What are the main business models in the digital twin market?
In our digital twin market deck, you will find everything you need to understand the market
SUMMARY
The main business models in the digital twin market are recurring software, capacity and consumption pricing, bundled industrial software, cloud or hardware pull-through, implementation services, and a smaller but growing layer of outcome-based contracts.
The strongest commercial model is not selling a digital representation by itself. The durable money usually comes from owning the engineering, operating or maintenance workflow that keeps customers using the twin after it has been created.
Subscriptions already dominate the industrial-software businesses where many twins live. Bentley, Dassault Systèmes and Siemens all show recurring revenue becoming the default commercial structure around engineering, PLM, simulation and infrastructure software.
Seat-based pricing is becoming less useful as a measure of digital-twin value. Autodesk, Bentley, AWS and Azure increasingly tie revenue to assets, data capacity, processing, API activity or cloud consumption, allowing customer spending to grow without adding more users.
Large incumbents have an important structural advantage because they do not need to sell the twin separately. Siemens can use twin capabilities to strengthen PLM and automation, Dassault can embed them in 3DEXPERIENCE, and Autodesk can connect them directly to BIM and building data.
Hyperscalers operate under a different logic. The explicit AWS or Azure digital-twin charge can be modest because the same deployment can generate years of additional spending on IoT ingestion, storage, databases, analytics, visualization and compute.
NVIDIA pushes that logic even further. Omniverse can now be free at the platform level while NVIDIA monetizes enterprise support and, more importantly, the GPU-heavy simulation, synthetic-data and physical-AI workloads that the platform encourages.
Implementation work remains necessary because industrial data is messy, but it is a weaker core model for a scalable software company. Bentley's latest reported economics show an enormous margin gap between subscriptions and services, making services more useful as a deployment engine than as the long-term profit pool.
Vertical digital twins appear easier to monetize than generic platforms because the customer can connect spending to a specific operational problem. Inspecting infrastructure, reducing HVAC energy use or monitoring equipment condition gives the buyer a much clearer economic reason to pay than simply having a better digital model.
The market is therefore moving toward a simple pattern: the underlying digital representation is becoming infrastructure, while the valuable commercial layers are shifting toward recurring workflows, asset-level analytics, consumption, computing and measurable operating outcomes.
What are the main business models in the digital twin market?
The digital twin market is mainly a recurring software and consumption business, with bundling, services, hardware pull-through and outcome-based contracts creating additional ways to make money.
The strongest proven model is recurring vertical software. Siemens, Dassault and Bentley show how valuable digital-twin capabilities become when they live inside software that customers use continuously for engineering, manufacturing or infrastructure operations.
Consumption is becoming the second major model. Autodesk scales Tandem pricing with the size and data intensity of the twin. AWS and Azure meter cloud activity. Bentley is going further by charging against API usage and physical assets. These models let revenue grow even when user counts barely change.
Bundling can be just as powerful for large incumbents. A digital twin can help sell or retain PLM, simulation, BIM, automation, cloud infrastructure and eventually hardware. NVIDIA's recent Omniverse licensing change shows how far that logic can go: the platform itself can even become free when the company has stronger monetization elsewhere in the stack.
Implementation services remain an important part of deployment, particularly in messy industrial environments, but their margins and scalability make them less attractive as the core business for a software company. Outcome pricing could eventually become much bigger because it ties fees to uptime, energy savings or production output, although the commercial evidence is still early.
The best digital twin businesses today own something beyond the twin itself: a recurring workflow, a valuable physical asset relationship, a stream of cloud consumption, a hardware ecosystem or a measurable operating outcome. The generic digital representation is gradually becoming infrastructure. Most of the durable money is being made above it, around it or because of what it allows the customer to do.
| Business model | How companies get paid | Current strength |
|---|---|---|
| Recurring software | SaaS, subscriptions, support and renewals | Strongest proven model |
| Capacity / consumption | Assets, data, API calls, compute or platform usage | Growing quickly |
| Bundled software | Digital twins increase adoption of PLM, CAD, BIM, simulation or automation | Very strong for incumbents |
| Hardware / cloud pull-through | Twin workloads drive GPUs, cloud services, controllers or equipment | Strategically powerful |
| Implementation services | Integration, data preparation, modeling and deployment | Important but difficult to scale |
| Outcome / equipment-as-a-service | Uptime, energy savings, production output or asset availability | Promising, still emerging |
Why is it so hard to say what counts as digital twin revenue?
