AMS Talks 2025-2026: The Human & The Machine in Carmaking

What an eventful year it has been… The AMS team has spent 2025 visiting factories from Mexico to Portugal, to Germany and beyond - and what we’ve seen is a definite shift away from the grandiose promises of centralised data lakes and monolithic IT platforms. In their place, a more pragmatic production approach has emerged, which prioritises speed, measurability and human agency over architectural ambition.

This was a central theme in our final AMS livestream of the year, Smart Factory Trends and Lessons for 2026: People, Process and Technology, where my colleague Nick Holt, (Editor) and I, Ilkhan Ozsevim (Deputy Editor) alongside industry expert Riddhi Padariya from the ISA, and supported by clips from Dr Hans Bernd Kettler from Bosch Mobility in America, and Murugan Boominathan from Magna IT, discussed what we have learned from a year spent on the front lines of manufacturing transformation.

What we observed across these factory floors was manufacturers grappling with what we at AMS call "open heart surgery": transforming century-old brownfield sites into flexible electric vehicle (EV) production hubs without halting output. The challenge represents the single most formidable operational task in automotive manufacturing today.

And make no mistake: The classical dog-eared production playbook no longer applies. As vehicle producers confront intensifying cost pressures and compressed development cycles, we have watched them discover that digital success cannot be measured by budget size but by tangible value delivered at the assembly station. The actual transformation demands more than new robots, it rests on ways of thinking about data, collaboration and most importantly, the role of the workforce.

On our visit to Volkswagen's Autoeuropa plant in Palmela in October, for example, we found that rather than pursuing a multi-million-euro software package, engineers at the site solved large problems with small solutions. At Palmela, they have developed the Welding Online Monitoring System for a modest €17,000 ($17,850), and using a custom algorithm to analyse real-time operational data from welding guns, the system now predicts abnormal behaviour and potential failure before quality problems materialise on the production line.

Inside Palmela: How VW transformed a 30-year-old factory for net zero

From geothermal networks to electrified paintshops and closed-loop metal recycling, Volkswagen's Palmela plant demonstrates how legacy factories can achieve significant decarbonisation—while confronting hard truths about polymer contamination and circularity limits.

The significance of this goes beyond cost efficiency. It signals what we might call the democratisation of IT-OT innovation, where the most valuable digital solutions need not originate from executive mandates but can emerge from shop-floor ingenuity. The monolithic data lake concept, once the dominant paradigm, has now yielded to agile, low-friction, employee-empowered approaches.

We found that the lesson is to empower the engineers closest to the process to solve their own problems. That is no less than one of the core shifts we have been seeing.

And predictive maintenance embodies this manufacturing wind-change. What has for a very long time been either reactive or schedule-based solutions has matured into sophisticated, AI-supported systems which essentially eliminate unnecessary downtime whilst avoiding wasteful interventions. With EV production requiring substantial capital investment in battery assembly, platforms and infrastructure, reducing expensive disruptions has become paramount. Throughout our site visits, we have observed manufacturers gaining a better handle on what they want from data, how to manage it, and what they are looking for, rather than simply collecting vast quantities of it. The correct data and implementations are far more important than big data. 

The greenfield advantage

Yet the brownfield-greenfield divide still remains stark. When we visited BMW's Debrecen facility in Hungary, conceived as part of its iFactory approach, we encountered a plant designed virtually with extensive digital planning. The IT-OT integration was holistic from inception, with data management baked into every process.

At this BMW smart factory, a central control unit functions as mission control, monitoring incoming components, coordinating logistics and tracking flow through the plant. And to connect it all up on a human level, every staff member carries identical work phones, ensuring communications reach everyone uniformly. Whilst the homogeneity might appear dystopian to some, it supports culture-building around technology adoption. You get a strong sense that data is controlled and actionable.

Inside BMW’s smart logistics strategy at Debrecen

BMW is redefining automotive logistics with direct delivery systems, real-time tracking, digital twins, and autonomous transport at its Debrecen plant. The OEM’s tech-driven approach integrates production, quality, and sustainability – setting a new standard for next-gen vehicle manufacturing.

Ford presents an illuminating counterpoint to this. Despite significant greenfield investments in BlueOval operations, the manufacturer has emphasised brownfield upgrades at facilities like Dearborn, taking an asset-light approach. When we spoke to Paul Stevens at our event, Automotive Manufacturing North America (AMNA) this year, he told us that if you visited the plant before and after implementation, you might not even notice physical differences, yet the changes prove remarkably effective. The end goal remains consistent, but the path is liable to diverge, substantially.

Sustainability ever-woven into production operations

Another major shift is in sustainability, where there is a clear indication that formerly peripheral concerns have migrated to the operational core. Throughout our travels, we have witnessed environmental management evolve from compliance-driven and retrospective approaches to active, real-time resource optimisation.

