Is the Digital Twin Market growing now?
In our digital twin market deck, you will find everything you need to understand the market
SUMMARY
Yes. The Digital Twin Market is growing now, with real adoption, expanding software spending and measurable industrial returns all moving in the same direction.
The market is much easier to see in deployments than in headline market-size estimates. Current 2025 estimates range from about $19 billion to $36 billion, which is too wide to treat any single figure as precise.
Manufacturing is still the center of gravity because the economics are unusually concrete. Throughput, downtime, capex, scrap and engineering time can all be measured, which makes it easier to justify further spending after a project works.
The installed base is meaningful but still far from saturated. PwC found 21% adoption in a broad operations survey, while ARC found 33% implementation among industrial respondents and another 36% planning to adopt within three years.
The strongest customer evidence is not that companies can build a twin, but that some are changing operating decisions around it. PepsiCo, for example, is moving toward proving major capital changes digitally before making them physically.
Digital twin growth is also becoming less dependent on one software category. Siemens, Dassault Systèmes and Bentley are growing the engineering and industrial platforms through which many twins are actually deployed, even when the invoice never says “digital twin.”
AI is widening the use case. A twin can give AI a realistic physical environment to reason about, while AI can make the twin more useful for simulation, optimization and automated decision support.
Physical AI may become one of the biggest new demand engines. Robots, autonomous machines and industrial AI systems need realistic simulated environments where they can fail cheaply before they are allowed to fail in a factory or warehouse.
AI data centers create a similar economic case. As power density, cooling complexity and hardware costs rise, virtual testing becomes cheap compared with the cost of getting a physical design wrong.
The biggest constraint is still integration. Connecting engineering models, sensors, operational systems and decades of inconsistent industrial data remains slow and expensive, especially for smaller companies without internal digital-twin expertise.
So the market is growing, but the standalone category may actually become harder to measure as it matures. Digital twins are increasingly becoming a standard capability inside simulation, PLM, IoT, infrastructure software, robotics and industrial AI rather than a neatly separated software market.
This market map, featured in our digital twin market deck, highlights top companies and startups in the digital twin market
What actually counts as a digital twin today?
Today, a digital twin is a digital version of a real product, machine, factory, building or physical system that stays connected to real-world information and can be used to understand, simulate or improve what happens physically.
The definition matters because “digital twin” has become a very loose label. A static CAD model of a machine is usually too limited. The same goes for a BIM model that represents a building but never changes once the building is operating. The more useful digital twins combine engineering models with sensor data, operational information or simulation so that someone can test what is happening now and what might happen after a change.
A 2025 academic review led by Michel Fett and Eckhard Kirchner compared 27 previous industry surveys and found that companies were already building twins with a mix of Java, Python, MATLAB, engineering software, sensors and other systems. There was no standard technology stack.
That makes this market unusually messy to measure. Siemens can sell a digital twin through its Xcelerator engineering stack, Bentley can build one through iTwin, Dassault Systèmes calls them Virtual Twins, and NVIDIA can provide the 3D and simulation layer through Omniverse. Much of the money therefore sits inside PLM, simulation, industrial software, IoT and infrastructure software rather than a neat product called “digital twin software.”
Is the digital twin market actually growing now?
Yes, the digital twin market is growing now, and the current evidence is much stronger than a collection of optimistic market forecasts.
More industrial companies are deploying digital twins, the large software platforms behind them are growing, manufacturers are expanding projects after seeing measurable results, and AI is creating use cases that barely existed a few years ago.
Siemens gives us one of the freshest checks on demand. In its latest reported quarter, Digital Industries revenue grew 10% on a comparable basis and its software business grew 15% to €1.8 billion. Across the first nine months of its fiscal year, Siemens said its broader digital business grew 18%.
Dassault Systèmes reported a different but similarly useful pattern. Total revenue grew only 4% in its latest quarter, while 3DEXPERIENCE software revenue grew 14% and cloud software revenue also grew 14%, mainly because of manufacturing customers. Bentley Systems’ latest quarter showed total revenue up 12.8%, subscription revenue up 13.6% and constant-currency recurring revenue growth of 12%.
