Digital Twin: from NASA's pairing technology in Apollo 13 to Grieves's concept (2002) and the industrial reality of 2026

A Digital Twin is a dynamic digital representation of a physical object, process, or real-world system, connected bidirectionally to its real counterpart through data flows. The technical definition matters because it distinguishes the digital twin from simpler digital models: a 3D model is not a digital twin; a simulation isn't one on its own either; what makes a system a digital twin is the real-time (or near-real-time) flow of data between the physical object and its digital replica, making it possible to monitor, simulate, and predict the behavior of the original.

The concept sounds modern, but it has half a century of conceptual history and two decades of academic formalization. For a brand or company that hears the term without context, knowing the lineage and the current state helps distinguish between real applications with demonstrated value and tech marketing with no substance.

The conceptual origin: NASA and Apollo 13, 1970

Although the term "Digital Twin" as such was coined decades later, the concept has a notable precedent at NASA during the Apollo era. For the spaceflight missions of the '60s and '70s, NASA maintained physical replicas of the spacecraft at mission control in Houston. When something went wrong in orbit, engineers could reproduce the problem on the ground-based replica, test solutions, and communicate them to the astronauts.

The most famous case is Apollo 13 in April 1970. When the oxygen tank explosion jeopardized the mission, Houston engineers used simulators and physical replicas to design the procedures that ultimately brought the astronauts home. This practice of "having a mirror system to test and diagnose on" is frequently cited as a conceptual precursor to modern digital twins, although NASA used the expression "pairing technology" at the time.

As digitalization advanced, the physical simulators were progressively replaced by digital simulations. But the essential idea —maintaining a model of the real system that can be manipulated without risk, synchronized with the original— was preserved.

The academic formalization: Michael Grieves, 2002-2010

Formal credit for the modern conceptualization of the Digital Twin goes to Michael Grieves, then a professor at the University of Michigan, who presented the concept at a Product Lifecycle Management (PLM) conference in 2002. The idea was first published in one of his books on PLM in 2003, still without its current name.

The term "Digital Twin" itself formally appears in 2010 when Grieves published a white paper that explicitly named the concept. The paper "Digital Twin: Manufacturing Excellence through Virtual Factory Replication" established the definition that has held: a digital twin has three essential components — the physical object, its virtual counterpart, and the data flow that connects them. In 2016 Grieves published "Origins of the Digital Twin Concept," recapping the history of the term he himself had coined.

NASA and John Vickers (then at NASA, later in one of the pioneering groups on digital twins) helped popularize the term in the industrial context between 2010-2012, particularly in their 2012 white paper on technologies for the NASA of the future. By the mid-2010s, "Digital Twin" had entered common industrial vocabulary.

The integration with Industry 4.0: Hannover Messe 2011

In parallel with the formalization of the Digital Twin, another movement emerged that would integrate the concept into a broader vision of industrial digitalization: Industry 4.0 (Industrie 4.0 in the original German).

The term was publicly introduced at Hannover Messe 2011 —the world's largest industrial trade fair— by a working group led by Henning Kagermann and backed by the German government as an industrial strategy. The idea: the fourth industrial revolution (after mechanization, electricity and mass production, and automation with electronics) would be the integration of the physical with the digital through technologies like the Internet of Things, big data, artificial intelligence, and digital twins.

Industry 4.0 became a reference framework for industrial investment in Germany, Europe, and eventually globally. China launched "Made in China 2025" in 2015 with parallel concepts. The United States, Japan, and other countries developed similar initiatives.

Within Industry 4.0, the Digital Twin positioned itself as a key component: the way data from the physical world is translated into digital representations that can be analyzed, simulated, and optimized.

The three essential components

To understand what really constitutes a digital twin (vs. things that are labeled as such without being one), it helps to return to Grieves's definition:

1. The real physical asset. The physical object, equipment, building, process, or system you want to replicate. It can be a wind turbine, an aircraft, a building, a production line, an entire city.

2. The virtual twin. The digital model of the asset. It can include 3D geometry, physical parameters, simulated behavior, and historical data. Fidelity varies according to use — from simplified models to exhaustive replicas.

3. The data connection. The bidirectional information flows between the physical asset and its virtual twin. Sensors on the asset send data to the model (telemetry, status, performance); the model can send commands to the asset (configuration changes, alerts).

It's the real-time (or near-real-time) data connection that distinguishes the digital twin from a mere 3D model or simulation. Without that synchronization, you have a digital model, but not a digital twin in the strict sense.

Types of Digital Twin: the fidelity hierarchy

The industry distinguishes between levels of digital twin according to fidelity and application:

Digital Model. A digital model only, with no connection to the real asset. Useful for design and planning, but not properly a digital twin because it lacks synchronization.

