Farooque Munshi, Partner at EY. Leads Data/AI for Advanced Manufacturing across the Americas. Focused on turning AI ambition into outcomes.
Three years ago, if you asked a Fortune 500 manufacturer who owned AI, you’d get a shrug, a vague gesture toward IT and maybe the name of a vice president of analytics four levels below the CEO. That answer doesn’t survive a board meeting anymore.
The titles vary. Chief data and AI officer, chief AI officer, chief enterprise AI and data officer. The title is the least interesting thing about the job. What’s interesting is what the person holding it is now expected to deliver and how badly that mandate has diverged from the role, even since 2022.
What The Job Actually Looks Like Now
There are four things on the plate, and the leaders I’ve watched succeed don’t get to set any of them down:
1. A Number: A real number tied to gross margin or yield or working capital, committed to the board and validated by finance. The era of the AI pilot that lives in a slide deck and never touches the P&L is closing. CFOs want attribution, and they want it from finance, not from the team that built the model.
2. The Data Foundation: This is the part nobody at the board level wants to hear about. Historian data, MES, ERP, supplier feeds, product telemetry, the whole unglamorous plumbing—none of the AI work on a factory floor functions without it, and most of the public failures trace back to someone trying to skip this step.
3. The Operating Model. What the central platform team owns, what gets pushed out to embedded data scientists, how you handle model risk and how you actually upskill the engineers and commercial people who have to use this stuff. That last piece is where most of these roles get stuck.
4. Being The Face Of The Thing: Board, customers, regulators and the workforce. That last audience is the one nobody talks about enough. In a manufacturing context, with union shops and a workforce that has watched automation cycles come through before, you don’t get to wing the conversation about what AI means for people’s jobs.
More and more of these leaders sit under a CEO, COO or senior engineering SVP, not a CIO. The job has stopped being about supporting IT delivery and started being about owning business outcomes that happen to need data and AI to work.
Three Appointments Worth Looking At
What follows are anonymized renderings of three real appointments at large manufacturers. The backgrounds and approaches paints a clear picture.
The Product-First External Hire
A legacy industrial manufacturer created its top AI role for the first time last year and went outside to fill it. The hire had two decades of AI work, mostly at enterprise tech firms and large platform companies, and heavy product orientation. The interesting thing wasn’t the résumé. It was the placement inside software and engineering, reporting to a senior engineering executive instead of IT. AI is treated as a product capability inside engineering rather than an analytics layer draped over the business.
The ‘Data Foundation First’ Leader
A century-old global manufacturer’s senior-most data and AI executive had done similar work at several other industrial companies before landing there. Their strategy starts from a thesis almost out of fashion: The data foundation matters more than the model. Under their leadership, the company rolled out a GenAI platform for its global workforce and built an internal academy that has trained thousands.
The use cases are deliberately unsexy. Quality issues surfaced from front-line feedback before they became recalls. Computer vision cleared up confusing engineering documentation on the line. They’ve said publicly that change management and adoption are bigger constraints than anything the model is doing.
The Platform Builder In A Regulated Business
A diversified industrial appointed a new data and AI leader into one of its regulated divisions, hiring externally from a technology-intensive manufacturer where they’d built integrated cloud platforms spanning data, AI, ML and security. In a regulated division, you can’t separate the AI agenda from the platform agenda. If the underlying platform isn’t auditable, the AI work doesn’t ship.
The three came in from different places, but all three are being treated as business leaders who need AI fluency, not technical leaders who need to learn the business.
The Hard Part
The hardest part isn’t the technology or the talent market (although the latter’s been brutal). It’s the gap between what the board’s been reading in The Wall Street Journal and what’s actually possible when you’re deploying these systems on shop floor equipment from the 1990s in unionized plants across regulatory regimes that don’t agree.
Leaders who over-promise in year one tend to be quietly gone by year two. Leaders who under-promise get accused of lacking ambition and tend to be gone by year three. The ones who last are those who educate upward constantly. They tell the board what’s possible this year versus next, where the company is actually ahead or behind and what the leading indicators look like before the financial impact shows up. It’s tedious work. It’s also the work.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
