The Rise of AI in Manufacturing

As global supply chains grow more complex and uncertain, agentic artificial intelligence may offer new ways for manufacturers to better manage risk and capture value in their organizations, according to a Deloitte report.

AI agents, which can reason, plan and act with autonomy, can perform routine activities that reduce cycle times and free up humans for other strategic tasks, according to Deloitte’s Resilient By Design: The Agentic Supply Chain. They also can conduct always-on monitoring and decision-making, as well as scale activities beyond the limits of human workers.

Victor Reyes, managing director of Deloitte’s Human Capital unit, told Manufacturing Dive that companies should reimagine their workflows around where they can most leverage AI agents, as well as identify the “unique parts of the jobs and the processes where only a human can make the call.”

Agentic AI adoption is already accelerating. A recent study from IBM found that more than half of the supply chain executives that it surveyed are already deploying AI agents to automate workflows.

Separately, Gartner estimated that by 2030, about 50% of cross-functional supply chain management solutions will use intelligent agents for autonomous decision-making. It also predicted in a more recent report that 40% of enterprise applications will have integrated task-specific agents by 2026, compared to less than 5% in 2025.

What’s resonating with companies today is the enhanced level of autonomy that the technology can bring, Reyes said, allowing for swift, proactive responses to dynamic commodity pricing, delivery timeframes and other factors.

“Now you have technologies that actually can be constantly monitoring and sensing those things and infusing that into your operations as they make decisions,” he said.

From a risk management standpoint, agentic AI can improve how manufacturers navigate geopolitical events and weather disruptions, as well as the frequent changes in tariffs and customs regulations.

The agents are able to evaluate an entire bill of materials, assess things at a component level and process and begin to take action, like initiating requests to alternative suppliers, Reyes said.

“It’s taking two or three or four steps in response to that sensing that’s quite different than just a dashboard that a human might have had before with some of this information and then have to figure out what to do with it,” he said.

One of the biggest challenges companies are facing is how to implement the technology with their current workforce in mind. Citing a separate Deloitte report, Reyes said about 93% of companies’ AI investments are going into the technology itself, while only 7% are going toward their people.

“You can throw all this technology in the mix, but if you’re not really taking the time to co-create a new set of roles with your workforce, you’re creating risk,” he said.

Deloitte identified four elements for executives to keep in mind that ensure sustained value creation with agentic AI, including data architecture, technology stack, workforce preparation and risk mitigation. Reyes said the first two are foundational or “evergreen” for companies, whereas the latter are important for conversations around where the technology can best fit within the organizations.

“What would you do if you could go out and hire 1,000 more people to run this organization? Well, that's the kind of impact that you can have with agentic AI,” Reyes said.

As manufacturing companies across the U.S. increasingly adopt artificial intelligence, the workflow is shifting from text-generating models to physical systems that use robots, sensors and processing capabilities. Despite the potential this presents, there’s a huge gap between experimenting with AI tools and wide-scale implementation of fully automated systems. 

One way companies are starting to bridge this divide is through testing sites. Notable examples launched in the last few years include Tata Consultancy Services’ partnership with Google Cloud on seven Gemini Experience Centers, Microsoft’s AI Co-Innovation Lab in Wisconsin and Deloitte’s Smart Factory in Kansas. These sites allow manufacturing companies to test physical AI and explore use cases before making any major shifts. 

The models differ based on what the manufacturing company is looking for and what the testing center can provide. Most of these collaborations tend to be project-based, focused on exploring new technological possibilities and solving specific operational challenges. 

The centers offer an immersive learning environment by connecting digital, physical and experimental elements on a fully operational production line, said Rohini Prasad, a principal at Deloitte’s Smart Manufacturing Business, via email.

“Manufacturers get to experience firsthand how cutting-edge solutions can be integrated and deployed with their own operations,” said Prasad. “Organizations can explore use cases, test integrations and develop road maps for scaling AI and digital transformation in a way that goes beyond pilots and into real operational impact.”

This step is necessary to ensure the process works before making a major investment.

These are probabilistic systems likely to make mistakes. So they need to be tested and safeguards must be put in place to understand and mitigate the risks, said John Harrington, chief product officer at HighByte, an industrial dataops software company working with manufacturers to contextualize plant floor data.

“There is a lot of variability across manufacturing plants, including the products they produce, the production lines, and the individual cells within those lines,” he added. “AI offers a way to manage this variability more effectively than rules-based systems, which require extensive custom programming to account for each difference.”

While AI tools seem better equipped to deal with variability, they also complicate training and testing the automation models. 