The digital twin market currently mixes several different businesses under one label, so a large share of what gets called “digital twin revenue” is really software, cloud, services or hardware revenue sold around the twin.
That becomes obvious when we look at the companies considered leaders in the market. Siemens talks extensively about comprehensive digital twins, yet customers usually pay Siemens for Xcelerator software, PLM, simulation, automation or services. Dassault Systèmes builds its virtual-twin strategy around 3DEXPERIENCE but reports revenue through software product lines, subscriptions, licenses and services. Bentley sells infrastructure digital-twin technology through engineering applications, iTwin platform usage and newer asset-analytics products.
AWS and Microsoft create another problem for market sizing. AWS IoT TwinMaker and Azure Digital Twins are directly billable digital-twin services, but a customer using them will often pay separately for storage, IoT ingestion, databases, visualization and compute. How much of that surrounding cloud bill belongs in the “digital twin market” becomes a judgment call.
For this analysis, we therefore follow the money rather than the label. We count the different ways companies get paid because a customer wants to create, connect, simulate or operate a digital representation of a physical asset. That gives us a much clearer view of the actual business models.
This market map, featured in our digital twin market deck, highlights top companies and startups in the digital twin market
What has changed in digital twin business models lately?
Digital twin monetization is moving toward recurring and consumption-based revenue while the underlying modeling layer is becoming cheaper to access.
Recent results from the large industrial software companies show that customers are still spending heavily on the broader digital stack. Siemens said its digital business grew 18% during the first nine months of fiscal 2026. Dassault Systèmes reported 14% growth in both 3DEXPERIENCE software revenue and cloud software revenue in its latest quarter. These are much broader businesses than digital twins alone, but the software environments where industrial twins live are still growing.
At the same time, the platform layer is becoming easier to access. NVIDIA made Omniverse free for development, production and redistribution in May 2026, while enterprise support remains paid. Autodesk Tandem now prices around the amount of assets and operational data inside a twin instead of simply counting users. Bentley is pushing further into consumption pricing, including API usage and charges linked directly to physical assets.
The direction is becoming clearer. Companies can still make good money from digital twins, but simply charging for access to a 3D digital representation looks less compelling these days. Revenue is moving toward the workflows, data, computation and operating decisions that happen around that representation.
If you want more recent data on this point, please see our latest digital twin market report.
Are subscriptions now the main digital twin business model?
Subscription software is currently the clearest proven business model around digital twins.
Bentley gives us one of the cleanest examples. In its latest quarter, the company generated $378.6 million of subscription revenue from $410.7 million of total revenue. Subscriptions therefore represented about 92% of the business, while annualized recurring revenue reached $1.54 billion and grew 12% at constant currencies.
Dassault Systèmes shows the same pattern from a much larger industrial-software base. During the first half of 2026, subscription and support revenue reached €2.30 billion and recurring revenue represented 83% of software revenue.
Siemens says roughly 80% of its Digital Industries software business is now recurring, including subscriptions, SaaS and maintenance. PLM, which is closely tied to Siemens' digital-twin strategy, is still moving toward SaaS.
These businesses are broader than digital twins, so those billions should not be treated as digital-twin revenue. What they show very clearly is how the companies building the largest industrial twin environments prefer to get paid: continuously, for as long as customers keep designing, building and operating their physical assets.
| Company | Current recurring-revenue evidence | What it shows |
|---|---|---|
| Bentley Systems | About 92% of latest quarterly revenue came from subscriptions | Infrastructure software is overwhelmingly recurring |
| Dassault Systèmes | 83% of first-half software revenue was recurring | Virtual-twin workflows sit inside a large recurring software base |
| Siemens Digital Industries Software | About 80% of the business is recurring | PLM and industrial software continue moving toward SaaS |
As this chart shows, and as featured in our digital twin market deck, search interest in digital twins has increased sharply
Why do Siemens and Dassault sell digital twins inside bigger software suites?