Those figures include much more than digital twins, so we should not turn them into a fake industry growth rate. Still, three major engineering-software ecosystems producing this kind of recurring growth tells us the software foundation beneath the digital twin market is expanding in the real economy.
As this chart shows, and as featured in our digital twin market deck, search interest in digital twins has increased sharply
How big is the digital twin market today?
We cannot defend one exact digital twin market size today. The most reasonable reading of the available estimates puts the 2025 market somewhere around $20 billion to $35 billion, with roughly $30 billion sitting near the middle of the range.
That range is wide enough to be uncomfortable. Among eight current estimates we reviewed, the lowest 2025 figure was $18.9 billion and the highest was $36.2 billion. The median was about $29.5 billion. Analysts can differ by more than $17 billion when supposedly measuring the same market.
The forecasts diverge even more. 360iResearch expects roughly 17% annual growth, while MarketsandMarkets expects almost 48%. Most other forecasts sit somewhere between the mid-20s and low-40s.
We would use these reports to answer one question only: do researchers see a growing market? Yes, overwhelmingly. They are much less useful for telling us whether digital twin revenue is exactly $29 billion, $35 billion or something else.
| Research firm | 2025 market estimate | Forecast CAGR |
|---|---|---|
| Global Market Insights | $18.9B | 41.7% |
| MarketsandMarkets | $21.1B | 47.9% |
| Fortune Business Insights | $24.5B | 35.4% |
| IMARC | $29.3B | 25.3% |
| The Business Research Company | $29.6B | 41.9% near-term |
| 360iResearch | $34.9B | 17.1% |
| Grand View Research | $35.8B | 31.1% |
| Mordor Intelligence | $36.2B | 36.0% |
If you want more recent data on this point, please see our latest digital twin market report.
Are companies actually using digital twins, or are most projects still experiments?
Digital twin adoption has moved well beyond experiments, but the technology is still far from being used by every company.
PwC’s 2025 Digital Trends in Operations survey found that 21% of respondents were using digital twins. That is meaningful adoption, although it also means roughly four out of five respondents were still outside the market.
An industrial survey from ARC Advisory Group gives us a more aggressive picture. ARC reported in 2026 that 33% of participating companies had already implemented digital twin and simulation technologies, while another 36% expected to implement them within three years.
The different results make sense. Digital twins have much more value for a company running factories, grids or heavy equipment than for the average corporate office. ARC’s industrial audience should therefore show higher adoption than a broader operations survey.
What interests us most is the pipeline. ARC’s results put almost seven out of ten industrial respondents either inside the market already or planning to enter it. We are no longer looking at a technology where adoption depends on convincing companies that the concept exists.
This chart, included in our digital twin market deck, shows annual VC investment in digital twin startups
Are digital twins actually saving companies money?
Yes, digital twins are producing large enough savings in real industrial deployments to justify more spending.
PepsiCo currently provides one of the clearest examples. Working with Siemens and NVIDIA, the company built detailed digital twins of selected manufacturing and warehouse operations. Siemens reports that the initial deployment increased throughput by 20%, cut expected capital expenditure by 10% to 15%, and allowed teams to find more than 90% of potential operational problems before physical implementation.
The project also changed PepsiCo’s internal process. At a Siemens event in 2026, PepsiCo manufacturing executive Steve Hoinka said the company wanted major capital decisions to be proven digitally before they were implemented physically. That is a much stronger commitment than running an innovation pilot.
Infrastructure projects show a similar pattern. Viscan used Bentley’s technology on Germany’s B29 infrastructure project and reported 60% faster data processing alongside a 20% reduction in project time. At Malaysia’s Bakun hydroelectric plant, Sarawak Energy reported that digital workflows reduced the time required for as-built modeling by 70% and cut onsite staffing needs by 30%.
