ABI Research projects industrial AI spending to increase from US$4 billion in 2026 to more than US$35 billion by 2035, with manufacturing software and robotics driving the greatest investment. How is the market evolving and where should technology providers focus their GTM strategies?
The International Manufacturing Technology Show (IMTS) only takes place every 2 years, and a great deal can change between events. The conversation around industrial Artificial Intelligence (AI), a defining theme this year, has shifted considerably since 2024.
Scouring the show floor at McCormick Place in Chicago for the second time this year, following Automate 2026, I expect isolated Generative AI (Gen AI) demonstrations to take a back seat to Agentic AI, Physical AI, and the broader role AI can play in industrial automation.
Manufacturers are asking harder questions now. Where can AI deliver measurable value? How much oversight does it require? And can it work within the factories they already operate?
Before making the trip, I wanted to address five questions that will shape the industrial AI conversation over the next few years.
Table of Contents
- How has the industrial AI conversation changed since IMTS 2024?
- Why is industrial AI becoming an ecosystem play?
- How will industrial machinery vendors adapt to the AI shift?
- Where can industrial AI deliver more autonomy without adding operational risk?
- Why is brownfield integration the next test for industrial AI?
How has the industrial AI conversation changed since IMTS 2024?
The industry is speaking a different language than it was 2 years ago. At IMTS 2024, much of the attention was on Gen AI and the ways manufacturers could use natural language interfaces to retrieve information or create content. Those capabilities still matter, but industrial organizations are now looking further ahead.
IMTS 2026 will place much more emphasis on Agentic and Physical AI. We are starting to talk about systems that can interpret information, determine what should happen next, and help initiate an appropriate response with little to no human intervention.
Consider a manufacturing dashboard. Traditionally, someone must know which screen to open, which data to review, and how to interpret them. AI can make that experience more dynamic. An operator could ask why a production line slowed during the previous shift and receive a focused explanation based on quality records, machine data, and operating conditions.
Quality analysis and root cause identification are two especially promising AI applications. When something goes wrong, the first challenge is understanding what happened. The second is determining how to resolve it quickly. Industrial AI excels at contextualizing events, enabling manufacturers to move through both stages faster. It also delivers relevant information to workers when they need it instead of requiring them to search across disparate systems.
Why is industrial AI becoming an ecosystem play?
ABI Research tracking data show a 218% increase in industrial AI activity between 2024 and 2025. By activity, we refer to partnerships, acquisitions, investments, and internal initiatives. The increase suggests that industrial AI is becoming an ecosystem play rather than something vendors can develop in isolation.
NVIDIA spearheads the industrial AI space, accounting for 56 total activities within the 2024 to 2025 time frame. Its influence extends from photorealistic engineering and simulation to autonomous robotics and edge applications. The result is automation that can tackle less predictable tasks, not just highly structured and repeatable processes that AI has typically been relegated to.
Just a few NVIDIA activities include:
- A partnership with PTC to bridge Product Lifecycle Management (PLM) actions with transformer models
- A US$1 billion investment in VAST Data to build the AI operating system for Graphics Processing Units (GPUs)
- A partnership with Deutsche Telekom and SAP to launch €1.1 billion Industrial AI Cloud for European industry
- Teaming up with Samsung to build an AI Factory to transform intelligent semiconductor manufacturing
- A partnership with Eli Lily to build an in-house AI supercomputer for drug discovery
I would be remiss not to mention the 42 and 31 activities we identified for Siemens and Microsoft, respectively. Or headline moves such as Schneider Electric’s US$3.1 billion acquisition of Cognite, Emerson’s acquisition of AspenTech, and SoftBank’s takeover of ABB’s robotics division.
Beyond these standout tech innovators, Systems Integrators (SIs) also play an essential role in deploying and scaling industrial AI. A technology provider may have an impressive AI model, robot, or software platform, but an experienced team must still make it work reliably within the customer’s operation. In this light, ABI Research’s Industrial Systems Integrators (SI) and Value-Added Resellers (VARs) market data finds that 60% of SIs possess a focus on industrial automation software and AI.
How will industrial machinery vendors adapt to the AI shift?
IMTS was once known as the International Machine Tool Show. Advanced machining remains central to the event, but its scope now extends across industrial AI, robotics, automation, additive manufacturing, digital manufacturing software, and connected factories.
The expansion of the IMTS agenda reflects how manufacturing technology is reaching a broader customer base. The industry has historically followed the 80/20 rule: roughly 20% of manufacturers are responsible for 80% of output. These larger enterprises have also tended to be the earliest adopters of advanced machines and technologies. Now that AI and automation are quite accessible to smaller organizations, their relevance is spreading to the wider manufacturing market.
Machine tool vendors have traditionally found themselves in a strategic conundrum. Their customers may say, “we want innovative capabilities.” But, at the same time, reliability and determinism cannot be compromised. A flashy demo means little if the resulting deployment introduces unpredictable behavior. One mistake can cause expensive downtime, damage equipment, or put workers at risk.
