Loading...
Loading...
Indiana has the second-highest manufacturing output per capita in the United States, and the state's industrial base runs on three sectors that could hardly be more different from one another: integrated steel production at Cleveland-Cliffs Burns Harbor and US Steel Gary Works along Lake Michigan, diesel and alternative-power engine manufacturing at Cummins in Columbus, and FDA-regulated pharmaceutical and medical device manufacturing anchored by Eli Lilly's $9 billion Indiana manufacturing expansion. Each sector has a distinct AI adoption profile, and the conditions in Indiana's industrial corridor — aging blast furnace infrastructure in Gary, precision-toleranced engine machining in Columbus, cleanroom pharmaceutical production in Indianapolis — create specific predictive maintenance and process control requirements that generic AI platforms rarely handle without significant customization. Cleveland-Cliffs' Burns Harbor facility is one of the last integrated steel mills in the United States operating a hot-strip rolling mill, blast furnaces, and a BOF steelmaking shop under the same roof — a capital-intensive, energy-intensive operation where AI-based furnace control and roll-force optimization translates directly to yield improvement on commodity margins that leave little room for inefficiency. Cummins, which has spent $1 billion globally on electrification and digital initiatives, is deploying connected-engine telematics and AI-based engine health monitoring from its Columbus headquarters that sets a technology standard its Indiana suppliers are expected to approach. Eli Lilly's 21 CFR Part 820 Quality System Regulation environment adds a pharmaceutical manufacturing dimension that few industrial AI vendors are equipped to navigate without specialized support.
Updated June 2026
The Gary-Hammond-Portage corridor in Lake County and Porter County is one of the heaviest industrial concentrations in North America — Cleveland-Cliffs Burns Harbor, US Steel Gary Works, ArcelorMittal (now Cleveland-Cliffs) Indiana Harbor, and dozens of downstream metals processors operate within a 15-mile stretch along Lake Michigan. Burns Harbor runs three blast furnaces, a basic oxygen furnace steelmaking shop, and a hot-strip rolling mill that collectively consume more electrical power than some small cities. AI applications that generate real returns here include blast furnace burden distribution optimization (using tuyere-level sensor data and burden probe measurements to feed ML models that reduce coke rate), hot-strip mill roll-force prediction (reducing cobbles and roll changes through computer-vision billet tracking and force-feedback models), and ladle lifecycle management (predicting refractory lining wear before costly breakouts). US Steel's Gary Works has been a pilot site for Carnegie Mellon University's integrated steel AI research consortium — one of the few ongoing academic-industrial AI collaborations in North American steelmaking — and the talent and methodology that program has generated is accessible to Indiana steel operators through licensing and consulting relationships. The practical challenge across Lake County steel facilities is that much of the installed instrumentation dates to the 1980s and 1990s, requiring data-layer integration work before any ML model can run. AI vendors who arrive without steel-process domain knowledge consistently underestimate this by 30–50% in their scopes.
Cummins Inc. operates its largest engine manufacturing complex in Columbus, Indiana — producing diesel, natural gas, and hydrogen-combustion powertrains across multiple plants in the Columbus metro area. Cummins' Connected Diagnostics platform, which uses AI to analyze engine telemetry from hundreds of thousands of on-highway and off-highway engines in the field, has created a feedback loop that informs manufacturing process quality AI in Columbus. Engine failures in the field that trace to specific machining variance at Columbus plants are now detectable within weeks rather than warranty-return cycles measured in months. The downstream effect on Indiana's engine supplier base is measurable: Cummins has formally required key Tier 1 suppliers to demonstrate AI-based process monitoring capability as a condition of long-term sourcing agreements, citing variability-reduction targets that manual SPC programs can no longer reliably meet. Suppliers in the Columbus, Indianapolis, and Fort Wayne manufacturing corridor — machined castings, fuel systems, turbochargers — are responding with IoT sensor deployments and ML-based statistical process control that would have been considered over-engineering five years ago. The Cummins Technical Center in Columbus also runs a supplier development program that provides validated process-monitoring architecture guidance to Indiana manufacturers who are trying to meet Cummins' AI-readiness requirements without the in-house capability to evaluate competing AI platforms.