“Production conditions change constantly, from machine performance to material differences to operator inputs, and models need to perform consistently despite that variability,” said Tim Beatty, president of Bullen Ultrasonics, via email.

Bullen is an Ohio-based precision machining provider working on specialty material components for industries such as aerospace, medical, automotive, defense and semiconductors. The company is working with an external partner to test data-driven AI models to increase machine precision and performance.

The testing lab environments are designed for experimentation and early-stage learning, which makes it possible to move faster, reduce upfront investment and access specialized expertise, he said. But this can be a double-edged sword.

“Those environments can reduce risk, but they don’t always reflect the complexity of day-to-day operations and may not work well when deployed in a real workflow,” said Assaf Melochna, president and co-founder of Aquant, an agentic AI platform for asset-centric manufacturers.

“Manufacturers are looking at things like safety, accuracy and reliability, and testing centers can be useful for evaluating those in a controlled environment,” he said. “But the real question is how well the system performs in actual workflows, where conditions are less predictable and the cost of getting it wrong can be high.”

Some manufacturers, like Bullen Ultrasonics, have found a solution to this by testing AI in their own machining environment using real data and customized goals.

“This ensures what’s being tested is directly relevant to how our business actually runs,” Beatty said. 

Another concern is how data gets collected, managed, shared and integrated with the physical AI systems to execute tasks. 

“AI is only as good as the data behind it, and in most factories, data isn’t perfectly clean or consistent across every machine, line, or department,” Beatty said. “If the data isn’t reliable, the outputs won’t be either.”

That’s why he warns manufacturers against overhauling the entire process with automation too fast. 

“A better approach is to focus on a single use case, such as reducing scrap on a single line or improving cycle time on a specific machine, and test AI there first.”

Due to this, Beatty sees the quickest wins coming from areas such as machining that already have a large amount of structured data available. This makes it easier to solve specific problems, like identifying patterns that lead to defects.

Aquant is already seeing such results with Makino, a global CNC manufacturer that is using the company’s platform to help technicians troubleshoot equipment issues faster. That’s why Melochna sees predictive maintenance as a strong use case. 

Such predictive maintenance systems can be used with agentic AI models that companies like Aquant are building to create automated bots that track operation patterns, identify issues and fix them with minimal human intervention. 

Demand for testing such use cases is growing as manufacturers move beyond experimentation and toward operationalizing smart manufacturing capabilities.

“The critical questions they are now trying to answer are less about whether an AI solution exists and more about how it fits into their organization,” said Prasad.

Prasad said industries with complex production environments and high automation potential — such as aerospace, advanced industrial equipment, energy and life sciences — have been early adopters seeing value from real-world testing centers, but organizations across all industries are exploring. 

No matter which area manufacturing companies choose to work with, she added, “the fastest evolution will happen in capabilities that deliver near-term ROI and can be scaled with existing infrastructure.”

Sponsored content

By L2L

The growing gap between knowing and doing

Manufacturing has a data problem.

Plants generate troves of data across machines, systems, people and processes. Despite an abundance of data, many manufacturers struggle to use it effectively.

Nonetheless, manufacturing data is valuable—especially when it translates to better decision-making. Deloitte reports that most manufacturing leaders plan to invest at least 20% of their 2026 budgets in technologies enabling advanced data collection and analysis.

But even with these investments, manufacturers haven’t seen substantial productivity gains in over a decade. According to the U.S Bureau of Labor Statistics, manufacturing productivity has been steadily declining since 2011.

In short, digital transformation was a double-edged sword. It made machines and data collection faster, but it also added technological complexity that hindered operational efficiency. This issue is compounded by the ongoing skilled labor shortage, which 80% of manufacturers already experience to a painful degree, according to the 2025 L2L Skills Report. 

This leaves manufacturers in a “data paradox.” They have more data across systems, assets and people than ever, but it hasn’t translated into clearer insight or faster action.

To get to the root of this sector-wide problem, we asked over 600 U.S. manufacturing leaders about their biggest challenges with industrial data, including:

• How data is collected in their facilities
• Data quantity and quality issues
• Barriers to real-time visibility and execution
• How technologies, including AI, help or hurt
• The cost of manufacturing data issues
• Strides toward better operational insight

Drowning in data: The current state of data collection

Today, manufacturing leaders have the tools and systems to collect more plant data than ever.

But more data hasn’t necessarily translated into better insights or faster action.

From Industrial Internet of Things (IIoT) sensors to complex dashboards, modern industrial technologies promise real-time visibility and clear guidance for process improvements. However, many manufacturers still lack a single source of truth for operational performance.

