The AI-driven transformation of manufacturing

The way in which manufacturers approach digital transformation varies widely based on their operational model, their position within supply chains, size and specialism. However, one thing unites the sector: nearly every manufacturing company surveyed for Forvis Mazars’ C-suite barometer study either has a dedicated strategy for digital transformation (74%) or is developing one (23%). 

AI increasingly plays a key role in these strategies. Two-thirds (63%) of senior leaders in the sector say they expect AI to have a major impact on their organisation, according to our C-suite barometer. Half (47%) believe that AI is now a key determinant of digital transformation success. For now, however, levels of AI project delivery remain relatively low. Only one-third (36%) of manufacturing executives say their organisation has undertaken major restructuring as a result of implementing AI during the past two years. 

Undoubtedly, the challenge of funding transformation in an asset-heavy industry characterised by long depreciation cycles plays a role here. Manufacturing companies tend to rely on a series of tightly interconnected processes, which can make a gradual approach to AI transformation problematic. In addition, concerns about ROI, skills shortages, legacy systems and AI maturity persist. 

However, there are signs that awareness of AI’s potential results in action. Relatively few large manufacturers accustomed to low levels of investment in digital transformation regard their existing approach as sustainable in the era of AI. Most intend to spend more on AI transformation in the near future. In addition, recent surveys from the U.S.-based Manufacturing Leadership Council detect widespread use of vision systems, machine learning and generative AI alongside increasing interest in agentic AI. Despite the specific challenges confronting manufacturers, AI may yet emerge as a significant catalyst for widespread transformation across the sector. 

New possibilities on the factory floor 

AI is creating a wide range of possibilities on the factory floor and beyond, including improved real-time forecasting, factory-floor automation, schedule optimisation and monitoring. For process or continuous manufacturers, predictive maintenance and time-series analytics that yield improvements in uptime and yield are proving attractive. In discrete manufacturing, attention is often focused on robotics, autonomous process systems and closed-loop production systems, which use sensor and machine data to fine tune the manufacturing process. 

“AI involves thinking about a company as the sum total of its workforce, skills, processes and capabilities. Increasingly, we will map the key elements involved and transfer responsibility for them to AI agents with the potential to orchestrate these tasks in a sequential fashion. The future of AI involves multi-modal, multi-agent orchestration across the entire business, supervised by human experts.”  Christian Segurado, Senior Manager, Forvis Mazars US

Identifying the right use case scenarios for investment remains the key challenge for senior managers. Confronted by the competing claims and rival solutions, a clear understanding of how AI can generate the best possible returns for their organisation is essential. This requires a consistent focus on major pain points that AI has a demonstrated potential to address. As many AI pioneers can attest, ‘shiny’ cutting-edge projects with vaguely specified outcomes can exert a powerful influence on management thinking. Likewise, it is worth interrogating arguments that your organisation should pursue transformation options deployed by rivals. AI is not a ‘one size fits all’ solution: what works for Company A will not necessarily work for Company B. 

A resolute focus on ROI metrics will help steer use case selection in the right direction. During the initial stages of corporate experimentation with AI, many stories emerged of apparently successful pilot projects that struggled to prove their value on the verge of full-scale deployment. Efforts to retrofit ROI justifications frequently fail. Typically, the original error lies upstream in the form of a poorly chosen use case scenario. 

Drawing upon the experience of neutral outsiders offers clear benefits. Gathering market intelligence and engaging in peer-based discussion should be the norm for senior managers discussing the potential of AI transformation. The ideal combination involves an in-depth understanding of both the technology and the practical steps required to deploy and manage it. Organisations armed with both have a much-improved chance of navigating their way to successful AI transformation.  

Resilience, compliance and digital supply chains 

Businesses leaders surveyed by our C-suite barometer identify four high-impact threats that they find hard to control: macroeconomics, competition, geopolitical instability and energy costs. In the manufacturing sector, a fifth threat is included on this list: supply chains.

Manufacturers have reacted to the need for more resilient supply chains, supported by dual sourcing and inventory buffers. Integrating new supply chain technologies has also become a key investment priority, cited by 44% of senior executives surveyed for our C-suite barometer. The resulting transformation of supply chain management involves far more than adapting to major episodes of disruption. It creates the potential for continuous dynamic risk management in three key areas: 

• Building continuously updated models that generate risk calculations based upon macroeconomic trends, trade flows, market conditions, logistics options and production capacity. 
• Using the same data sources to refine operations by aligning production capacity with intelligence on the availability of materials, components and buyer behaviour. 
• Optimising the balance between resilience and efficiency in supply chains by real-time monitoring and adjustment of inventory levels depending on levels of sourcing risk and supplier performance. 

Notably, regulation is encouraging the digital transformation of supply chains. The EU’s Carbon Border Adjustment Mechanism (CBAM) and pending enforcement of the Corporate Sustainability Due Diligence Directive (CSDDD) are obliging large companies to quantify and address risks that extend beyond their own facilities. In the U.S., food manufacturers have been adapting the traceability demands of the Food Safety Modernisation Act (FSMA) and pharmaceutical manufacturers are navigating the final implementation phases of the Drug Supply Chain Security Act. 

