Picture this. Your COO approved $3.2 million in AI spend eighteen months ago. Three pilots are running and the predictive maintenance model looks promising on paper. The quality inspection tool gets mentioned in every town hall and the demand forecasting engine generates beautiful dashboards.
But the board sees something entirely different: $3.2 million spent. $0 in documented savings. $0 in measurable revenue impact.
You have two weeks before the next board meeting. And right now, you are not entirely sure how to explain the gap.
What makes this harder to explain is that it isn’t an isolated failure. Grant Thornton’s 2026 AI Impact Survey says, 950 senior leaders across 10 industries, confirms this is a structural, sector-wide problem. And manufacturing is at the centre of it.
Between February 23 & March 18, 2026, Grant Thornton surveyed 950 senior business leaders across 10 industries, including 100 manufacturing executives.
Among manufacturing respondents, zero reported significant revenue uplift from AI initiatives, compared to 12% of executives across all other industries. Zero reported significant cost savings, again versus 12% overall. Nearly half (47%) said AI has delivered only a modest uplift.
Manufacturing has arguably deployed AI more deeply than most sectors, into:
The problem isn’t the technology. It’s what happens, or doesn’t, after the pilot ends.
Even though 48% of manufacturers are piloting AI, only 10% have fully integrated it into operations. Only 39% are scaling AI across multiple functions, ten points below the broader survey average.
The investment is there. The production infrastructure, governance, integration, accountability; largely isn’t.
The single most actionable finding from Grant Thornton’s survey doesn’t make headlines the way the revenue numbers do, but operationally, it’s the more urgent one.
Only 7% of manufacturers have a tested AI playbook. That means 93% of manufacturers running live AI, across quality control, demand forecasting, and predictive maintenance. They have no validated framework for what happens when a model fails, who owns the outcome, or how AI outputs connect to financial results.
Every week without one is a liability the balance sheet hasn’t priced yet.
More than three-quarters of executives (78%) lack confidence their organization could pass an independent AI governance audit within 90 days. Only 22% of operations leaders have a fully developed AI strategy.
The pattern is familiar: buy the platform, run the pilot, wait for the vendor to explain what comes next. That loop doesn’t produce returns, it produces a longer list of stalled initiatives.
Understanding the revenue gap requires looking at what happens between a successful pilot and a production-grade deployment. Three structural barriers consistently prevent the transition.
AI systems don’t operate in isolation. They need to connect with Manufacturing Execution Systems, ERP platforms, and plant-floor data sources that were built across different decades, by different vendors, on incompatible architectures. As Tezo’s analysis of operational latency in manufacturing shows, even accurate AI recommendations can take 3.5 hours to reach the shop floor when manual re-entry between systems is required. That lag eats the ROI before it can be measured.
Three in four boards have approved major AI investments. Only 52% have set clear governance expectations around them. Just 54% have integrated AI risk into ongoing oversight.
When ownership is unclear, models drift quietly. No alerts fire. The first sign something is wrong is often a production anomaly or a missed forecast, by which point the cost is already real.
Boards approve AI investments expecting margin impact. Operations teams report back on adoption rates, models deployed, and hours saved. That measurement mismatch, between what’s being tracked and what the business needs to improve, is where most ROI conversations quietly break down.
The manufacturers closing the gap aren’t doing anything exotic. They’re doing the unglamorous work of building the infrastructure that makes AI accountable.
Every production AI deployment needs a tested failure protocol, what triggers a model rollback, who authorizes it, and how operations continue in the interim. Right now, only 7% of manufacturers have one. Build it before. The cost of building it after is measured in production hours, not planning hours.
Organizations with fully integrated AI are nearly four times more likely to report revenue growth than those still piloting, 58% versus 15%. The gap is in the data architecture that connects AI decisions to financial outcomes. Without that link, ROI exists but stays invisible, which, from a board’s perspective, is the same as not existing.
Many plants still run on approval structures built before AI was a factor, where a quality deviation flagged in seconds waits hours for sign-off. The COO’s role is to define which decisions AI can own outright, which need human review, and what conditions determine each path.
Stop reporting on AI activity. Report on AI impact. Define three to five measurable outcomes, throughput recovered, defect rate change, unplanned downtime reduction, and connect every live deployment to at least one. If a deployment can’t show movement on a business metric within 60 days, treat it as a research project and resource it accordingly.
The playbook above is not theoretical. Tezo partnered with a leading global manufacturer of conveyor belts, operating across 29 countries, to move their AI from manual, fragmented processes to a production-grade, multi-agent automation system. The results were not incremental. Quote turnaround time accelerated by 99.9%. Order processing, inquiry management, and quote generation, functions that previously required multi-day manual handling across international teams. These now run through AI with seamless automation at global scale.
This is what the Grant Thornton gap looks like when it closes: AI embedded into critical business workflows with outcomes that are visible, measurable, and defensible in any board room.
The shift required three structural changes: end-to-end process:
Ready to identify which of your pilots is production-ready? Book a 45-minute AI Diagnostic with Tezo’s manufacturing practice.
Manufacturing contributes approximately $2.9 trillion annually to U.S. GDP, employs 13 million people, and faces mounting pressure from tariff uncertainty, labor shortages, and supply chain fragmentation. AI isn’t a productivity luxury here, it’s an operational necessity.
Grant Thornton’s data makes the cost of inaction concrete. A sector running at scale without governance, accountability, or a clear pilot-to-production path isn’t being cautious, it’s accumulating unpriced risk.
The COOs who build that path, and explain it clearly to their boards, are the ones still in the room when results arrive. Contact us to identify which of your pilots can hit production in 90 days, and which need immediate restructuring.
Don’t miss out on the latest tips, tools, and tactics at the forefront of HR and Employee
We love what we do so much and we're always looking for the next big challenge, the next problem to be solved, the next idea that simply needs the breath of life to become a reality. What's your challenge?