Nearly half (47%) of the manufacturing leaders surveyed report an even split between automated and manual operational data collection. Data is captured through a variety of methods, most commonly:

• Quality Management Systems QMS: 62%
• Machine sensors or equipment data: 55%
• Manufacturing Execution Systems MES) or production systems: 52%
• Manual input by frontlines: 50%
• Maintenance management systems: 48%
• Spreadsheets: 48%

Despite the availability of automated data collection solutions, half of manufacturers at least partially rely on manual data capture (e.g., spreadsheets, whiteboards, paper), especially on the front lines. Moreover, 11% state that they capture operational data almost entirely by hand.

Even with heavy investment in IIoT and sensors, there’s still a disconnect between what systems capture and what’s actually happening on the shop floor. When 55% of manufacturers use automated data collection, yet 50% rely on manual frontline logs, the industry isn’t running a digital operation. It’s a hybrid system where machine data sits idle while managers wait for a paper trail to give them enough context for decision-making, leading to operational friction and lost profits.

In the next section, we’ll uncover why industrial organizations have trouble turning data into action.

Starving for insights: Barriers to trustworthy, actionable data

Operational efficiency, top-quality products and optimal workforce performance all require a connected, real-time view of shop floor operations. Yet, this level of visibility still isn’t the norm.

In fact, only three in 10 manufacturing leaders confirm that their operational data reflects the shop floor in real time.

Why is this number so low?

One reason is that tool and system usage are inconsistent. While 89% of respondents report having a standard set of tools across all teams and levels, 63% say that usage is inconsistent, even if it’s required.

That’s less than half of the respondents who consistently use the data solutions provided for them.

However, standardizing tools isn’t the same as standardizing processes. When tools are introduced as point solutions rather than integral parts of a system, they become extra tasks instead of efficiency boosters. These issues often arise when leadership is misaligned with the way frontline teams work on a daily basis. Implementing new software alone won’t fix broken processes.

When data lives in different places, teams have trouble accessing and acting on it in a timely manner. This scenario leads to more manual data input and assessment, resulting in operational delays. It’s also where the gap between data and action becomes a permanent part of company culture.

Organizations often attempt to unify this data with legacy systems (e.g., ERP, MES, or EAM). While these solutions are necessary for modern plants, they act as systems of record—only capturing what went wrong—instead of systems of action that provide real-time guidance. Manufacturers that can’t turn static data into a live workflow struggle to hit production targets, let alone make significant productivity improvements.

The cost of manufacturing data challenges

The impact of inconsistent and disconnected data on the plant floor is palpable.

Nearly two-thirds of respondents say frontline supervisors spend at least an hour of their shift pulling, cleaning and reconciling data. Some even report supervisors spending over four hours per shift simply trying to make sense of operational data.

When team leads spend half their time reviewing or cleaning data, they aren’t managing the shop floor. Their preoccupation with data handling opens the floor to safety risks and other problems, which can cause massive operational delays.

When turning data into insight takes this long, problems persist for hours or even days. Only 9% of respondents say they can usually find the root cause of a shop floor issue in real time. This means that the other 91% spend their time fixing and documenting preventable and costly problems. Moreover, 65% say that root causes are inconsistently documented.

Another consequence of inconsistent information capture is the loss of valuable know-how, particularly tribal knowledge. Eighty-eight percent of leaders say important operational information disappears when experienced employees change roles or leave.

This makes it much harder to train new employees, maintain standards and avoid repeat issues. The departure of knowledgeable workers results in inexperienced new hires starting with little to no role-specific knowledge and a much steeper learning curve.

Therefore, it’s no surprise that unused and inaccessible insights have resulted in increased financial, material and time-related costs in the past 12 months for respondents:

• Fourty-five percent report more scrap and rework
• Fourty-three percent report frustration among teams
• Fourty-three percent report repeated issues
• Fourty-three percent report missed production targets
• Fourty-two percent report longer resolution times

Aside from inconsistent documentation, respondents cite the following issues as the biggest barriers to turning data into action in their plants:

• Inconsistent processes - cited by: 32%
• Manual data collection - cited by: 32%
• Data quality issues - cited by: 31%

The prevalence of data challenges in modern facilities brings up an important question: If data is primarily handled by software, how much of the problem is technological? 

Siloed data isn’t just a headache for IT teams. It’s a productivity killer that has long-term consequences for industrial organizations. Yet, teams are still expected to produce while tackling new manufacturing challenges, all with old and inefficient data gathering methods.

In the next section, we’ll delve into the effectiveness of advanced technologies, like artificial intelligence AI and how they affect the actionability of data.

Does modern industrial technology help or hurt?

