NTT DATA is not a startup experimenting with AI on the side. It is one of the world's largest IT services firms, with close to 200,000 employees globally and annual revenue above $30 billion. So when it reports that Codex completed a complex technical task in 30 minutes that previously took five engineers three days, that number carries real weight.
To understand what happened, it helps to know what Codex actually does. Unlike a regular AI assistant that answers questions, Codex can receive a goal, investigate the problem independently, run processes, test its own work, and revise it, all without a human directing each step. NTT DATA used it to automate a full incident analysis for a critical system. That kind of work normally requires experienced engineers who know the system well, understand what went wrong, and can read through logs and data to trace the cause. Codex did it in half an hour.
The result spread quickly inside the company. Senior leaders took notice, and NTT DATA accelerated the rollout. Today, Codex is live across roughly 9,000 employees, and importantly, many of those users are not engineers at all. Non-technical staff are using it to pull data from credit card statements and populate expense forms, build simple internal tools, analyze data in spreadsheets, and automate repetitive file-management tasks. Work that previously required a request to the IT department.
This is the part that matters most for any business operator reading this. The old pattern in professional services was simple: more revenue required more people. You won a bigger contract, you hired more staff to deliver it. AI is breaking that link directly. NTT DATA said so plainly: the company is moving from growing by adding headcount to growing by doing more with AI.
That shift is not unique to NTT DATA. Across the IT services sector, AI is compressing what used to take teams of people into tasks a single employee with the right tool can handle. One industry analysis described it as "the biggest operating model rewrite since offshoring." The companies that built their growth on selling human hours are now having to explain to clients why they should pay the same price for work that takes a fraction of the time.
For NTT DATA's clients, which include banks, insurers, manufacturers, and governments across 70 countries, this matters in two ways. First, if their service provider can deliver the same output faster, they should expect prices to come down or scope to go up. Second, the tools NTT DATA is using internally are the same tools it is offering to its clients, meaning the efficiency gains can flow into the client's own operations.
Non-developers now account for about 20% of all Codex users globally, and that group is growing three times faster than the engineering users. That pattern confirms what NTT DATA is already seeing: the value of this kind of AI is not limited to technical teams.
The governance question is real. NTT DATA built clear rules before expanding access: which data can be used, which systems Codex can connect to, what requires a human to review before anything is finalized. That scaffolding matters. Tools that can take independent action inside a company need boundaries, and companies that skip this step tend to create problems that are harder to fix than the time saved was worth.
The practical takeaway is this: if a 200,000-person company built on selling human expertise is actively redesigning its business around AI doing the work, every organization that buys services from companies like NTT DATA, or competes with them, is already inside this story whether they realize it or not.