These figures come from companies and vendors involved in the deployments, so we would not average them into a universal digital twin ROI. What they show quite clearly is that the technology can already pay for itself through higher throughput, shorter engineering cycles, lower capex and less field work.
| Deployment | What the digital twin was used for | Reported result |
|---|---|---|
| PepsiCo | Factory and warehouse redesign | Throughput +20%; capex down 10–15%; 90%+ of potential issues found before implementation |
| Viscan | Infrastructure reality capture | Data processing 60% faster; project time down 20% |
| Sarawak Energy | Hydroelectric plant operations | As-built modeling 70% faster; onsite staffing need down 30% |
Which industries are driving digital twin growth right now?
Manufacturing still drives most digital twin activity, but the market is spreading quickly into energy, infrastructure, shipbuilding, data centers and robotics.
The geographical spread is also broader than it used to be. North America remains the biggest region in most market estimates, generally accounting for roughly one-third of current spending, but the actual deployments now stretch across Europe and Asia.
HD Hyundai, for example, is building a digital shipbuilding platform in South Korea using Siemens technology. The companies recently expanded that work into a low-triple-digit-million-dollar agreement for a repeatable AI-powered digital shipyard model, including U.S. shipbuilding. TSMC has used NVIDIA Omniverse technology in semiconductor-fab design and construction. Ola Electric has used virtual factory technology in India. Sarawak Energy is using digital workflows in Malaysia.
Those projects also cover very different physical systems. We are seeing ships, semiconductor fabs, factories, electrical networks, roads, dams, warehouses and AI data centers represented digitally.
The market is becoming less dependent on one narrow industrial niche. The common denominator is an expensive physical system where planning mistakes, downtime or inefficient operation cost real money.
This chart, included in our digital twin market deck, breaks down Neara's playbook in digital twins
Is manufacturing still the biggest digital twin opportunity?
Manufacturing remains the strongest digital twin market today because factories make the economics unusually easy to understand.
A manufacturer can test the layout of a production line before installing anything, simulate bottlenecks, predict equipment problems, optimize material flow and train robots without stopping a real factory. Every improvement can eventually be measured in throughput, downtime, scrap, engineering hours or capital spending.
That is why Siemens, Dassault Systèmes, ABB, Rockwell Automation, PTC, Schneider Electric and NVIDIA are all spending heavily around virtual factories and industrial simulation.
The recent PepsiCo deployment is particularly useful because it connects virtual design directly with plant economics. As discussed above, PepsiCo says its early work produced a 20% throughput improvement and a 10% to 15% capex reduction. HD Hyundai’s recent nine-figure Siemens agreement pushes the same logic into shipyards, where the physical product and production process are even more complex.
Manufacturing also benefits from decades of CAD, simulation and product-lifecycle data that other industries often lack. Digital twins therefore fit into workflows engineers already use rather than requiring an entirely new operating model.
Are energy and infrastructure becoming serious digital twin markets?
Yes, energy and infrastructure are becoming major digital twin markets because grids, transport systems, water networks and large physical assets are getting harder to operate with static engineering models.
Utilities are a good example. Modern electricity networks have distributed solar, batteries, electric vehicles, smart meters and increasingly volatile demand. Operators need to understand what is happening across thousands or millions of physical points rather than look at a design that was produced years ago.
Xcel Energy has built a digital representation of its electricity network around data from meters, transformers, feeders, SCADA systems and other operational sources. Its advanced meters alone can generate roughly six billion data points per day. The utility uses these models for problems such as transformer loading, voltage analysis and grid planning.
Schneider Electric and ETAP are currently pushing further into this market with physics-based grid digital twins for utilities and critical infrastructure. Saint John Energy in Canada has also opened a digital twin of its network to researchers and technology providers so they can test new energy applications against a realistic system.
Bentley offers a useful financial cross-check. The company sells engineering software across roads, rail, water, energy and other infrastructure, with iTwin increasingly acting as its digital-twin layer. Its latest subscription revenue grew 13.6% and recurring revenue grew 12% in constant currency.
Infrastructure is becoming one of the clearest places where digital twins are moving from design software into day-to-day operations.
This chart, included in our digital twin market deck, shows annual funding in digital twin startups
Are digital twin software vendors actually growing?