Today’s plants rely on multiple software systems to keep operations running. But when these systems exist in silos, teams have a difficult time piecing together a complete, real-time view of plant performance. Many manufacturers add new tools and solutions to their tech stacks to alleviate this issue. However, the result is often tech fatigue: more layers for frontlines to manage.

Our respondents’ feedback underscores this problem. Three-quarters report relying on multiple systems to perform their daily work, creating an environment that’s disjointed and overly complicated:

• Sixty-five percent say their operational tools are redundant
• Sixty-four percent have a fragmented tech stack
• Fifty-three percent say their tech stack is too complex
• Fifty-eight percent report that some of their tools create more friction than clarity

Advanced data collection and analytics solutions are designed to streamline reporting and decision-making. However, they tend to produce the opposite effect if they don’t accurately reflect plant floor activities or provide a holistic picture of operations.

And it’s impossible to talk about industrial technology without mentioning AI, a highly promising innovation when applied to manufacturing environments. But to deliver actionable insights that drive tangible outcomes, AI tools must be combined with unified shop floor workflows.

Eighty-eight percent of site-level leaders say their teams use AI tools, with the most-reported use cases being: data analysis (65%), automated dashboards (68%) and predictive insights (41%).

Most manufacturers (87%) believe AI has the potential to significantly improve shop floor productivity, even if turning AI’s outputs into action remains a challenge. Here’s a quick breakdown of our respondents’ AI-related observations:

Top-cited benefits and challenges of industrial AI

Benefits

• Faster insights: 52%
• Increased productivity: 51%
• Improved visibility: 48%
• Less manual data analysis: 47%
• Reduced downtime: 47%

Challenges

• More support is needed to turn AI findings into action: 87%
• Integration challenges limit AI’s impact: 79%
• Difficulty aligning AI tools with existing workflows: 70%
• Lack of frontline trust in AI’s outputs: 68%

It’s clear that the usefulness of AI applications on the shop floor depends on the quality, integration and organization of a plant’s data. But without a foundation of unified, accurate and up-to-date datasets, AI applications become extra tools that add complexity instead of clarity to manufacturing processes.

Industrial software should make data more actionable, helping plants unlock operational insights that drive efficiency and productivity. Without a live and complete view of operations, manufacturers will continue to deal with bottlenecks, excessive downtime, reporting delays and other barriers to operational efficiency and productivity.

Better, more connected data = better outcomes

So, what can better operational insights do for manufacturers?

According to our respondents, the biggest benefits of better operational insights include:

• Improve quality: 52%
• Improve cost savings: 51%
• Boost workforce productivity: 50%
• Reduce downtime: 48%
• Improve training and skill development: 44%

Knowing exactly what’s occurring and when is critical not only for creating better products with greater speed and efficiency but also for enabling long-term process improvements.

The path forward: Recommendations for manufacturing leaders

The "manufacturing paradox" of being data-rich but insight-poor is not a failure of effort, but a failure of architecture. To break the cycle of declining productivity and digital fatigue, leaders must shift their focus from simply collecting data to orchestrating action.

Today, most plants are still focused on building and maintaining a system of record instead of one that enables proactive action. Data is only the fuel for change on the plant floor. Instead of simply collecting data, manufacturing leaders need to focus on better ways to drive productivity and efficiency.

To make the shift from a system of record to a system of action, leaders can start with these four strategic pillars:

1. Centralize data for a single source of truth. Operational data is only an asset when it is centralized and reflects the shop floor in real time. Currently, only 21% of leaders find it very easy to access the data they need. By moving away from fragmented spreadsheets and manual logs, which 50% of plants still rely on, teams can eliminate the "information chase" and focus on execution.
2. Transition from systems of record to systems of action. Traditional software like ERP, MES and CMMS often acts as a digital filing cabinet, merely recording what went wrong yesterday. To drive true OEE gains, plants need an action-based system that unifies disparate data and instantly notifies the right person to solve the right problem at the right time.
3. Systematize tacit knowledge digitally. With most leaders reporting that critical operational knowledge disappears when experienced employees leave, the resulting information gaps pose a significant business risk. Leaders should implement digital platforms that capture and standardize best practices, making this tacit knowledge permanent part of the plant’s “digital DNA.”
4. Reduce complexity to empower AI. The potential for AI to improve productivity is often throttled by fragmented technology stacks. By reducing technological complexity and unifying workflows into a single interface, teams provide AI with the clean, high-velocity data it needs to move beyond simple reporting and start providing prescriptive guidance for the shop floor.