Siemens and Dassault can make more money by putting digital twins inside CAD, PLM, simulation and automation workflows than by asking customers to buy a separate twin product.
A useful industrial twin needs much more than a visual model. A manufacturer may need the geometry of a machine, its bill of materials, engineering requirements, production configuration, simulation results, maintenance history and live operating data. Companies such as Siemens and Dassault already control large parts of that information through the software customers use every day.
Dassault's virtual-twin strategy therefore sits inside 3DEXPERIENCE, where customers can design, simulate and manage products across their lifecycle. Siemens goes even further because its software can connect product engineering with factory automation and control systems. The digital representation can follow an asset from design into production and eventually into operation.
That makes the twin useful without requiring a separate buying decision every time. The customer already needs PLM, simulation or engineering software, and better digital-twin capabilities give it another reason to expand usage and renew the broader platform.
This bundled model is particularly hard for a new standalone vendor to attack. The incumbent already owns the customer relationship, the engineering data and the applications where much of the work happens.
If you want more recent data on this point, please see our latest digital twin market report.
How do digital twin vendors charge: per user, per asset or per usage?
Digital twin pricing is moving away from simple seat counting because the cost and value of a twin often grow with assets, data and computation rather than with the number of people looking at it.
Autodesk Tandem is a good example. Autodesk explicitly uses capacity-based pricing. Its paid tier includes capacity for assets and data points, while users and facilities can be unlimited. Additional asset and data-point capacity can be purchased as the twin gets larger.
Bentley's iTwin Platform combines a recurring platform fee with usage credits. Its current Standard tier starts at $199 per month and Premium at $499 per month, with additional credits priced at $1.20. Storage, data movement, processing and other platform activities consume those credits.
Bentley is also experimenting further down the stack. In its 2025 year-end discussion, the company described asset-consumption revenue from infrastructure inspections. For electrical distribution poles, Bentley said each digital-twin inspection and simulation could generate revenue in the low double digits per physical asset. That creates a completely different scaling mechanism: more infrastructure inspected means more revenue, even if the number of software users stays flat.
We therefore see three increasingly common pricing units today: the size of the twin, the amount of platform activity and the number of physical assets being analyzed.
| Example | What the customer pays against | How revenue grows |
|---|---|---|
| Autodesk Tandem | Assets and data capacity | Larger or more data-heavy twins |
| Bentley iTwin Platform | Subscription plus usage credits | More storage, processing and platform activity |
| Bentley Asset Analytics | Physical assets inspected or analyzed | More towers, roads, poles or other assets |
| Traditional enterprise software | Seats or subscriptions | More users and broader deployment |
This chart, included in our digital twin market deck, shows annual VC investment in digital twin startups
How much money can AWS or Azure make from one digital twin?
For AWS and Azure, the digital twin service itself can be a relatively small bill; the bigger opportunity comes from all the cloud infrastructure running around it.
AWS makes this unusually easy to see through its own TwinMaker pricing examples. A small factory containing 800 entities produces a sample TwinMaker bill of $197.53 per month under standard pricing. A 5,000-entity warehouse comes to $649.74 per month. That works out to roughly $2,400 and $7,800 a year respectively.
Those figures exclude several surrounding AWS services. The same examples can require IoT SiteWise to ingest and store industrial time-series data, S3 for 3D files and Managed Grafana for dashboards. Other workloads can bring databases, analytics, AI and additional compute into the architecture.
Microsoft Azure follows a similar consumption model. Azure Digital Twins has no upfront fee; customers are charged for operations, messages and query units. An idle twin produces no Digital Twins bill at all.
AWS and Microsoft do not need every digital twin API to become a six-figure software contract. A twin can pull an industrial customer deeper into a cloud environment where multiple services keep generating consumption revenue for years.
Are digital twin implementation services a good business on their own?
Digital twin implementation services are often necessary, but the economics look much weaker than software when services become the main product.