Yes, the companies most exposed to digital twin workflows are growing today, although none of them reports a clean “digital twin revenue” number.
The latest company results are fairly consistent. Siemens’ software business inside Digital Industries grew 15% in its latest quarter. Dassault Systèmes’ 3DEXPERIENCE and cloud software revenues both grew 14%. Bentley’s subscription revenue grew 13.6%.
The comparison becomes more interesting when we look at the underlying recurring businesses. Siemens said organic software ARR reached €5.7 billion and grew 11%. Dassault reported an annual run rate of €4.4 billion, up 6%. Bentley ended its latest quarter with $1.536 billion of ARR, up 12% in constant currency.
We should keep the interpretation narrow. Siemens sells EDA, PLM and industrial software. Dassault sells far more than virtual twins. Bentley sells engineering applications across the full infrastructure lifecycle. Their growth does tell us something important about the customers paying for digital-twin capabilities: engineering and industrial software budgets are expanding rather than collapsing.
| Company | Latest relevant growth | Digital twin exposure |
|---|---|---|
| Siemens | Software revenue +15%; organic software ARR +11% | Xcelerator, simulation, PLM, Digital Twin Composer |
| Dassault Systèmes | 3DEXPERIENCE +14%; cloud software +14% | Virtual Twins and 3DEXPERIENCE |
| Bentley Systems | Subscriptions +13.6%; constant-currency ARR +12% | iTwin and infrastructure engineering |
Is AI making digital twins more useful now?
AI is making digital twins much more useful because companies can increasingly move from watching a physical system to testing and recommending what should happen next.
Older digital-twin projects were often built around monitoring, predictive maintenance and engineering simulation. Those are still useful. The newer opportunity is to give AI systems a realistic model of the physical environment they are supposed to understand.
Siemens’ Digital Twin Composer is an example. The software combines engineering data, simulation and real-time information so teams can test production changes virtually. PepsiCo is already using AI agents inside this environment to simulate and refine changes before implementing them physically.
Dassault Systèmes is moving in the same direction with AI-powered Virtual Companions inside 3DEXPERIENCE. NVIDIA approaches the problem from another angle, using Omniverse and its simulation stack to give robots and industrial AI systems environments where they can train.
Generative AI could also make parts of the digital-twin process cheaper. Models can help generate assets, clean data and automate modeling work that previously required specialists. That may reduce the price of creating a twin while making accurate physical data more valuable.
AI is expanding the addressable use cases. The twin gives AI something concrete to reason about, while AI makes the twin useful for more than visualization and human-led simulation.
This chart, included in our digital twin market deck, compares the main business model options for digital twin enterprise software platforms
Is physical AI creating a new wave of digital twin demand?
Physical AI is creating one of the strongest new reasons to build digital twins today because robots need somewhere safe and realistic to learn.
A language model can make millions of mistakes during training without breaking a machine. A warehouse robot or autonomous forklift does not have that luxury. Real-world training is slower, more expensive and potentially dangerous.
Digital twins solve part of that problem by giving robots a simulated factory, warehouse or production line where large numbers of scenarios can be tested before deployment.
NVIDIA’s recent manufacturing push shows how quickly this use case is spreading. Caterpillar is using Omniverse for factory and supply-chain digital twins. Lucid Motors is using factory twins for planning, optimization and robotics training. Toyota is creating a twin of its Georgetown facility to test automation scenarios. TSMC is using Omniverse in fab design while developing robotics for its Arizona facility. Wistron is using NVIDIA technology to digitally test systems assembled in Texas.
NVIDIA has also expanded its Mega Omniverse blueprint for factory-scale twins, with Siemens among the first industrial software providers supporting it and FANUC and Foxconn connecting robot models.
This creates demand for digital twins even when the customer starts with a robotics problem. As more physical AI gets deployed, simulation infrastructure becomes part of the robotics stack.
If you want more recent data on this point, please see our latest digital twin market report.
Are AI data centers becoming a real digital twin market?
Yes, AI data centers are emerging as a valuable digital twin use case because the physical infrastructure has become enormously complex and expensive.
