98% of Manufacturers Are Exploring AI. Only 20% Feel Ready. Here's the Gap in Between.
If you run a manufacturing operation, you've heard the pitch a hundred times: AI is transforming the shop floor, the supply chain, the forecast. What you don't hear as often is how few manufacturers actually feel prepared for it. That gap, not the hype, is the more useful number to plan around.
The short answer: 98% of manufacturers are exploring or actively considering AI-driven automation. Only 20% feel fully prepared to use it at scale. That's not a small gap. It's the defining fact of where manufacturing actually stands with AI right now, and it explains a lot about why so many AI initiatives stall out between the pilot and the plant floor.
What "exploring but not ready" actually looks like
The 98%-vs-20% number gets thrown around as a headline stat, but the detail underneath it is more useful than the topline:
Seven in ten manufacturers have automated half or less of their core operations. Most of the plant is still running on manual process, not because leadership doesn't want automation, but because the last mile of implementation is harder than the pitch decks suggest.
Only 40% have automated exception handling: the process of catching and correcting the orders, shipments, or production runs that don't go as planned. Manufacturers themselves cite this as one of the most disruptive manual processes they run, and it's still mostly done by hand.
78% have automated less than half of their critical data transfers: the flow of information between ERP, MES, and planning systems that AI actually needs to be useful. You can't run good AI forecasting on data that's still moving between systems via export and re-import.
Put together, this isn't a story about manufacturers being behind or resistant. It's a story about the AI layer having nothing solid to stand on. Systems aren't talking to each other cleanly enough for automation to extend past the pilot stage.
Why the gap exists
The honest reason most manufacturers stall here isn't lack of interest (the 98% figure rules that out). It's that the obvious path to "AI-ready" looks like a full technology overhaul: rip out the ERP, rebuild the data architecture, retrain the team, and hope the new system delivers what the old one couldn't. That's an expensive, risky, multi-year bet, and for a mid-market manufacturer already running lean, it's a bet that's easy to keep deferring.
The overlooked alternative is narrower and less disruptive. Instead of replacing the systems of record, connect an intelligence layer on top of what's already running: pulling data from the existing ERP, cleaning up the handoffs between systems, and applying forecasting and automation to the processes that are still manual. That doesn't require the operation to stop and rebuild. It requires the data that's already being generated to actually get used.
The market is moving whether or not any single plant is ready
This isn't a niche concern. The manufacturing operations management software category (production scheduling, workflow automation, operational intelligence) was valued at roughly $22 billion in 2025 and is projected to more than double to over $52 billion by 2030. Manufacturing digital transformation spending broadly is projected to approach $1 trillion by 2031. Whatever pace an individual plant moves at, the surrounding market is building toward the assumption that manufacturers will close this gap, not stay in it.
Trade volatility is accelerating the timeline further. Tariff swings and shifting trade policy through 2025 and into 2026 have pushed manufacturers who import raw materials or components to reconsider supplier diversification and lead-time planning, both of which depend on having reliable, real-time data rather than a monthly spreadsheet reconciliation. The manufacturers under the most pressure to close the AI-readiness gap are often the same ones dealing with the most supply chain disruption.
Who this actually affects
This isn't limited to large enterprise manufacturers with dedicated data science teams. It's most acute in the mid-market: manufacturers running established ERPs (Dynamics GP, older Epicor, Sage, or heavily customized legacy systems) with multi-SKU production, bill-of-materials-based manufacturing, and import-dependent supply chains. That's the segment where the data almost always exists somewhere in the business, but doesn't yet move cleanly enough between systems for AI to work with it.
Frequently asked questions
What percentage of manufacturers are using AI? 98% report exploring or considering AI-driven automation in some form. Actual scale-readiness is far lower: only about 20% feel fully prepared to deploy it at scale, and roughly 70% have automated half or less of their core operations.
Why do so many manufacturers struggle to scale AI past a pilot? Most commonly, it's a data problem rather than a technology-adoption problem. Exception handling and cross-system data transfers (the connective tissue AI needs to operate on real-time information) remain largely manual at most manufacturers, which limits how far automation can extend even when the appetite for it is there.
Do you need to replace your ERP to close this gap? No. The more common and lower-risk path is adding an intelligence and automation layer on top of the ERP that's already in place, rather than a full system replacement. That approach uses the data already being generated instead of requiring a multi-year rebuild before any AI value shows up.
Is this gap specific to any one industry within manufacturing? The data spans manufacturing broadly, but it shows up most clearly in mid-market operations (multi-SKU, BOM-based, import-dependent businesses) where legacy systems and manual handoffs are most common and the cost of a full ERP replacement is hardest to justify.
Where this leaves you
The 98%-vs-20% gap isn't a reason to rush into a system overhaul, and it isn't a reason to wait it out either. It's a reasonably precise description of where the actual bottleneck sits: not interest, not budget in the abstract, but the plumbing between the systems that already hold the data. Closing that gap tends to start smaller than people expect: not a new ERP, but a layer that finally puts the data your systems are already generating to work.
Sources: Manufacturing AI and Automation Outlook 2026 (Redwood Software, via PR Newswire); Manufacturing Operations Management Software Market Report 2026 (The Business Research Company); Manufacturing Modernization: Four Trends to Watch in 2026 (Forvis Mazars).