Bentley's latest quarter gives us a useful real-world comparison. The company generated $22.4 million from services and spent almost $20.0 million delivering them, leaving an approximate gross margin of 10.8%. Its subscriptions and licenses generated $388.3 million against $54.0 million of direct costs, equivalent to roughly an 86% gross margin.
The gap is enormous. Integration work can require engineers to clean data, connect old systems, configure models and adapt workflows for one customer. Software can then be sold repeatedly without recreating all of that work each time.
There is still money to be made in digital-twin consulting, especially for systems integrators and engineering firms whose business already revolves around projects. For a scalable software company, though, services work best when they get the customer deployed and open the door to years of recurring software or consumption revenue.
| Bentley latest quarter | Revenue | Direct cost | Approx. gross margin |
|---|---|---|---|
| Subscriptions and licenses | $388.3M | $54.0M | 86.1% |
| Services | $22.4M | $20.0M | 10.8% |
This chart, included in our digital twin market deck, breaks down Neara's playbook in digital twins
Why did NVIDIA make Omniverse free if digital twins are valuable?
NVIDIA can afford to give Omniverse away because industrial digital twins create demand for paid support and, more importantly, accelerated computing.
Since May 2026, companies can use Omniverse for development, production and redistribution without buying an NVIDIA AI Enterprise subscription. NVIDIA's documentation, updated again recently, makes the split explicit: community support is available for free users, while enterprise support requires NVIDIA AI Enterprise.
The paid enterprise layer can still be meaningful. NVIDIA currently lists a one-year self-managed AI Enterprise subscription at $4,500 per GPU, with larger multi-year and support arrangements available.
The bigger economic opportunity sits around the workloads themselves. Physically accurate simulations, synthetic-data generation, robotics development and large industrial environments can consume substantial GPU resources. NVIDIA benefits when more developers build these applications on an ecosystem designed around accelerated computing.
A GPU used by a factory simulation cannot automatically be counted as “digital twin revenue” in a market-size chart. But from NVIDIA's point of view, giving away the software foundation can make perfect commercial sense if it expands the number and size of GPU-intensive physical-AI workloads.
If you want more recent data on this point, please see our latest digital twin market report.
Is industrial connectivity its own digital twin business?
Industrial connectivity is a real digital twin revenue layer, although recent moves show that companies can separate it from the engineering software that sits above it.
PTC provides the clearest example. The company completed the sale of its Kepware industrial-connectivity and ThingWorx IoT businesses to TPG in 2026, receiving $523 million in cash at closing. PTC said the divestiture would let it focus more heavily on its Intelligent Product Lifecycle strategy.
That is a useful transaction because Kepware and ThingWorx sit close to the plumbing required for connected industrial twins. Their sale for more than half a billion dollars shows that connectivity has standalone value, while PTC's decision to divest also shows that owning every layer of the stack is not essential.
Autodesk has chosen a more integrated structure but still separates the products commercially. Tandem manages the digital twin itself, while Tandem Connect handles external connections through plugins, message brokers and other integrations. Tandem Connect has its own capacity limits and paid offering.
Connectivity can therefore support a standalone business, a separate module or a feature inside a larger platform. The winning structure depends on whether the vendor has enough scale and differentiation at that layer to justify owning it.
This chart, included in our digital twin market deck, shows annual funding in digital twin startups
Do vertical digital twins make more money than generic platforms?
Vertical digital twins look easier to monetize today because customers can connect the bill to a concrete operating problem.
Bentley is actively moving in this direction. After acquiring Talon Aerolytics and Pointivo technology, the company said its Asset Analytics business had reached an approximately $50 million revenue run rate. The offering focuses on real infrastructure jobs such as inspecting telecom towers, roads and electrical distribution poles. Bentley also said Talon had already demonstrated profitable growth and a pipeline of enterprise-wide contracts.
The model becomes even clearer in buildings. Trane Technologies' BrainBox AI technology automatically adjusts HVAC systems according to predicted building conditions. In a deployment covering more than 600 Dollar Tree stores, Trane reported annual savings of almost 8 million kWh and more than $1 million in energy costs.