A large AI facility has to coordinate compute racks, networking, electricity, cooling, building systems and backup infrastructure. Power densities are rising quickly, and getting the thermal or electrical design wrong can delay billions of dollars of hardware.
NVIDIA has been building this idea into its Omniverse DSX architecture, where companies can model AI factories before and after construction. Partners around these workflows include Siemens, Schneider Electric, Vertiv, Cadence, Jacobs and PTC.
The broader spending environment is already moving fast. Siemens said its Smart Infrastructure data-center business recorded triple-digit order growth during the first nine months of its fiscal year, reaching roughly €6 billion of orders.
Those €6 billion are obviously not digital twin sales. They tell us that one of the world’s largest industrial software vendors is seeing enormous growth in the physical systems where simulation and digital-twin tools can have unusually high value.
As AI infrastructure gets larger, denser and more expensive, the cost of testing designs virtually becomes small compared with the cost of making a physical mistake.
This chart, featured in our digital twin market deck, illustrates how revenue is divided among customer segments in the digital twin market
Do the big engineering-software acquisitions tell us anything about digital twins?
Yes, the recent engineering-software acquisition wave shows that simulation and digital engineering have become strategically valuable assets.
Siemens completed its acquisition of Altair Engineering for an enterprise value of about $10 billion. Altair brought simulation, high-performance computing, data science and AI capabilities that fit directly into Siemens’ comprehensive digital-twin strategy.
Synopsys went even bigger by completing its roughly $35 billion acquisition of Ansys. The combination brings semiconductor design and multiphysics simulation into the same software company. Synopsys described the addressable engineering-software opportunity around the combined group at roughly $31 billion.
Bentley acquired Cesium, a 3D geospatial platform whose technology can represent large physical environments. Bentley is combining Cesium’s geospatial layer with iTwin so infrastructure customers can work with digital representations that extend beyond an individual bridge, road or building.
Calling the entire $45 billion-plus combined value of the Siemens-Altair and Synopsys-Ansys deals “digital twin M&A” would be misleading. Simulation software serves many other purposes. The acquisitions still show that large software companies are paying heavily for the physics, engineering data and computation that make advanced digital twins possible.
Is digital twin software really a standalone market?
Digital twins are increasingly sold as a capability inside larger software platforms, which makes the standalone “digital twin software market” harder to defend.
Microsoft offers Azure Digital Twins and AWS continues to offer IoT TwinMaker, so dedicated products certainly exist. Specialist digital-twin vendors also operate across manufacturing, buildings and infrastructure.
Yet many large deployments these days combine several products. A company might use Siemens Teamcenter for engineering data, simulation tools for physics, NVIDIA Omniverse for a 3D environment and an IoT platform for real-time machine information. Another company might use Bentley’s engineering stack and iTwin rather than buying anything explicitly labeled “digital twin software.”
AWS offers an interesting clue here. TwinMaker remains available, while AWS has ended support for several adjacent specialized industrial and simulation services, including IoT Analytics, IoT Events and SimSpace Weaver. Customers appear increasingly comfortable assembling these capabilities through broader cloud and data platforms.
This is why the category can feel bigger operationally while becoming less distinct commercially. Digital twins are spreading through engineering and industrial software even when the invoice never contains the words “digital twin.”
If you want more recent data on this point, please see our latest digital twin market report.
This chart, included in our digital twin market deck, shows how industrial digital twin platform technology has evolved over time
Are digital twin standards finally making deployment easier?
Digital twin standards are improving interoperability today, but connecting all the underlying industrial data is still hard.
The 2025 review of 27 industry surveys found the same complaints appearing repeatedly: inconsistent communication standards, incompatible datasets, different definitions for the same information and difficulty linking engineering systems with operational systems.
Some of that friction is starting to ease. The Asset Administration Shell is becoming a standardized way to describe industrial assets and their data. OpenUSD is also gaining traction as a common framework for composing complex 3D environments.
The Alliance for OpenUSD now includes major industrial and software companies, and the group has started the ISO certification process for its Core Specification. Siemens, NVIDIA, Bentley/Cesium, PTC, Schneider Electric and other industrial ecosystems increasingly support OpenUSD-based workflows.