In 2025, industrial machinery was the leading industry for AI activity (21% of all activities tracked). The payoff from these deals will be on full display at IMTS. The differences between supplier offerings will become increasingly evident.
One of the top things I will be watching is how machine tool providers balance AI applications with the operational discipline their customers are accustomed to. Notable vendors across North America and Europe that I anticipate seeing, and would welcome the opportunity to connect with, include Haas Automation, TRUMPF, GROB Systems, and Flow International. For vendors based out of the Asia-Pacific region, I look forward to seeing the latest announcements from Mazak, Okuma, DMG MORI, and a few others.
Where can industrial AI deliver more autonomy without adding operational risk?
Easing and accelerating automation for a broader set of tasks and scenarios is the goal for autonomous systems. This transition will occur gradually.
The first stage keeps a human in the loop. AI recommends or acts, and a person validates the result. The next stage puts the human on the loop. The system operates more independently, while someone monitors its performance and intervenes when necessary. Eliminating human oversight entirely will take longer, particularly in safety-critical or production-critical environments.
This means the first workflows to approach greater autonomy will likely be those where the consequences of an AI error are relatively low. Documentation is one of the strongest near-term opportunities.
Shift changeovers are a good example. At the end of a shift, an operator must record what happened so the next person understands the condition of the equipment, any unresolved issues, and what requires attention. The quality of that information can vary depending on how busy the operator was or how consistently the form was completed.
AI can create an initial summary using production records and operator input, identify missing information, and prepare it for human review. Field service workflows offer a similar opportunity. In both cases, more complete records create better context and traceability over the life of a machine.
That information can become valuable beyond the individual workflow. We often frame these tools as ways to understand what went wrong, but AI can also help manufacturers understand what went right. If one factory achieves an unusually strong result, AI can document the conditions and practices that contributed to it. Production owners can then determine whether that success can be repeated at another site. This will be particularly attractive to both operational business leaders and Chief Technology Officers (CTOs)/Chief Information Officers (CIOs) tasked with scaling digital transformation projects globally.
This is why automated documentation stands out as a relatively low-risk, high-productivity use of AI. Instead of asking 1 person to write 5 documents, AI might allow that person to review 10 or 20 drafts. The technology changes the economics of the workflow by increasing what each employee can accomplish in a given day.
Why is brownfield integration the next test for industrial AI?
Over the next several years, AI will make industrial automation easier to configure and deploy. Advances in Physical AI will reduce the manual programming required for some robotic applications, improve how Autonomous Mobile Robots (AMRs) navigate dynamic environments, and allow collaborative robot arms to adapt to a wider range of shop floor tasks. Companies advancing these capabilities include NVIDIA, Palladyne AI, and Siemens on the software and platform side, alongside robotics vendors such as ABB, FANUC, KUKA, Universal Robots, and Yaskawa.
Making individual technologies smarter is only part of the challenge. Most manufacturers operate brownfield factories filled with equipment of various ages from multiple vendors. These systems were not necessarily designed to communicate with one another, making a complete replacement of existing infrastructure unrealistic.
Industrial AI adoption will happen incrementally. It may initially touch 5% of a factory, then expand to 10%, 15%, or 20% as the organization gains confidence and identifies additional opportunities. Throughout that process, most production will continue to depend on existing equipment and workflows.
I compare it to renovating a house while continuing to live in it. New electrical systems and smart controls cannot be designed in isolation. They must work safely with the structure already in place. Industrial AI faces the same requirement.
By the next IMTS, I expect the conversation to shift from making individual technologies more intelligent to connecting that intelligence across the wider operation. Digital twins will be integral to that progression. Historically, manufacturers used digital twins primarily to understand how a product would look or perform. Automakers, for example, can use them to accelerate the design of new vehicle prototypes.
The next step is connecting the product to the machines building it, the process moving it through production, and the factory surrounding the entire operation. Siemens’ Digital Twin Composer reflects this direction by bringing separate systems of record together within a more comprehensive digital twin environment.
AI becomes more useful when it can understand how one action may affect quality, throughput, maintenance, and downstream operations. At IMTS 2026, I will, therefore, be looking beyond the most dramatic demonstrations. The more consequential innovations will be those that help manufacturers introduce AI into real facilities, preserve human oversight where necessary, and expand adoption without disrupting the dependable systems already keeping production moving.
Let’s Connect at IMTS 2026
ABI Research will be attending IMTS 2026 to watch the most transformational tech trends shaping the industrial and manufacturing markets. I, along with six of my colleagues, will be available for brief discussions on industrial AI, digital twins and simulation tools, product design software, and the latest advancements across Manufacturing Execution Systems (MES), Software-Defined Automation (SDA), robotics, PLM, and other systems. Schedule a meeting with us!
Ben Weaver