At BMW's San Luis Potosí plant in Mexico which we visited last October, we were introduced to the site’s digital energy control room, which aggregates real-time data across energy, water and CO₂ systems, using predictive analytics and machine learning to anticipate inefficiencies. The approach fuses digital tools with the facility's green and lean objectives, transforming sustainability from an afterthought into an operational imperative across the plant.

And since the energy control room automatically captures sustainability footprints across departments, it automatically digitally dismantles structural silos by consolidating disparate resource use - and disparate departments - into unified visibility. Site-wide, teams can observe how their actions map to outcomes, fostering accountability and engagement. We have also watched JLR tackle similar challenges in some of its legacy plants. It can, indeed, be a capex challenge, but it also saves money in the long-run. Again we see that often, relatively small projects can make a meaningful difference, and they are typically data-driven.

BMW’s San Luis Potosí plant in Mexico is driving global sustainability with solar power, water reuse, and AI-driven carbon reduction

BMW’s San Luis Potosí facility in Mexico exemplifies sustainability in action. From solar energy to water recycling and AI-driven efficiency, the plant demonstrates how local, effective green innovation can reshape automotive production on a global scale. Following a site visit, AMS delves into the facility’s world-leading sustainability intiatives. Ilkhan Ozsevim reports.

Passing the torch: Breaking tribal knowledge in carmaking

2025 has shown us that cultural transformation proves as challenging as technological implementation. For decades, automotive manufacturers struggled with tribal knowledge - with deep expertise being trapped in functional silos with limited cross-pollination. That paradigm is fracturing under competitive pressure.

"Our industry is in a big phase of change, and it's about speed," Dr Hans Bernd Kettler explained at AMNA. "We have a lot of new technologies and we need to be fast to compete, including with the pace we see in China. We see breakthroughs like electrification and automated driving. This requires resources and investment, and it doesn't allow us not to be lean or not to be fast, or to make expensive mistakes."

Kettler emphasised that whilst technology dominates discussions, success hinges on people. "We need colleagues and teams to think end-to-end, not just functionally. We do this through cross-functional rotations, building rotations into early training, and forming multifunctional teams with an end-to-end logic."

At AMNA, we also heard a striking example from Trent Randalls, Engineering Manager at BorgWarner. He assembled a cross-functional team of approximately nine people spanning engineering, production, product development, safety and maintenance. The group physically dismantled and reassembled existing components alongside prototypes at a technical centre. This hands-on approach enabled everyone to comprehend differences that theoretical approaches would have missed, who as a result were able to anticipate challenges and propose improvements. It built trust and team cohesion, making time-poor individuals willing to invest effort.

Randalls has been consistent about this in his conversations with AMS: digital tools matter, but collaboration and cross-functional teams have been crucial to making progress.

Smart Factories need smarter people

As automotive manufacturers accelerate digital transformation, the industry faces a fundamental reckoning about workforce development. Four industry leaders reveal how they are reshaping roles, building skills and bridging the gap between automation ambitions and human capability.

Bake it in: The clean slate advantage

Another central topic we encountered time and again, is that legacy manufacturers face a structural disadvantage against new entrants, particularly from China. As AMS has written about extensively this year, the edge enjoyed by companies like Nio and BYD stems not from superior robotics but from collapsed development processes and rigorous design for manufacturing principles.

Whilst legacy vehicle producers operate on five-to-seven-year major model cycles, Nio has reportedly delivered new EV designs to market in approximately 120 weeks, roughly half the time. And we see two factors that enable this velocity: vertical integration and digital-first agility.

Companies such as BYD manufacture core components internally, eliminating friction between design and production. This of course, compels concurrent working between teams and streamlines complex, expensive systems. And with new market entrants, digital-first agility treats vehicles as unified digital products, embedding manufacturing software engineers into design from the earliest stages. We have seen industry estimates suggesting that a substantial share of eventual product cost becomes locked in early during development.

When AMS spoke to NIO about their collaboration with Comau on an e-drive module production line, the approach was striking. The company brought the integrator into the earliest design stage, sharing CAD drawings and effectively building the line parallel to product development, before final financial contracts were completed. Trust and tight integration meant that whenever Nio altered a design, Comau could adapt immediately. Physical parts and physical lines matched in significantly compressed timeframes.

The lesson that automakers should glean from this is that the most agile operations treat collaboration not as a request, but as a structural mandate. Competing in the EV era requires legacy players to adopt a clean slate mindset and make simultaneous engineering an enforceable pillar of production. That’s why new entrants are on the rise while legacies lag behind.

