Executive summary
Industrial software is reaching an inflection point. Platforms are larger than ever, M&A activity has seen surges in the last 5 years, and marketing decks promise AI will deliver new frontiers on closing the loop between data and autonomous operations. But insights from the plant floor tell us a different story: operators still struggle with disconnected systems, data lakes that produce little insight, and AI pilots that have yet to move KPIs. The growing maturity of digital operations has brought clarity to the building blocks that will enable autonomous operations – a contextualized data foundation, interoperability across systems, bidirectional communication at the sensor level – but that same clarity reveals just how long the road ahead still is. The gap between products and promises is a multi-billion-dollar opportunity for software incumbents, disruptors, and investors.
The industrial software landscape today
Industrial operations have been redefined roughly once a generation. Mechanization in the early 19th century replaced labor-intensive human tasks while computerization in the mid-20th century brought centralized control to industrial machines. Over the last two decades, digitalization has transformed the volume and role of data in industrial processes. The resulting software stack can be seen in Figure 1. Each of these phases of industry has been marked by new technology bringing productivity gains.
Figure 1: The industrial software ecosystem across the production lifecycle
The question is whether the current shift in industrial software will deliver the defining productivity gains of autonomous operations or simply a more sophisticated version of the data-collection exercise that came before it. AI capabilities have reached a point where real industrial use cases are emerging, and agents are proving return on investment in other end-markets, yet scaled industrial deployments have lagged. These factors and the overall industrial economic tailwinds – reshoring, energy transition, aging infrastructure – are driving significant capex by software customers. Corporate and private capital have been eager to chase this opportunity as platform vendors and private equity compete for a limited set of scaled assets.
The industrial software vendor ecosystem can be usefully divided into four archetypes that vary in breadth of capabilities, stack positioning, and origin. Understanding the strategic posture of each is essential to understanding the landscape of solutions and where value is accruing.
Figure 2: Industrial software competitor segmentation and ecosystem participation
- IT / ERP – players who entered the industrial world from the top of the stack anchored in financial, supply chain, and workforce solutions and home of enterprise data. These players have lobbied for industrial companies to lift data into the cloud where they can control the data and begin to build the foundations of AI infrastructure. Their advantage is scale and c-suite relationships, while still developing credibility handling operational and engineering tasks due to lack of domain expertise.
- OT Control – these are vendors who entered the stack from the bottom. Honeywell, Rockwell, Emerson, etc. started with hardware and the control systems that physically run the facility. This creates a natural moat of installed base, domain knowledge, and safety certifications that are hard to replace. OT Control players have expanded their participation across the industrial software ecosystem through organic and inorganic platform development (see figure 4). Yet, stitching together acquired products into a cohesive platform has proven difficult, leaving most portfolios fragmented and lagging software pure-plays.
- OT Lifecycle – players who historically anchored the design portion of the production lifecycle, specializing in engineering tools used for design and simulation of the facility (e.g., CAD, PLM, EAD). Their strategy is increasingly to extend these design-time tools into runtime operations, positioning the digital twin as the connective tissue between how an asset was engineered and how it actually performs.
- Point Solutions – specialists who offer best-of-breed capabilities in one segment of the ecosystem (see figure 1). These competitors differentiate themselves through depth of capabilities for a specific industry, software product, or – increasingly – AI integration. The best of these assets earn premium valuations as they are difficult for platforms to replicate / build organically.
These four archetypes serve a customer base that is itself far from uniform. Industrial software buyers span process, hybrid, and discrete manufacturing end-markets. Nuances in data maturity and process workflows affect the adoption level and applicability of solutions within different end-markets. Process industries have historically anchored OT control platforms, discrete industries have pronounced needs for OT lifecycle platforms controlling production lines, while hybrid industries can often be the most demanding buyers (e.g., life sciences and pharmaceuticals) requiring tight integrations between engineering, operations, and enterprise systems.
Given all the changes and shifts in the landscape, the question remains – which archetype produces the winner? The answer hinges on whether owning the entire workflow is necessary to capture value, which solutions will be commoditized by AI, and what role today's plant systems (e.g., DCS, PLC) will play in an increasingly software-defined future.
Data constraints and the limitations of AI today
Before we can identify who will win in the future of closed-loop process control, we must first take stock of where we sit today.