Still, a common 3D format only solves part of the problem. A factory can have decades of equipment IDs, maintenance records, sensor names, ERP data and engineering models that were never designed to work together.
The technology needed to build digital twins is mature enough. Cleaning and connecting the customer’s physical-world data remains one of the slowest parts of deployment.
Are digital twins still mainly for big companies?
Digital twins are still much easier to justify for large companies with expensive factories, infrastructure or equipment.
PepsiCo can spend heavily on simulation because a few percentage points of extra throughput across a large manufacturing network are worth a lot of money. HD Hyundai can justify a nine-figure digitalization agreement because shipyards and vessels are extraordinarily complex assets. Utilities can justify network twins because a grid problem can affect thousands of customers.
Smaller manufacturers face a tougher calculation. Building a high-quality digital twin can require engineering software, sensors, integration work, specialists and clean historical data before the company sees a return.
The industry surveys reviewed by Fett and Kirchner reinforce this. Larger companies were more likely to have internal digital-twin expertise, while smaller organizations depended much more on outside providers. Several surveys also found strong demand for simpler and more standardized tools.
This is one of the biggest remaining growth opportunities. If vendors can turn digital twins into repeatable applications that a mid-sized company can deploy without a custom engineering program, the number of potential customers increases dramatically.
In our digital twin market deck, we identify pain points entrepreneurs should prioritize
Is the digital twin market growing partly because everything gets called a digital twin now?
Yes, loose definitions probably inflate some digital twin market estimates, so we should separate real adoption from category marketing.
The language has expanded quickly. Dassault talks about Virtual Twins. Siemens increasingly links digital twins with the industrial metaverse and industrial AI. NVIDIA places them inside physical AI. Infrastructure companies mix digital twins with BIM, reality capture and asset management.
Some of these combinations are legitimate. A modern factory twin genuinely does combine engineering models, sensor data, AI and simulation. Separating every technology into its own market would be equally artificial.
The problem comes when the same software spending can be counted several ways. A simulation package might appear inside the simulation market, the engineering-software market, the industrial-AI market and the digital-twin market depending on who publishes the forecast.
That explains why current market estimates disagree by nearly twofold on today’s market size and even more on future growth rates.
We trust the deployment evidence more than the giant TAM forecasts. Companies are putting digital twins into factories, grids, shipyards and robotics programs. That activity is real even if nobody can measure the category to the nearest billion dollars.
If you want more recent data on this point, please see our latest digital twin market report.
What could stop the digital twin market from growing?
Digital twin growth could slow if companies keep needing expensive custom integration for every deployment.
Integration is already one of the biggest complaints. PwC’s 2025 operations survey found that 47% of respondents struggling to get expected value from technology investments cited integration complexity, while 44% cited data problems. Those issues hit digital twins particularly hard because the model depends on accurate information from physical equipment and several business systems.
Skills are another bottleneck. The industry-survey review published in 2025 found that many companies still relied heavily on outside expertise, with one underlying survey reporting that 72% of respondents needed additional IT skills.
Maintenance also gets underestimated. Creating a convincing model for a pilot is one job. Keeping that model accurate after machines are replaced, software changes and factories are reorganized is much harder.
The standalone category could also become less visible even while usage grows. Customers may increasingly buy industrial AI, robotics, simulation or engineering platforms where digital-twin functionality comes bundled into the broader system.
For now, these issues look more like limits on the speed of adoption than reasons for the market to reverse.
This chart, included in our digital twin market deck, illustrates how revenue is divided geographically across Europe, Asia, North America, Africa, and South America in the digital twin market
Is the Digital Twin Market growing now?
Yes. The Digital Twin Market is growing now, and the evidence is strong enough that we would call the growth real rather than speculative.
Current adoption surveys show a meaningful installed base with plenty of companies still planning deployments. Real customers are reporting measurable improvements in throughput, capex, engineering time and field work. Meanwhile, the latest results from Siemens, Dassault Systèmes and Bentley show double-digit growth in several of the engineering-software businesses through which these projects are actually deployed.