Watch The Full Livestream Now On Demand

Smart Factory Trends and Lessons for 2026: People, Process and Technology

In this AMS livestream, our editorial team and industry experts reflect on the year’s most significant developments and discuss what they mean for the future.

Scaling tech without losing the human touch

And we saw that Magna's approach to digitalisation reveals the complexity of scaling innovation whilst preserving local needs. During our livestream, a clip of Murugan Boominathan showed him sharing insights that resonated with what we have observed across multiple facilities.

"When we introduce new things, especially when operations don't have experience, the first question is, 'What is going to change tomorrow?'" Boominathan explained. "The best way is to introduce change slowly and at a constant pace. Start with a small proof of concept with the team on the floor. Engage them early so they get experience with the tool, compare it with what they do today, and adapt to the new one. That lays down a path to adopt change steadily."

Boominathan stressed the importance of process-first thinking. "We also work with leadership to make sure we can adapt changes from one place to another and scale effectively. On cross-functional teams, Magna has several product lines with different processes. We start with the process. Identify what drives output, group processes, and build solutions for the process rather than deploying a generic solution."

There was a lot to be learnt at AMNA. There, automotive leaders emphasised capturing tacit knowledge from experienced workers nearing retirement and transferring it to younger teams. Some suggested upskilling programmes, design thinking, catalyst teams and encoding expert knowledge into models and large language models. The approach proves people-centric because workers shape the models rather than vice versa.

We have also seen Toyota Motor Manufacturing UK addressing skills shortages at its Burnaston plant, where more than 300 technicians approach retirement. The manufacturer has implemented a hybrid apprenticeship programme with Rockwell Automation and a local college, incorporating two years of classroom training with real hardware and simulation software alongside behavioural competencies. The academy upgraded to current Rockwell equipment to reflect modern systems.

We interviewed Stephen Heirene, Industry Consultant at Rockwell, who emphasised the critical importance of training reflecting real-world applications. The Burnaston upgrade, he said, ensured hands-on experience resembles what learners encounter in factories. Indeed, this people-first approach runs through everything we have observed this year.

Rockwell's Stephen Heirene on bringing future mobility to factory floor training

Rockwell Automation's Stephen Heirene explains how the company is helping automotive manufacturers bridge the gap between legacy systems and digital transformation through training academies that prepare engineers for the realities of modern production.

Battery production's persistent bottlenecks

And we see that despite progress, EV battery pack production still remains technically demanding. During our livestream, Riddhi Padariya offered valuable insights based on her experience with battery packs and Tesla Megapacks. Having worked mostly on battery packs at Tesla's Austin and Nevada operations, she identified several persistent challenges that keep raising their heads.

Watch the full video to find out more

However, in a clear motif across the industry, Padariya pointed to the learning that manufacturability needs to be considered from inception. A Tesla Model 3 or Model Y pack can contain 2,500 to 3,000 cylindrical cells, she said. So engineers must decide whether to build smaller modules that assemble into packs or construct large pack sections with different configurations. Failing to think through these choices, forces later station retrofits because equipment cannot meet yield requirements.

AI as co-pilot, not oracle

We also see Artificial Intelligence's role in manufacturing continuing to evolve, though cautiously. Throughout our reporting this year, we have watched vehicle producers construct internal versions of ChatGPT, closed-loop systems using proprietary data to avoid external contamination. BMW has pursued this approach, supporting democratisation by making data more accessible internally.

Yet AI models prove only as effective as the data fed into them. Legacy datasets remain large and messy, though manufacturers have gained better command of the challenge. 

BMW's purchasing and supplier network division is piloting an AI purchasing agent, a customised multi-agent system that aggregates, analyses and prepares relevant data to make complex procurement processes more efficient. It functions as an intelligent co-pilot for buyers and strategists, with ramifications for production.

It’s clear to us that next year in 2026, AI in production involves not single-task robots but interconnected autonomous agents handling enterprise complexity. These are systems that can shift AI from an operator tool to a strategic asset, optimising value chains rather than merely assembly lines. On factory floors, no doubt, AI will converge more closely with automation. Predictive maintenance will also absorb more of this capability, sifting through legacy datasets to extract what matters and applying the right information to solve appropriate challenges. That growth will accelerate in 2026.

We heard at an event earlier this year, including from Fraunhofer, that a very large share - 80%! - of AI projects fail. Whatever the precise scope of that statistic, it serves as a sobering reminder that the human layer remains indispensable.

As we advance into 2026, the message from our year of reporting proves consistent: no matter how automated and digitalised manufacturing becomes, people must remain central. Technology serves as an enabler, but culture, collaboration and empowerment determine whether transformation succeeds or stalls. The smartest automotive factories will be those that balance silicon with heart.