A key barrier unique to industrial settings is the complexity of OT data. While in other settings the default data type is transactions, documents, or customer data, in the industrial world, timeseries data is the standard. Temperature, pressure, vibration, and voltage are streamed from sensors and control systems at sampling rates from seconds to sub-millisecond intervals.
This data is usually difficult to work with for many reasons: sampling rates differ between tags, sensors drift and can report false values that skew averages, and clocks across systems are not always synchronized. A single mid-sized refinery can generate billions of timeseries points per day, streaming from hundreds of different assets with tag naming conventions that can be unique to each asset / data stream. Tag naming conventions like “FI-101.PV” carry value that is only unlocked once decoded by a person or system that understands its structure.
The dominant response to this complexity of timeseries by vendors over the last decade has been to lift OT data off historians into the cloud, land it in a data lake, and let data science teams work with it. However, by the time timeseries reaches the cloud, the crucial context of the data’s connection to physical operations in which it was generated is lost. Cloud infrastructure solves a storage and compute problem, but this is a sidestep from the real binding constraint.
Context in the industrial environment has many layers. To unlock the full value of a datapoint it must be connected to the asset, process, engineering, and event context. The asset hierarchy tells us what sensor sits on which piece of equipment at which plant; the process context contains what operation was running and what product was being made; engineering information connects the design specifications and control logic governing that asset; event context adds maintenance and alarm details. This context exists, yet is rarely captured and assigned to datapoints within current data structures as it is scattered across historians, engineering systems, maintenance management systems, paper documents, and technicians’ knowledge. The stitching together of these systems that are physically siloed and recording data in different syntaxes has largely been pushed down the road with hopes that data lake cloud migration would solve these problems.
Figure 3: Evolution of industrial data storage solutions
But simply putting the data in the same place does not mean it is contextualized. This fact becomes clear the moment customers try to unlock real value from their data – this is exacerbated by AI that is trained or deployed on top of industrial data. After several years of heavy investments and no shortage of vendor enthusiasm, ROI has been sobering. The set of wins has clustered narrowly within three categories: conversational assistants, training tools shortening the runway for new technicians to replace an aging workforce, and anomaly detection in equipment behavior. These are real, but adjacent to plant operations rather than embedded within them. Use cases that would provide meaningful change to operational economics and productivity are prescriptive maintenance that prevents failures, autonomous process optimization that outperforms a skilled operator, generative engineering compressing design cycles, or agentic workflow coordination. The gap to unlocking these high-value applications is not a model capacity gap – it is an input gap. Without cleaned, contextualized data, even the most advanced models are reduced to pattern-matching on noise.
M&A and the battle for scale
The who, what, and when of investments across the industrial software landscape can point towards the why (investment thesis) that is shaping the future of manufacturing. To understand the deals influencing industrial software, let’s look at a few examples of where companies and investors are placing their bets:
OT Vendors Building Out Software Platforms – companies that were built on industrial hardware, control, and engineering products are aggressively expanding their software product portfolio through inorganic activity. Few companies better exemplify this pattern than Schneider Electric and their acquisitions of AVEVA (completed 2023, ~$10.8B total), OSIsoft (acquired by AVEVA, 2020, ~$5.0B), RIB Software (2020, $1.5B), ETAP (2021, ~$0.3B), and more. Competitors on the OT control side (e.g., Emerson + AspenTech) have pursued similar actions to build out platforms with capabilities across the plant stack and end-markets (see figure 4).
Industrial Operators Investing in Data Foundations – a second, quieter pattern involves industrial operators putting capital directly into the data layer sitting beneath software applications. In 2022, Saudi Aramco acquired a 7.4% stake in Cognite, a leading industrial data fabric, for over $100M at a valuation of over $1.5B. In 2023, Shell, in partnership with Norwegian VC firm Idekapital, invested $90M in Kongsberg Digital, a provider of unified industrial data workspaces upon which advanced digital twins are built, at a valuation of over $0.5B.
Nobody knows where pain points sit within the software stack better than the customers of software products – the players investing directly in vendors are trying to unlock real insights and value. The investment by two of the largest customers into foundational data elements highlights the importance that data ownership and contextualization will play in the next era of industry.
Private Equity Focus on Point Solutions – the third pattern is private capital competing for point solutions that can differentiate themselves and scale revenues. A clear recent example is Blackstone and Vista Equity Partners’ joint acquisition of Energy Exemplar, an advanced electricity grid simulation vendor, for approximately $1.1B in 2024. Energy Exemplar serves over 500 utilities across 79 countries exemplifies the kind of assets PE is chasing: category-leading point solutions in structurally growing end-markets with high recurring revenue.