Turn data chaos into clear, actionable insights

L2L’s Connected Manufacturing Operations Platform is a system of action that replaces fragmented silos with operational clarity. By unifying maintenance, production and skills into one interface, we eliminate the "information chase," helping the world’s leading manufacturers move from post-mortems to real-time execution with measurable ROI in as little as 16 weeks.

While AI and automation seem to be the biggest trends in the industry, Intrinsic Chief Technology Officer Brian Gerkey recently shared a striking statistic: 80% of U.S. manufacturing facilities have zero automation.

In spite of discussion about the potential benefits of this technology, it’s still far from widespread in the United States, let alone at the level of fully automated factories as seen in countries like China and Japan.

“There is no doubt interest is high across the board, but execution is where things get difficult,” said Jeff Burnstein, president of the Association for Advancing Automation. The group’s research also shows that while a strong majority of manufacturers believe AI will be critical to their future, only a small percentage say it’s widely deployed today.

Deloitte’s 2025 Smart Manufacturing and Operations Survey showed similar results. An estimated 92% of manufacturers surveyed said they believed smart manufacturing will be the main driver for competitiveness over the next three years.

Yet only about 29% of manufacturers reported already using AI or machine learning at the facility or network level, and only 24% had deployed generative AI. Looking ahead over the next two years, 41% of respondents said they planned to prioritize factory automation investments.

The bottlenecks holding back AI adoption

Manufacturers are still building the foundational capabilities required to scale AI and automation, said Tim Gaus, a principal and smart manufacturing leader at Deloitte.

Despite nearly three in four companies planning to deploy agentic AI within two years, only one in five reported having an equipped model, according to the company’s State of AI in the Enterprise report.

“Many organizations are still working with fragmented legacy systems and data that is not structured for AI use,” said Jasmeet Singh, executive vice president and global head of manufacturing at Infosys.

According to Singh, it often comes down to digital maturity. Manufacturers that have already modernized their core systems, invested in cloud and built strong data foundations are moving faster on AI. Singh said such companies are better positioned to scale beyond pilots because their data is ready to support advanced use cases.

Singh added that a key shift where many manufacturers struggle is the transition from pilot projects to measurable business outcomes, because they want a clear return on investment before committing significant funds.

“In many cases, earlier proof-of-concept efforts did not translate into enterprise-wide impact, which has slowed broader adoption,” he said.

The disappointment often comes from how AI was implemented rather than the technology itself, something Singh called using “AI for AI’s sake.”

But even these “failed” projects can be valuable, according to Burnstein, because they give “a much clearer understanding of where the real constraints are, which leads to more successful implementations the next time.”

Automation considerations

Most automation today is built for highly standardized, repeatable processes, said Stefan Nusser, chief product and commercial officer at Intrinsic, Google's robotics software company. But manufacturing often involves variability, customization and evolving processes, which can make automation expensive to test and implement.

Even when they’re established, Nusser said the processes may run for only a short period despite the significant bandwidth required to set them up. This may be because automation is just used for a specific task but not scaled across the whole factory. Companies may also realize the technology is not feasible or profitable for their current business model, or their manufacturing needs change over time and require adaptation to newer processes.

Major bottlenecks are commonly seen with mid-sized manufacturers that have the resources to experiment with AI and automation solutions but tend to wait until the value is clear before making large-scale investments. On the other hand, larger organizations have more resources to pursue aggressive growth, but their scale and legacy systems can slow things down, Singh said. 

That’s why Nusser advocates for incremental adoption. He cited the example of autonomous mobile robots that are flexible and less risky to implement, compared to a legacy, conveyor-based solution that requires overhauling the entire structure and workflow. The latter may seem nearly impossible, according to experts, considering that shutting down the system for even a few days could cause losses and supply chain disruptions in the millions. 

History repeats itself

This is not the first time the manufacturing industry has experienced a complicated technology transition. It also happened with the adoption of simpler robots.

“The technology faced resistance because of initial high costs and hesitancy around change management, but eventually, sectors like automotive manufacturing saw huge benefits after prioritizing the foundations needed to make it effective,” Gaus said. 

AI and advanced automation are following a similar trajectory. Early adopters are demonstrating clear productivity gains, he said, with some companies reporting 10% to 20% improvements in production output and employee productivity.

“As these results become more visible, we can expect adoption to accelerate across the industry.”

According to a recent PwC report, manufacturers could triple their current automation levels by 2030. Some use cases showing the strongest potential for adoption of automation in the next few years include data capture and analysis, demand forecasting and quality control.

A recent report from the International Federation of Robotics shows that trend has already been building in recent years. An estimated 542,000 industrial robots were installed across the world in 2024, more than double the number 10 years prior. U.S. installations reached a peak of 44,000 units in 2023.