Digital Shadow. A model connected to the real asset with a one-way data flow (from the asset to the model). The model is updated with real data but does not control the asset. A common approach in passive monitoring.

Digital Twin. A model with a bidirectional connection. The model receives data from the asset and can send commands or configuration changes. This is the strictest technical definition.

Additionally, Grieves distinguishes by scale:

Digital Twin Prototype (DTP). Before manufacturing the physical product, models of the digital prototype with all its specifications.

Digital Twin Instance (DTI). Each individually manufactured product has its own specific digital twin, which tracks its individual behavior throughout its life.

Digital Twin Aggregate (DTA). The aggregation of all DTIs for fleet-level or population-level analysis. It makes it possible to identify global patterns (all type X engines have problem Y at 5,000 hours).

This granularity matters because the applications vary according to the level.

Real use cases of Digital Twin in 2026

Energy: wind turbines and power plants. Siemens Energy and GE Vernova maintain individual digital twins of every wind turbine they sell. Sensors monitor vibration, temperature, output, and weather conditions. The digital twin compares real behavior with expected behavior, predicts failures before they occur (predictive maintenance), and makes it possible to optimize parameters to maximize production. The savings are significant — a documented 20-30% reduction in unplanned maintenance in fleets with a mature digital twin.

Aviation: engines and aircraft. Rolls-Royce maintains digital twins of every engine it sells, with sensors that send data during flight. The company has shifted from selling engines to selling "flying hours" as a service (the "Power-by-the-Hour" model developed since the '60s), where the digital twin is central to managing maintenance efficiently. Pratt & Whitney and GE Aerospace operate similar models.

Manufacturing. Companies like Siemens, Bosch, and Schneider Electric offer digital twin platforms for entire factories. Mercedes-Benz uses digital twins of its production lines to simulate changes before implementing them. BMW has spoken openly about its transformation with NVIDIA Omniverse to create digital twins of its plants.

Construction: BIM and digital twins of buildings. Building Information Modeling (BIM), in use since the '90s, has evolved toward digital twins of buildings. Companies like Autodesk (with Construction Cloud), Bentley Systems (with iTwin), and Trimble make it possible to maintain a synchronized digital model during the construction and life of the building. The revised European EPBD (Energy Performance of Buildings Directive) of 2024 explicitly mentions a "Digital Building Logbook" — conceptually close to the digital twin.

Smart cities. Singapore has developed Virtual Singapore, a digital twin of the entire city with detailed 3D modeling and real-time data. Other cities (Helsinki, Boston, Shanghai) have similar initiatives. The usefulness: simulating the impact of urban planning changes, optimizing traffic, planning emergency response, modeling climate effects.

Healthcare. Digital twins of individual organs (heart, lungs) are used in medical research and, increasingly, in personalized surgical planning. Dassault Systèmes has developed the "Living Heart Project." The concept is being extended experimentally to "patient digital twins" for personalized medicine — still in early research in 2026.

Retail and logistics. Amazon operates digital twins of its fulfillment centers. Walmart has developed digital twins of stores to optimize layout and stock. Maersk and other shipping companies operate digital twins of their fleets.

Digital twin platforms.

• Microsoft Azure Digital Twins (launched 2018, GA 2020).
• AWS IoT TwinMaker (launched 2021).
• Siemens Industrial Edge / Mendix.
• NVIDIA Omniverse (launched 2020) for cases with requirements for realistic 3D visualization.
• Bentley iTwin for infrastructure.
• Ansys Twin Builder for advanced engineering.
• PTC ThingWorx (especially with industrial IoT).
• Dassault Systèmes 3DEXPERIENCE.

The prices of serious implementations are significant — typically hundreds of thousands to millions of dollars for medium-sized projects.

What a Digital Twin is NOT (despite the marketing)

There is considerable inflation of the term in marketing and consulting. It's worth distinguishing:

Static 3D models. Traditional BIM with no data connection. Useful for design, not a digital twin.

Interactive visualizations. Virtual tours, product configurators. Different, valuable applications, but not a digital twin.

Dashboards with data. Showing metrics in real time is monitoring, not a digital twin (it lacks the underlying model that makes it possible to simulate and predict).

Isolated simulations. Simulation software that doesn't connect to real data from the asset is simulation, not a digital twin.

Generic "Digital Transformation." Any digital project gets labeled "digital twin" if it sounds modern. Grieves's technical definition is narrower.

The industry has started to talk about a "Digital Twin Maturity Model" to distinguish between levels. Gartner and other analyst firms publish rankings and descriptions of what constitutes a true digital twin vs. a partial approximation.