Equally informative are deals that did not happen. In early 2024, Schneider Electric entered advanced talks to acquire Bentley Systems, an infrastructure engineering software platform. The deal collapsed later in the year over a combination of valuation concerns and reluctance to relinquish control of a software company to a hardware-led acquirer.
Figure 4: Industrial software available assets and consolidated platforms by revenue
This near miss reveals a structural feature of the industrial software market: there are very few scaled software assets in the $150M to $500M revenue range available at any given time (see figure 4). Once companies cross the threshold of scale, credibility, and recurring revenue they are quickly absorbed into either larger platform players, or private equity firms paying a premium for category leadership. Scaled industrial software assets trade at premium multiples because the supply is constrained, and the window between "interesting" and "acquired" has compressed.
The Battle for Defensibility
The winners in the next decade of industrial software are likely to be players that focus on the high-value, strategic assets that are critical to winning – these include data contextualization, system interoperability, and bidirectional communication at the sensor level – assets need to move forward on the path to an autonomous future s
Figure 5: Industrial software strengths and exposures
For IT and ERP vendors, a key battle will be overcoming the barrier presented by OT timeseries complexity and ability to move down the software stack. The IT vendors that win will be those that move beyond selling infrastructure to actually doing the work of integrating, streaming, and cleaning data from industrial assets – work that requires deep domain expertise. In conversations with some of the largest global IT firms, executives have stressed the limited success they have achieved in this endeavor despite investments in software development and human capital.
For OT Control vendors, extraction and ownership of data from their installed hardware will present a defensible moat for a limited timeframe. Future customers will place increasing value on not only data extraction but also real-time communication back to devices. The ability to read and write back to plant level sources will be table stakes for enablement of autonomous operations – the question is whether real-time OT APIs will be commoditized by the new wave of vibe-coders within customers’ IT departments.
The challenge for OT lifecycle vendors when trying to offer a platform for autonomous operations is extending across the asset lifecycle from engineering to real-time operations. Vendors will try to either own operations solutions themselves or more likely integrate heavily with real-time products. This will require interoperability (communication) across products and platforms offered by different vendors – a problem yet to be solved consistently today.
For OT control and lifecycle vendors both, being early to deliver valuable insights from embedding AI into products / platforms will be key to growing share. While surface-level applications (e.g., conversational assistants) are enough to satisfy customers today, they will be eager to jump platforms once higher value AI applications are realized. If not for their own desires, then to appease shareholders.
When assessing point solution vendors, key criteria will include the degree of embeddedness in real-time workflows, and ability to turn data access into insights. Each point solution will eventually fall into one of three buckets: 1) solutions that are negatively impacted / replaced by AI, 2) solutions forced to evolve by AI, and 3) solutions that benefit from AI. Understanding why analytics will be replaced while systems of record retain value is key for making strategic investments in the coming years.
The Long Road Ahead
For the first time, the building blocks of autonomous industrial operations are becoming more clear. The clarity of these building blocks is itself a large step with vendors now knowing what they need to build and customers knowing what they want. And yet that same clarity reveals the distance still to travel. The data layer is still developing and fragmented, AI applications are still separated from operations largely due to reliability and explainability concerns, and data extraction from sensors and devices is far outpacing the ability to communicate in the other direction. These capabilities remain at a horizon that is still years away – not quarters. For corporate strategists and private investors, a wide field of view and understanding of the history of industry will help identify fads from assets that will compound over time.
Figure 6: Defensibility assessment by segment of the software ecosystem / stack
In the near /medium-term, defensibility will be driven by operational embeddedness / integration depth and data moats. Solutions with the largest data moats are those that have access to or create proprietary data, high regulatory barriers, or large network / ecosystem effects. Operational embeddedness includes complexity / specialization of workflows, operational criticality, and real-time output. Software solutions that rank highly across these technological and operational barriers will be the most insulated from disruptions due to AI and agentic workflows and will accrue value in the near-medium term.
AI, agents, and autonomous operations present both the greatest threat and opportunity for industrial software investors and vendors in a generation. How your company will be affected is just as much a strategy challenge as it is a technology one.
If your organization is faced with this shift, Red Chalk Group is here to help. Contact us to learn more.