Their findings echo Gaus’ predictions: “One thing we consistently see is that once manufacturers observe successful use cases — especially from their peers — adoption tends to accelerate very quickly.”

Agentic AI and robotics are among the most transformative opportunities for efficiency gains in modern manufacturing. But as automation scales, it raises a question that may not be top of mind for manufacturers: What are the energy costs?

According to a recent PwC survey, 81% of executives plan to increase artificial intelligence investments over the next three years, and 93% "believe America’s industrial advantage will be built on intelligent systems.” Moreover, 46% of energy and industrial professionals report that their firms are investing in renewable energy generation and storage, with roughly one-third expecting to achieve energy independence by 2030.

Meanwhile, agentic AI and robotics are becoming safer, smarter and more affordable. This comes as the manufacturing workforce faces an accelerating talent shortage driven by an aging labor pool and loss of institutional knowledge.

For companies in the early stages of AI and robotics deployment, energy costs can appear almost negligible. 

Agility Robotics, which has deployed its humanoid robot Digit for clients including Amazon, GXO and Toyota Motor Manufacturing Canada, estimates the power cost runs about $1 per shift. 

"The real complexity in physical automation isn't power — it's integrating into real-world workflows, safety systems, and operations on the floor," said Jonathan Hurst, cofounder and chief robot officer, via email.

That framing resonates with manufacturers newer to automation. But there may be another reality to consider for those that are looking to scale up their agentic AI or robotics quickly: increased energy use.

“As you scale up and these things become a bigger part of your production, it might be significantly more,” said Jackie Bakalarski, a principal focused on sustainability and ESG at Avetta.

Bakalarski recommends thinking through if low energy figures will still apply after factoring in your specific considerations.

"Once you normalize that cost across the infrastructure, the building, running your own smaller power plants, etc. That's more than probably a dollar a shift over time.” 

Energy is no longer linear with output

Perhaps the most important insight for manufacturers is that the traditional relationship between energy consumption and production output has fundamentally shifted.

There was a time when energy would increase alongside productivity. That may no longer be the case, according to Bakalarski . 

"Energy demand used to be very indexed to output," she said. "Now those things have shifted to a degree. So your output can go up a lot, while your energy demand only goes up maybe marginally.” 

That decoupling is part of why some feel agentic AI is a compelling option for manufacturers.

"My personal goal is to bring at least 30% productivity of an employee,” said Anoop Mohan, chief product and technology officer at Augury. "And if we can do that, that's for me the role of what we call AI agents."  

Augury's Mohan describes a world in which AI agents work continuously alongside reliability engineers, maintenance teams and operations staff. These agents can interpret sensor data, flag potential failures before they occur and recommend actions in real time. The results are less unplanned downtime and smoother production runs, which in turn carry an energy efficiency dividend: machines running at optimal performance draw less power than those operating in a degraded state awaiting repair.

Planning power as a parallel track

Where companies stumble, Bakalarski notes, is in failing to plan energy infrastructure alongside production changes.

"If you're planning to change your production, and you're adding robotics or going to be doing something that creates a bigger energy pool, you need to be planning your energy.” 

Automotive manufacturers have historically led in this discipline by modeling production environments digitally — including energy load sequencing — before committing changes to the factory floor. As the AI industry evolves, manufacturing companies increasingly compete for energy resources with data centers, according to a recent AlphaStruxure survey. Risks resulting from resource constraints include energy shortages and even delays in opening new sites when energy is unavailable to support the facility.

Another factor is the promise of alternative sources of energy. 

A recent Deloitte Energy and Industrials report, which focused on a broad range of sectors such as cleantech manufacturing, data centers and AI, said “renewables are in a race with other clean generation options to fill the resource gap.” A possible advantage of renewables is that they may help boost overall energy availability and eventually lower cost. 

With all of these factors in mind, experts say the manufacturers best positioned to capture the productivity promise of agentic AI and robotics are those treating energy not as an afterthought but as a core input in the investment decision from day one. 

Food manufacturers have opportunities to use advanced technologies, such as artificial intelligence, to improve the quality and safety within their operations, Drew Gaputis, a principal at Deloitte, said during the virtual Food Manufacturing Summit on March 25.

Gaputis was joined by Shay Luo, co-founder at a stealth AI startup, to discuss how food manufacturers are investing in ways to make their operations more efficient using these advanced technologies. 

“We’re seeing tools such as computer vision, automation [and] anomaly detection,” Gaputis said. “These are things that can be implemented in the manufacturing process to really help companies be more proactive in terms of potential safety pitfalls.” 