Those examples are easier to sell than a generic promise to “build a digital twin.” A telecom operator can pay to inspect towers faster. A utility can pay to understand the condition of poles. A retailer can pay to cut energy bills.
The customer rarely cares which percentage of the value came from the 3D model, the AI layer, sensor data or simulation. What it can understand immediately is whether a recurring operational problem became cheaper or easier to manage.
If you want more recent data on this point, please see our latest digital twin market report.
Can digital twin vendors charge for results instead of software?
Outcome-based digital twin pricing is credible now, but we still do not see enough evidence to call it a mainstream business model.
The idea is gaining attention inside industrial software. Siemens argued in a 2026 industrial-machinery analysis that connected digital twins and digital threads can give machine builders enough operational visibility to charge for uptime or units produced. That turns a machine sale into an ongoing equipment-as-a-service relationship.
The economics are attractive in theory. A manufacturer that helps a customer avoid millions of dollars of downtime could potentially capture more value from a performance contract than from another software seat. Energy optimization creates a similar opportunity because the savings can be measured directly against utility bills.
Recent deployments show that these outcomes can now be large enough to matter. The BrainBox AI deployment across more than 600 Dollar Tree locations generated more than $1 million in reported annual energy savings. Yet the public evidence does not show Trane simply taking a percentage of those savings as its standard pricing model.
Digital twins clearly make outcome contracts easier to measure and manage. The market has much stronger evidence today for subscriptions, capacity pricing and consumption fees than for widespread revenue-sharing or pay-for-performance contracts.
This chart, included in our digital twin market deck, compares the main business model options for digital twin enterprise software platforms
Are perpetual licenses disappearing from digital twin software?
Perpetual licenses are fading fast around digital twin software, although large industrial vendors have not eliminated them completely.
As seen above, Bentley's latest quarter was already overwhelmingly subscription-based. Perpetual licenses contributed just $9.7 million, equivalent to about 2.4% of total quarterly revenue.
Dassault Systèmes remains somewhat more mixed, but recurring revenue still represented 83% of first-half software sales. Siemens says about 80% of Digital Industries Software is recurring and continues moving parts of PLM toward SaaS.
The direction fits the product. A useful operational twin keeps changing as the physical asset changes, sensors produce new information and engineers update models. Vendors increasingly want the commercial relationship to continue for the same period.
Perpetual licensing will probably survive in customers and industries that need on-premise deployments or slower software transitions. It is becoming a secondary model rather than the center of digital-twin economics.
Why is a generic digital twin startup so hard to build today?
A horizontal digital twin startup has a difficult position today unless it owns a valuable workflow, dataset or distribution channel.
Competition now comes from several directions at once. Siemens and Dassault can add twin capabilities to engineering platforms that customers already use. Autodesk can build Tandem around BIM and construction data. AWS and Azure can offer the underlying twin infrastructure as another cloud service.
The base technology is also getting cheaper. As discussed earlier, NVIDIA now allows companies to build and deploy commercial Omniverse applications without paying for the core software. Startups can benefit from that, but so can every competitor.
Meanwhile, PTC's decision to sell ThingWorx and Kepware while concentrating on its product-lifecycle strategy shows that even a major industrial software vendor can choose depth in a workflow over owning a broad horizontal IoT platform.
The better startup position is increasingly specific: inspecting power infrastructure, optimizing building energy use, monitoring a particular type of industrial equipment or automating a costly engineering process. A generic platform can still work, especially with strong distribution or proprietary technology, but “we create digital twins for anything” has become a much harder pitch.
If you want more recent data on this point, please see our latest digital twin market report.
This chart, featured in our digital twin market deck, illustrates how revenue is divided among customer segments in the digital twin market
So what are the main business models in the digital twin market?
Today, the digital twin market is mainly a recurring software and consumption business, with bundling, services, hardware pull-through and outcome-based contracts creating additional ways to make money.
The strongest proven model is recurring vertical software. Siemens, Dassault and Bentley show how valuable digital-twin capabilities become when they live inside software that customers use continuously for engineering, manufacturing or infrastructure operations.