The market is also gaining new sources of demand. Manufacturing remains the biggest use case, while power grids, infrastructure, shipbuilding, robotics and AI data centers are adding another layer of growth. Physical AI may be especially important because robots need simulated environments in which they can train and fail cheaply.
We would be much more cautious with the headline market forecasts. Current estimates of the market’s size differ by almost twofold, and the predicted growth rates vary from roughly 17% to nearly 48%. Digital twins overlap too heavily with simulation, PLM, IoT, BIM and industrial AI for those numbers to be treated as precise.
The stronger conclusion comes from what companies are actually doing. Customers are moving digital twins into operational workflows, software platforms around them are growing, and some companies are expanding projects after seeing clear financial results.
So yes, this market is growing. What remains uncertain is how much of that growth will eventually belong to a distinct “digital twin” software category, because the technology is increasingly becoming a standard capability inside the much larger industrial-software and AI stack.
If you want more recent data on this point, please see our latest digital twin market report.
OUR METHODOLOGY
This analysis tests whether the Digital Twin Market is growing now by looking at the parts of the market that can be observed directly: adoption, live deployments, customer outcomes, engineering-software growth, expansion into new use cases, strategic transactions, standards work and current market estimates.
We do not treat a single market-size forecast as the answer. Digital twins overlap heavily with PLM, simulation, IoT, BIM, industrial software, infrastructure software and industrial AI, so published estimates often measure different boundaries. We use the forecast range mainly to show the direction researchers expect and the amount of uncertainty around the category.
Adoption evidence comes from two different populations. PwC’s 2025 Digital Trends in Operations survey provides a broader operations view, while ARC Advisory Group’s industrial survey is more concentrated on companies where digital twins are naturally useful. We interpret the difference between their adoption rates as a difference in sample mix rather than a contradiction.
For operating evidence, we prioritize deployments with reported outcomes rather than showcase projects that only demonstrate technical feasibility. PepsiCo, Viscan and Sarawak Energy are useful because the reported results connect digital-twin use to throughput, capex, engineering time, project time or onsite staffing.
Company growth figures are used as a cross-check, not as a digital-twin revenue estimate. Siemens, Dassault Systèmes and Bentley all sell much broader engineering-software portfolios, but their latest software, subscription and recurring-revenue growth helps show whether the commercial platforms carrying digital-twin capabilities are expanding.
We treat AI, physical AI and AI data centers as new demand layers rather than separate proof that the whole market is growing. NVIDIA’s manufacturing and robotics deployments, Siemens’ Digital Twin Composer work and Schneider Electric/ETAP’s AI-factory and grid-twin projects show how the same underlying capability is moving into newer physical systems.
Strategic acquisitions are also kept in their proper scope. Siemens’ Altair acquisition, Synopsys’ Ansys acquisition and Bentley’s Cesium acquisition are not counted as digital-twin market revenue. They are evidence that simulation, engineering data, geospatial context and physical-system modeling have become strategically valuable parts of larger software stacks.
Standards work is included because interoperability affects how quickly the market can scale. We use the Industrial Digital Twin Association’s Asset Administration Shell specifications and the Alliance for OpenUSD’s Core Specification as current references for the effort to make industrial assets and 3D environments easier to connect across systems.
Key sources used for this analysis include: PwC’s 2025 Digital Trends in Operations survey, ARC Advisory Group’s digital twin adoption survey, Siemens’ Q3 FY2026 results, Dassault Systèmes’ Q2 2026 results, Bentley Systems’ Q2 2026 filing, Siemens on PepsiCo and Digital Twin Composer, NVIDIA on manufacturing, robotics and physical AI, Schneider Electric and ETAP on grid digital twins, Siemens on the Altair acquisition, Synopsys on the Ansys acquisition, the Industrial Digital Twin Association’s Asset Administration Shell specifications, and the Alliance for OpenUSD’s Core Specification.
This chart, included in our digital twin market deck, shows annual VC investment in digital twin startups