Gaputis added that companies’ ability to implement AI and automation to manage food safety proactively will help contain issues as soon as they arise and before they become widespread.

“I think that's something that we see companies investing in that will continue to occur,” he said.

Improving food safety and compliance

Advanced technologies can also help food manufacturers comply with U.S. traceability regulations, Gaputis said. A traceability rule under the U.S. Food and Drug Administration's Food Safety Modernization Act requires the agency to label certain foods that necessitate additional recordkeeping to protect public health, according to the FDA’s website.

The rule aims to speed identification and rapidly remove potentially contaminated food from the market to reduce foodborne illnesses and deaths. 

The traceability rule’s compliance date was initially set to begin Jan. 20. However, the FDA proposed extending the compliance date to July 20, 2028. Additionally, an appropriations law passed in November 2025 ordered the FDA not to enforce the rule before the new deadline.

Food manufacturers that take advantage of AI and other digital technologies will gain a “cleaner view” of the end-to-end supply chain, Gaputis said.

“That degree of visibility, I think, helps not only provide accountability, but [also to] identify issues and address those issues again in a very proactive manner,” he said.

As food manufacturers approach the traceability rule’s 2028 implementation date, Gaputis said he expects they will continue using the advanced technologies as well as “better managing” their supply chain “from an organizational standpoint.”

Luo added that the technology would help with predictive safety management, as AI is “extremely good” at considering many factors, “probably more than what human brains can process.”

“This can be leveraged for safety management and predict what’s coming, what would be a risk before it materializes.” 

Technology challenges foodmakers face

As food manufacturers implement AI and other technologies into their operations, they may overlook a few things and encounter challenges, speakers said.

Gaputis said some of the issues he’s seen involve data and “articulating” return on investment from an executive level.

“You can realize quick savings and productivity on an individual or maybe a functional level quite easily,” Gaputis said. “But thinking about large-scale investments in the ROI required for full-scale process transformation and organizational changes may be a little bit longer than your typical technology implementation.” 

His clients have also needed to design new or reimagined processes and operate models “holistically” to get the best use out of AI.

“We've seen a lot of a lot of large-scale CPG companies start to think about, ‘How do I define major processes by which our organizations are run? Create clarity around those processes and even owners from an end-to-end perspective across functions?’” Gaputis said. “And then using that as a mechanism to really reimagine the work, so that companies can gain full maximum leverage from the application of technology.”

He added that manufacturers need to consider shifting from applying technology to reinventing the organization now, a move that “leading companies” are quickly making.

Companies also overlook their employees’ concerns about working with AI. Luo said change management and gaining floor-level buy-in are “critical to the success of any of the digital transformations.”

“I think let's just be realistic about that and be very upfront with our workers and be radically transparent about the roadmap ahead and how you plan to use it,” Luo said.

She added that it’s more about helping people understand how AI can enable their work, rather than feeling threatened. 

“The current models are not as good as human judgment, so still, governance and human oversight are extremely important,” Luo said. “How to design a system that humans are in the loop for necessary override is more important than just letting the system run automatically.”

Luo said she doesn’t think using technology is the solution to replace human workers. Rather, it’s about how companies can use technology to address issues such as safety and tasks humans do not want to perform.

While AI is “very good” at performing repetitive and routine tasks, she added that she doesn’t think the technology is at a level where it can replace human workers.

“Those are the more valuable use cases of AI than just simply replacing people,” Luo said.

In August, a cyberattack forced automobile manufacturing giant Jaguar Land Rover to shut down production for a month. The company reported around $260 million in cybersecurity costs in addition to $650 million in losses owing to the production disruption, U.S. tariffs on imported cars and the phasing out of older models. 

The unexpected halt of the manufacturer’s highly automated production lines in the UK, which typically work on about 1,000 vehicles a day, also disrupted a broader global supply chain. Unions and officials estimated thousands of workers could be laid off, and smaller suppliers could go bankrupt due to the sudden loss of business. 

Incidents like these are growing as manufacturers increasingly digitize their processes. Experts are raising concerns that cybersecurity measures have become an afterthought as more companies rapidly adopt artificial intelligence and cloud systems for efficiency, profitability and reduced reliance on human labor.

According to the 2025 Deloitte Smart Manufacturing Survey, 57% of the 600 executives surveyed at large U.S. manufacturing companies reported using cloud systems. An estimated 29% said they are using AI and machine learning at the facility or network level. 

In 2024, North America was the largest market for cloud-based manufacturing infrastructure, accounting for almost 50% of the global share, according to a report by Market Research Future.

While this rapid growth seems promising from a production perspective, experts warn that the pace of tech adoption has far surpassed the cybersecurity measures required to keep these systems running safely. 