Under these circumstances, management by spreadsheet is being replaced by cloud-based systems of record and technologies that enable real-time reporting and transparency. Agentic AI is also making its presence felt. In food manufacturing, for example, time-sensitive compliance actions such as product recalls, which often involve complex data analysis, have become an ideal deployment opportunity. Manufacturers approaching supply chain transformation should prioritise the following possibilities for digital transformation: 

• Digital supply chain management requires a single source of truth for all production data produced by ERP systems and quality management systems (QMS). Increasingly, manufacturers are replacing paper-based records with tamper-proof digital records detailing suppliers, batch data and distribution routes. 
• The use of AI and advanced analytics require manufacturers to standardise the quality and governance of data used for compliance. Accurate metadata, data protection compliance and time-stamped records of due diligence actions will increasingly be required. 
• Manufacturers should investigate the potential of 2D codes, RFID and Internet of Things (IoT) to automate compliance checks and reduce dependence on line-of-sight confirmation in warehouses and production facilities. More broadly, EPCIS 2.0 and similar emerging technologies are designed to enable frictionless sharing of ‘what, when, where and why’ data between supply chain partners, creating the potential for detailed supply chain auditing. 

Charting the way forward 

The fact that relatively few manufacturers have so far moved ahead with early-stage AI projects may turn out to be beneficial in the long term. Many now enjoy the opportunity to learn from the mistakes of pioneers. The key lesson involves the need to prepare the conditions required for successful AI deployment. 

For example, most manufacturers operate back-office systems that aggregate and manage data from across the business, including enterprise resource planning (ERP), manufacturing execution systems (MES) and specialised software for operations management, product lifecycle management and quality management. The need to modernise these systems arises for multiple reasons, including the need to integrate financial reporting or a software vendor’s withdrawal of support for a legacy product. Frequently, core system upgrades can become the centrepiece of more broad-based digital transformation, as in the case of a Forvis Mazars client who recently completed a two-year project involving new systems for field service, production, service and a full ERP and CRM migration. Whatever the scale of the project, it is becoming vital for manufacturers to take the future requirements of AI into account when upgrading these systems and the infrastructure that supports them. 

Focusing on core systems should be accompanied by investment in enabling technologies including data management, IT/OT convergence and cybersecurity. Yet according to our C-suite barometer, only 33% of manufacturers describe investment in these key enabling technologies as a top priority. Many more manufacturers now need to move ahead with preparing for AI deployment in these areas. Appropriate funding should be allocated to the following: 

• Because AI places significant strain on ageing IT infrastructure, transformation teams need to objectively assess the maturity of existing systems, investing where necessary in cloud-native or SaaS platforms that will accelerate the deployment of AI solutions in the future. 
• Focus on creating the conditions for all core systems to generate well-governed data. AI systems need clean, standardised and well-annotated datasets to function at scale. Whether your organisation’s data is centralised or decentralised, it needs to meet the key rules of being findable, accessible, interoperable and reusable (FAIR). 
• Building the internal skills and teamwork required in areas like AI, IT/OT convergence and cybersecurity is vital. Upskilling, new recruitment and cross-functional teamwork will all make essential contributions to success. Shifting the focus of recruitment and training towards the skills and knowledge required to integrate AI with manufacturing is becoming a necessity. 

 Manufacturing companies that have already pressed ahead with large-scale AI innovation will understand the importance of these key initiatives. Many have faced initial struggles to scale pilot projects and made the necessary investments required to build, rather than buy, solutions. The positive news for these manufacturers is twofold. The experience of deploying emerging technology will have generated the knowledge and skills required to capitalise on AI’s future potential. Just as importantly, we are now entering a new phase in AI’s evolution, characterised by a far better understanding among software vendors of what it takes to deploy AI solutions that reliably address manufacturers’ needs under real-world conditions on the factory floor. 

CES insight lens: manufacturing innovation 

Each year, the Consumer Electronics Show (CES) attracts 150,000 attendees to giant exhibition halls in Las Vegas to examine thousands of new products and services. CES is a vast shop window for the technology and consumer electronics industries. What did this year’s show tell us about emerging trends in manufacturing? 

CES has evolved beyond being a trade show focused on consumer tech. For several years, industrial technology has featured on the show floor. In 2026, for the first time, manufacturing was the subject of a substantial dedicated conference programme at the event. 

As companies adopt the dynamic, real-time approach to data required by AI, digital twin technology is maturing rapidly. At CES, Siemens displayed its Digital Twin Composer, which links data from CAD-based technical designs with real-time operational data to create advanced simulation environments. Athina Kanioura, global chief strategy and transformation officer at PepsiCo, outlined how deployment at one of the company’s oldest legacy sites enabled a production redesign that increased efficiency by 20% within three months. Kanioura noted the existence of internal analysis suggesting that similar redesigns implemented across all of the company’s manufacturing plants could reduce annualised capex by 10% to 15%. 

Robotics have always been part of the agenda at CES, chiefly because of the technology’s scene-stealing demonstration potential. This year, the emphasis shifted from future possibilities to execution. Industrial applications were prominent, including robots designed to navigate hazardous environments, work on large irregular structures and operate in tight, restricted spaces. 

Robotic hands with finely tuned motor skills were a particular focus, including DexH13 from PaXini, based upon a four-finger multi-joint structure able to manipulate a 5kg payload with advanced tactile sensing. Long confined to apparently impressive demonstrations, intelligent robot hands like these now offer sufficient precision for deployment in real-world use cases. In common with other B2B vertical sectors, these use case scenarios are increasingly supported by an ecosystem of software, data solutions and credible options for maintenance and support. 

Innovations like these point to a future in which the upstream phase of manufacturing operations is designed, simulated and secured by digital twin technology, while downstream production is addressed by robotic solutions offering an increasingly wide span of applications. The benefits of manufacturing innovation at scale are well understood: less downtime, accelerated production and more refined capital allocation. However, the remaining limitations are just as clear: cybersecurity, data governance, platform dependency and, above all, a shortage of skills in engineering, simulation and advanced robotics. It is hard to avoid the conclusion that these technologies sit on the verge of industrialisation. Further progress will depend upon the ability to integrate them, secure them and maintain them.