Consumption is becoming the second major model. Autodesk scales Tandem pricing with the size and data intensity of the twin. AWS and Azure meter cloud activity. Bentley is going further by charging against API usage and physical assets. These models let revenue grow even when user counts barely change.
Bundling can be just as powerful for large incumbents. A digital twin can help sell or retain PLM, simulation, BIM, automation, cloud infrastructure and eventually hardware. NVIDIA's recent Omniverse licensing change shows how far that logic can go: the platform itself can even become free when the company has stronger monetization elsewhere in the stack.
Implementation services remain an important part of deployment, particularly in messy industrial environments, but their margins and scalability make them less attractive as the core business for a software company. Outcome pricing could eventually become much bigger because it ties fees to uptime, energy savings or production output, although the commercial evidence is still early.
The best digital twin businesses today own something beyond the twin itself: a recurring workflow, a valuable physical asset relationship, a stream of cloud consumption, a hardware ecosystem or a measurable operating outcome. The generic digital representation is gradually becoming infrastructure. Most of the durable money is being made above it, around it or because of what it allows the customer to do.
| Business model | How companies get paid | Current strength |
|---|---|---|
| Recurring software | SaaS, subscriptions, support and renewals | Strongest proven model |
| Capacity / consumption | Assets, data, API calls, compute or platform usage | Growing quickly |
| Bundled software | Digital twins increase adoption of PLM, CAD, BIM, simulation or automation | Very strong for incumbents |
| Hardware / cloud pull-through | Twin workloads drive GPUs, cloud services, controllers or equipment | Strategically powerful |
| Implementation services | Integration, data preparation, modeling and deployment | Important but difficult to scale |
| Outcome / equipment-as-a-service | Uptime, energy savings, production output or asset availability | Promising, still emerging |
OUR METHODOLOGY
This analysis looks at how companies actually make money from digital twins rather than trying to force the market into a single product category. We separated the main economic mechanisms around digital twins — recurring software, capacity and consumption pricing, bundled industrial platforms, cloud and hardware pull-through, implementation services, connectivity and outcome-based models — and looked for recent commercial evidence for each one.
We prioritized observable business-model evidence rather than broad market-size estimates. That included company financial disclosures, recurring-revenue mixes, current pricing structures, licensing changes, acquisitions and divestitures, usage-based models, customer deployments and measurable operating outcomes. Company-level figures are used to show how a model works, not as a proxy for total digital twin market revenue.
For financial economics, the main sources include Bentley Systems investor disclosures, Dassault Systèmes investor relations, and Siemens' disclosures around its Digital Industries and software businesses. These sources support the analysis of recurring revenue, subscriptions, services economics and the broader industrial-software environments in which digital twins are sold.
For pricing and monetization mechanics, key sources include AWS IoT TwinMaker pricing, Azure Digital Twins pricing, Bentley iTwin Platform pricing, and Autodesk Tandem. These sources show the shift from traditional seat pricing toward assets, data capacity, API activity, processing and broader cloud consumption.
We also used official product and strategy documentation to understand how digital twins are bundled into larger ecosystems. Important sources include Siemens Digital Twin, Siemens Xcelerator, Dassault Systèmes 3DEXPERIENCE, Bentley iTwin Platform, NVIDIA Omniverse, and NVIDIA AI Enterprise.
For industrial connectivity and vertical operating models, we used PTC disclosures and product material around ThingWorx and Kepware, together with Bentley's Asset Analytics disclosures and Trane Technologies' material on BrainBox AI. Those examples help separate the value of the underlying twin from the revenue generated by connectivity, infrastructure inspection, building optimization and other specific operating workflows.
Our final characterization of each model as established, growing or emerging reflects the consistency and commercial maturity of the evidence across these sources. The core question throughout was simple: when a customer wants to create, connect, simulate or operate a digital representation of a physical asset, what exactly does the vendor charge for, and what makes that revenue grow?
This chart, included in our digital twin market deck, shows how industrial digital twin platform technology has evolved over time
Related blog posts