Are AI and cloud systems making manufacturers vulnerable?

The IBM X-Force Threat Intelligence Report 2025 states that manufacturing has been the most-attacked industry by cybercriminals for four years straight. 

“The biggest cybersecurity risk in manufacturing right now comes from the amount of connectivity being introduced into environments that were never designed for it,” said Nick Nolen, vice president of cybersecurity  strategy and operations at Redpoint Cyber, which works with manufacturers. “The real challenge is the mismatch where manufacturing is modernizing quickly, but the underlying systems and processes weren’t originally built with cybersecurity in mind. That is what creates the openings attackers look for.”

Manufacturers used to fly under the radar when it came to cyberattacks, mostly because their systems weren’t online, said Todd Moore, VP of encryption at Thales. Until recently, he added, systems were designed for performance with no cybersecurity measures as digitizing was not even a consideration.

Now, as companies integrate advanced technology like AI and cloud systems with old infrastructure that isn’t equipped for digital operations, Moore said this has become an issue.

“Security is often bolted on to these systems rather than built with secure-by-design principles, making manufacturers vulnerable to everything from ransomware and malware to phishing and even denial-of-service attacks,” he said.

Adoption of AI and cloud systems amplified these risks by significantly expanding the attack surface area, according to multiple experts. Nolen said manufacturers have a much broader attack surface than most people realize.

“Modern manufacturing relies heavily on third-party integrators, connected machines, vendor-supplied software, and data exchange between business units,” he said. Each of these touchpoints introduces another opportunity for compromise. 

“Once an attacker gains entry into even a small corner of the environment, the interconnected nature of manufacturing systems can allow them to move quickly toward more sensitive areas,” Nolen said. “Attackers are going after whatever piece of the digital environment is easiest to get into, and as more manufacturing data moves into the cloud for AI and automation, those systems become a bigger and more attractive target.”

The biggest risk here, according to Kevin Albano, global head of X-Force Threat Intelligence at IBM, is the possibility of unauthorized access to the sensitive data manufacturers upload to AI and cloud systems.

“To mitigate this, manufacturers need to treat AI datasets as high-value assets,” he said. “That means classifying and protecting sensitive data across cloud, on-premises and hybrid environments, encrypting all personally identifiable information at rest and in transit and implementing strong key management.”

Cybersecurity measures to prevent attacks

Heavily monitoring AI usage can be difficult to manage as many manufacturers don’t fully understand the components their vendors are using behind the scenes, which creates security blind spots, said Ferhat Dikbiyik, chief research and intelligence officer at Black Kite. 

The company’s 2025 Manufacturing Report echoed IBM’s findings, showing manufacturing being the biggest cyberattack target for four years in a row. 

“Manufacturers need visibility into which vendors have access to production data, where AI is being used and how systems could connect back into [operational technology],” Dikbiyik said. This becomes complicated when the people involved in production and supply chain use AI and cloud tools informally, without security teams knowing what’s being uploaded or which models vendors are relying on, he said.

“Once you upload any sort of design or process information into an AI tool, you need to be confident about where that data is going, who can see it, and how it might be used,” Nolen said. “Do you know what your vendor is doing with that data? Do you know where it is stored? How long is it retained? Is it being used to train their own models? Many companies still don’t have clear guidelines on what is safe to upload and what should stay local.”

That’s why encryption is essential, according to Moore.

“Manufacturing organizations can begin with thoroughly classifying data and assessing risk to determine where vulnerabilities lie within hybrid or cloud environments,” he said.

The process, known as data classification, includes labeling information based on its sensitivity so that teams know what needs the highest level of protection.

Another concern is cloud systems being hacked because they “centralize sensitive design files, recipes, production parameters and supplier information, which means a single compromised cloud account can ripple across multiple plants,” Dikbiyik said. That’s why, he said, “companies need proper segmentation between IT, cloud, and operational systems so that a breach in a digital tool can’t cascade into production.”

Significant upfront costs required to set up and maintain security systems are deterring many manufacturers from implementing such measures. However, according to Nolen, "more spending doesn’t equal more security." 

Nolen helps manufacturers calculate the probable financial loss in the event of an attack, which helps them make informed decisions about how much they might need to spend, either for prevention or in the aftermath of an attack.

As part of this process manufacturers have to weigh the cost of investing in cybersecurity programs that must evolve with technological advancement versus the risk of an attack exacerbated by increased digitization. 

“I am never in favor of slowing innovation for the sake of cybersecurity,” Dikbiyik said, “but I always believe that we can make sure the digital factory of the future is built on a security model that matches its complexity.”