help find and train-up folks who haven't adopted Codex, and for granular usage controls to keep spend predictable," according to Reuters.
Oksenhorn added that the new tools are "helping us faster scale productivity of our employees while keeping safeguards in place." His comments reflect a broader shift. Companies are no longer just asking whether AI is useful — they are asking how to make sure every employee uses it, and how to stop any one team from blowing the budget.
Credit-Based Pricing Is About to Become the New Standard
Thursday's update is not just a one-off feature drop. OpenAI has signaled that credit-based pricing for all agentic AI workflows will become the standard by July 6, 2026, ending many "unlimited" trial periods previously given to Enterprise and Business users, according to Economic Times. The base cost for ChatGPT Enterprise typically runs $45–$75 per user per month, with a minimum of 150 seats — roughly $108,000 per year at minimum.
Analysts compare the moment to the early days of cloud computing. "Visibility was the only way to stop bill shock" when AWS launched its cost monitoring tools, according to researchers at Intellectia.AI. The new Cost API is expected to fuel a wave of third-party "AIOps" tools — software that manages AI budgets across multiple vendors like OpenAI, Anthropic, and Google — all at once.
Sentiment
Mostly PositivePublishers
15
Articles
38
Reach
53