Eli Lilly's $9 billion Indiana manufacturing investment — spanning facilities in Indianapolis, Lebanon, and Branchburg — is one of the largest pharmaceutical capital programs in U.S. history, and it's creating an AI implementation environment in Indiana that is fundamentally different from steel or automotive. FDA's 21 CFR Part 820 Quality System Regulation (transitioning to 21 CFR Part 4 / ISO 13485 alignment under the 2024 Quality Management System Regulation) requires that software used in manufacturing quality decisions — including AI tools used for process analytical technology (PAT), in-line spectroscopy, and equipment qualification — be validated as part of a computer software assurance (CSA) program. This is not a documentation exercise: it means that AI vendors must participate in installation qualification, operational qualification, and performance qualification testing under Lilly's quality system before any AI output can be used to release or reject pharmaceutical product. The validation burden is significant and expensive — a PAT-connected AI deployment at a Lilly facility that would take 90 days at a non-regulated manufacturer can take 12–18 months to fully qualify. But the upside is equally significant: a fully validated AI that predicts batch yield from in-process sensor data in a biologics manufacturing suite can save tens of millions of dollars per year at Lilly's production volumes. Indiana's broader life sciences manufacturing cluster — which includes Cook Pharmica (now Catalent) in Bloomington, Roche Diagnostics in Indianapolis, and Zimmer Biomet in Warsaw — faces the same 21 CFR 820 / QSR environment, creating a statewide demand for AI vendors with pharmaceutical manufacturing validation credentials.
Connecting AI systems to existing business infrastructure and workflows
Workflow automation using AI, including Make.com-style automation and RPA
Predictive models, data analysis, and ML pipeline development
Image recognition, object detection, video analysis, and visual inspection systems
Based on deployments at Cleveland-Cliffs Burns Harbor and comparable facilities, the three highest-ROI applications are: blast furnace thermal efficiency optimization (AI-based control of burden distribution and blast parameters, typically delivering 3–7% coke-rate reduction worth $8–20M annually at large facilities), hot-strip mill cobble and roll-change prediction (computer vision and force-model ML reducing yield loss, worth $3–8M annually in scrap reduction), and predictive maintenance on BOF converter vessel lining (ultrasonic thickness modeling extending campaign life by 5–15% and avoiding catastrophic breakouts that cost $500K+ in lost production). The data infrastructure work to get these models running typically adds 40–60% to the stated vendor cost.
Cummins communicates AI readiness expectations through its Supplier Quality Manual and Annual Supplier Development Review process. Tier 1 suppliers are scored on process capability indices (Cpk targets tied to critical characteristics) and on the monitoring-system evidence they can provide that those indices are being maintained in real time — not just in periodic audits. Suppliers who can demonstrate continuous IoT monitoring with ML-based control chart alerting score significantly better than those relying on manual gauge sampling. In Columbus and Fort Wayne, several machined-casting and precision-turning suppliers have told us the Cummins scoring system was the direct trigger for their AI investment decisions — the contract renewal risk was more compelling than any ROI calculation.
For a process analytical technology (PAT) or in-process control AI tool at a Lilly-regulated facility, the validation cycle — covering IQ, OQ, and PQ under Lilly's internal quality system — typically runs 9–18 months from vendor selection to approved use in production. The FDA's 2022 Computer Software Assurance guidance for 21 CFR Part 11 and Part 820 has streamlined some documentation requirements, but Lilly's internal quality standards still require extensive performance qualification data across the range of process conditions the AI will encounter. Vendors who arrive without pharmaceutical validation experience regularly underestimate this timeline by 50–100%, which creates schedule and budget overruns that have soured several Indiana pharma operators on AI implementations that were technically sound.
A mid-size Indiana machined-parts manufacturer deploying AI-based statistical process control and equipment health monitoring across a 50–150 machine shop floor should expect $150,000–$350,000 in total first-year cost, including sensor hardware, edge compute, software licensing, and integration services. The Cummins Supplier Development Program can offset some of this through co-development agreements if the manufacturer is willing to share process data with Cummins' engineering team. IMEC (Indiana's MEP affiliate, similar to Illinois IMEC) provides subsidized AI readiness assessments that help manufacturers scope these deployments and negotiate vendor contracts before committing.
No — and the gap is most acute at the intersection of manufacturing process knowledge and ML engineering. Purdue University's industrial engineering and mechanical engineering programs produce strong process-domain graduates, and Indiana University's Kelley School of Business produces analytics talent, but the overlap in manufacturing-specific AI skills is thin relative to the volume of projects being launched by Cummins, Lilly, Cleveland-Cliffs, and their supplier ecosystems simultaneously. In practice, most Indiana manufacturers are using hybrid teams — internal process engineers who own the domain knowledge working alongside external AI specialists who own the ML infrastructure. Remote-capable vendors with physical presence in Indianapolis, Columbus, or Fort Wayne are in the highest demand.
Reach Indiana businesses searching for industrial AI expertise.
Get Listed