OpenAI made a quiet but significant move this month: it restructured how enterprise teams pay for Codex, shifting from fixed seat pricing to a pay-as-you-go model that removes the financial barrier for organisations that want AI coding assistance without committing to a full subscription for every team member.

The change matters beyond the pricing mechanics. The adoption numbers behind it tell a more important story.

What Changed

Teams on ChatGPT Business and Enterprise can now add Codex-only seats to their workspaces with no monthly seat fee. Usage is billed purely on token consumption, meaning teams pay for what they use rather than buying a fixed allocation upfront.

Alongside this, OpenAI cut the annual price of ChatGPT Business from $25 to $20 per seat. For a team of 50, that is a $3,000 reduction in annual software spend — not transformative on its own, but a signal that OpenAI is competing hard for enterprise wallet share.

To accelerate adoption, OpenAI is also offering $100 in credits for each new Codex-only team member who joins and starts using the tool, up to $500 per team.

The Numbers Behind the Move

The pricing restructure is not driven by generosity. It reflects where demand is already going.

More than 2 million builders now use Codex every week. The number of Codex users within ChatGPT Business and Enterprise grew 6x between January and April 2026. More than 9 million paying business users now rely on ChatGPT for work.

OpenAI is lowering the entry cost because it has already validated that teams who start using Codex tend to deepen their usage. The pay-as-you-go model is designed to remove the “I am not sure if we need this” hesitation and let usage prove the value.

Why This Is the Real Inflection Point for AI Coding in Business

For most of 2024 and 2025, AI coding tools were positioned primarily as a developer productivity play. You saved hours on boilerplate code, you shipped features faster, you reduced context switching. The value was real but narrow.

What has changed in 2026 is who is using these tools. The 6x growth in Codex users inside enterprise ChatGPT accounts is not all software engineers. It includes data analysts building automation scripts, operations teams writing SQL queries, finance professionals building Excel macros they can now prompt in plain language, and project managers generating code to connect tools that never talked to each other.

AI coding is no longer just a developer tool. It is becoming a fluency requirement for knowledge workers who want to work with data.

What This Means for Business

The cost of experimentation just dropped. With pay-as-you-go pricing and $100 per-seat credits, a 10-person team can now trial Codex in a real workflow for close to zero cost. There is no longer a meaningful financial reason to delay evaluation.

The skills gap is closing — slowly. The 6x growth in Codex users suggests a lot of non-developers are beginning to interact with code through AI interfaces. This is healthy for businesses but creates a new challenge: knowing how to verify, test, and extend what the AI produces requires a baseline of data and technical literacy. Tools get more accessible; the judgment needed to use them well does not get easier.

The “AI is too expensive” objection is weakening. A common reason businesses have delayed AI adoption is that enterprise software feels expensive relative to uncertain ROI. A $20/seat tool that a team can trial on a token consumption model changes that calculation. The question is no longer whether you can afford to try it but whether your team has the skills to get genuine value from it.

Vendor consolidation risk is real. OpenAI is pricing aggressively to capture enterprise adoption while competitors like Anthropic, Google, and GitHub Copilot are all competing in the same space. Businesses that make decisions purely on short-term cost will find themselves renegotiating contracts as the market matures. The better question is which tool fits how your specific team actually works.

The EDNA Perspective

At Enterprise DNA, we have watched the data skills gap for a long time. For years, the limiting factor was access to tools. Most organisations could not justify the cost of enterprise analytics software for more than a handful of people.

That calculation has now inverted. The tools are accessible. ChatGPT, Codex, and a dozen other AI assistants are available to anyone with a browser. What businesses are discovering is that access without understanding creates noise. Teams generate outputs they cannot validate, automate processes they do not fully understand, and build dependencies on AI decisions without the judgment to know when the AI is wrong.

The companies that will capture the most value from the shift to AI-assisted work are the ones investing in data literacy alongside tool adoption. Not instead of tools, but alongside them.

If your team is starting to use AI coding tools, that is a good first step. The next step is making sure your people understand enough about data, logic, and process to direct the AI well and catch its mistakes. That is what separates teams that get a productivity bump from teams that genuinely operate differently.