If you have been wondering whether the AI adoption gap between leading companies and everyone else is real, OpenAI just published data that answers the question clearly. It is real, it is measurable, and it is getting wider every quarter.

On May 7, OpenAI introduced B2B Signals — a recurring report that measures how AI is actually diffusing across businesses, based on privacy-preserving, aggregated data from enterprise use of OpenAI products. This is not survey data asking executives what they plan to do. This is telemetry from what companies are actually doing.

The headline finding: frontier firms — the ones treating AI as a core operating capability rather than a productivity tool — now use 3.5 times as much intelligence per worker as typical firms. Twelve months ago, that gap was 2x. The companies ahead are pulling away.

What “Intelligence Per Worker” Actually Means

OpenAI defines intelligence per worker as a composite of how frequently employees use AI, which capabilities they use, and how deeply those capabilities are integrated into actual workflows. A sales rep who uses ChatGPT to rewrite emails counts for less than an analyst running automated research workflows, which counts for less than an engineering team running autonomous Codex agents on their production codebase.

The 3.5x gap is not mainly about whether a company has turned on Copilot. It is about whether AI is embedded in the highest-value work.

The most striking data point: frontier firms send 16 times as many Codex messages per worker as typical firms. Codex is OpenAI’s AI software engineering agent. This tells you that the frontier-firm gap is sharpest in agentic, autonomous work — not just AI-assisted tasks where a human stays in the loop, but AI doing structured work independently.

That distinction matters. Agentic AI is where the real productivity gains compound. If frontier firms are 16x ahead on that specific capability right now, the operational gap will be far larger in 12 to 18 months once those agents have had time to accumulate domain knowledge and handle more complex work.

The Gap Went From 2x to 3.5x in One Year

One year ago, OpenAI’s Signals data showed frontier firms using roughly 2x as much AI intelligence per worker as typical firms. That number is now 3.5x. The gap has not stabilized — it has accelerated.

This is worth sitting with. It means the companies that were ahead in April 2025 did not coast. They kept investing, kept deploying, kept expanding what AI handles inside their organizations. Meanwhile, typical firms continued using AI at the individual-productivity level: drafting emails, summarizing documents, answering questions.

There is nothing wrong with those uses. But they do not compound. They do not change how the organization operates. And they do not close the gap with frontier firms — because frontier firms moved past those uses a year ago.

What Frontier Firms Are Actually Doing

From OpenAI’s B2B Signals data and prior research, the pattern is consistent across industries. Frontier firms tend to share several characteristics:

They run AI on processes, not just tasks. Rather than having individuals use AI when they remember to, frontier firms have embedded AI into how their core processes work. Customer inquiries, financial reporting, product research, code deployment — AI handles defined steps in these workflows automatically.

They use agentic AI for multi-step work. This is where the Codex gap shows up. Instead of asking AI one question at a time, frontier firms have agents that can plan, execute, and verify multi-step tasks without constant human supervision.

They measure AI impact on outcomes, not activity. The question is not “how many employees use AI?” but “what outcomes did AI produce this quarter, and what does deployment look like next quarter?”

Why the Gap Will Keep Widening

The data shows the gap growing from 2x to 3.5x in 12 months. There is no structural reason to expect it to stop.

The most advanced AI applications — autonomous agents, process automation, knowledge management — are getting easier to deploy. Barriers to entry are falling for tools like Claude, GPT, and Gemini. But the gap is still widening because the companies ahead are deploying new capabilities faster than typical firms are catching up to old ones.

Frontier firms are already running multi-agent workflows. Typical firms are still debating which AI tool to standardise on.

The compounding nature of AI adoption means early movers accumulate advantages: AI-optimised data, trained institutional knowledge baked into agent workflows, teams skilled at working alongside automated systems. These advantages do not disappear when a competitor finally starts deploying AI. They need to be rebuilt from scratch.

What This Means for Business

The B2B Signals report should change how business leaders think about their AI timeline. The question is no longer “should we invest in AI?” Every company of any sophistication has answered that question. The real questions are:

How far behind are we? If your company is at the “typical firm” level — individual tools, no agentic workflows, no AI embedded in core processes — you are operating at roughly 3.5x less AI capability per worker than your most advanced competitors. That gap will affect speed, cost structure, and eventually quality of output.

What does it take to become a frontier firm? The data suggests it requires three things: strategic clarity about which processes AI should own, deployment of agentic (not just assistive) AI in those processes, and organizational capability to manage and improve AI performance over time.

What is the cost of waiting? If the gap went from 2x to 3.5x in 12 months, another 12 months could push it to 5x or 6x. The organizations most likely to close the gap are those that start now, not those that wait for the technology to mature further.

---

Enterprise DNA works with businesses at every stage of this curve — from teams building their first data literacy foundations through EDNA Learn, to organisations deploying autonomous AI agent workforces through Omni Ops and getting strategic clarity through Omni Advisory.

The B2B Signals data confirms what we have seen firsthand: the firms pulling ahead are not doing anything exotic. They made a decision to move from AI tools to AI operations, and they started doing it 12 to 18 months ago.

The question for every business leader reading this is which side of the 3.5x gap you want to be on in another 12 months.

Talk to our team about where your business sits on the AI adoption curve.