When the Digital Twin makes sense (and when it doesn't)

It makes sense when:

• The physical asset is expensive or critical. Turbines, aircraft, power plants, industrial buildings — the cost of failure justifies the investment in a digital twin.
• There are many sensors and lots of data available. A digital twin without data is a model, not a twin.
• The asset's behavior is complex. Systems with nontrivial dynamics benefit from a simulation that approximates reality.
• The organization is digitally mature enough to manage the system. A digital twin without a team capable of operating it is an underused asset.
• There is a clear business case. Reduction of unplanned maintenance, operational optimization, simulation of changes — quantifiable benefits.

It doesn't make sense when:

• The asset is cheap and low-risk. A digital twin of a mobile phone is overkill.
• There is no data infrastructure. Without sensors, without synchronization, you can't maintain a true digital twin.
• The organization just wants to "look digital" without a real operational need. It leads to wasted investment in systems nobody uses.
• The ROI can't be demonstrated. If after 2 years of implementation there's no clear saving or benefit, it probably wasn't the right case.

Common mistakes in Digital Twin projects

Confusing a 3D model with a digital twin. Buying modeling software and calling it a digital twin. Without a bidirectional connection to real data, it's not a twin.

Underestimating the cost of maintenance. A digital twin requires continuous updating of the model as the physical asset changes (maintenance, upgrades, modifications). Without that updating, the twin ages and loses fidelity.

Not defining the specific use case. "We're going to build a digital twin of the factory" without a specific objective (reduce downtime, optimize energy consumption, simulate layout changes) leads to a project with no clear success metric.

Buying a platform without a strategy. Digital twin platforms are powerful and expensive. Without a clear strategy, they remain a costly demo.

Underestimating integration complexity. Connecting sensors, legacy systems, ERP, MES, and databases into a coherent architecture is a significant engineering project.

Lack of internal talent. Serious digital twins require teams with knowledge of the asset's physics, data engineering, simulation, and programming. Without that talent, total dependence on expensive external consulting.

Not considering cybersecurity. Connecting physical systems to IT networks creates an attack surface. Documented cases of attacks on industrial infrastructure (e.g., Stuxnet 2010) show the risk.

Ignoring the organizational learning curve. A digital twin changes processes. People who used to make decisions by intuition now have to make decisions based on what the twin shows. That cultural transition is a project in itself.

Digital Twin and creative operations: a less obvious connection

For a traditional brand or agency, the digital twin may sound far from everyday creative work. But there are areas where the concept is starting to be applied:

Realistic 3D configurators. Brands like BMW, Audi, and IKEA let the consumer see the product in its context before buying. When the model is connected to real inventory (yes, it's available), customization data (yes, this combination is viable), and subsequent telemetry (yes, the customer actually bought what was configured), it approaches the logic of the digital twin.

Synchronized virtual spaces. In e-commerce, retail, and events. A virtual tour of a real store with synchronized stock, a hybrid event with a bidirectional feed between in-person and digital. For immersive experiences built on these models, see virtual reality (VR).

Audiovisual production. The concept of virtual production (popularized by The Mandalorian since 2019 with LED walls) uses digital twins of sets to produce content. Brands with frequent production can adopt similar tools.

Industrial design and prototyping. Brands with a complex physical product can use digital twins to iterate on the design before physical prototyping.

For most brands, the digital twin is not an immediate priority. But knowing the concept helps identify opportunities when the case justifies it, and helps you avoid being misled by vendors who offer a "digital twin" when what they deliver is a 3D model with good sales presentation.

The connection with creative operations

Where digital twin really is relevant for creative operations is in the management of brand assets when there's a physical product: 3D models that reflect the real product with all its variants and configurations, synchronized with inventory and catalog data. Large brands with extensive physical catalogs are building "digital product twins" to use in marketing, e-commerce, virtual retail, and so on.

That coordination, when it happens, fits within the broader discipline of creative operations: content production generates content from the digital twins, brand management ensures coherence between the physical product and the digital representations, and the digital assets live in the corresponding library.

At Polimake, the logic of digital asset management applies to any digital representation of a product: Studio coordinates campaigns that use these assets, Studio produces derivative material, and Media stores both the raw assets and the derivative versions for different channels.

---

If you lead technology, product, marketing, or strategy and you arrived here looking for an answer about Digital Twin, the most useful thing you can take from this article is probably the combination of three ideas: the digital twin has a precise technical definition (a bidirectional connection to real data, not just any digital model), its real value is in industries with expensive, critical assets where the ROI of optimization justifies the significant investment, and much of the marketing around the term in 2026 inflates what is really simpler. Distinguishing between a true digital twin and digital models labeled as such protects against misguided investment.

To round this out, digital assets covers the management of digital assets in general, render covers the underlying visual representation, and the cloud covers the infrastructure on which digital twins operate.

Quick references

• The cloud — the underlying technical infrastructure.
• Digital assets — management of digital assets in general.
• SaaS — a related software model.
• Render / rendering — the visual representation underlying digital twins.
• 3D and modeling — the technical foundation of digital twins with a visual dimension.