This article is written in my own name and does not represent the official position of the company I work for. It is based on publicly available presentations and publications.
How do you envision the future?
Imagine a surgeon saving a child whose heart should never have been beating. Exoskeletons that restore the ability to walk to those who have lost it due to disability. Drones delivering a vital vaccine to the most remote village on the planet.
Imagine an end to traffic fatalities. Habitats that produce their own air and food to prepare for the conquest of Mars. Trains capable of competing with airplanes to build a more sustainable world.
And speaking of energy: imagine a revolution capable of decarbonizing entire cities using nothing but tidal power.
I’m sorry to have to tell you this: you’re living in the past.
The heart? Recreated since 2014, beat by beat, by the Living Heart Project. Walking? Since 2021, nearly 100 people have taken more than a million steps thanks to SPARK, Biomotum’s exoskeleton. The vaccine? In 2018, a Swoop Aero drone delivered it into the hands of a nurse in Vanuatu for the very first child in the world to be vaccinated this way.
The road? Since 2005, BMW has crashed its cars a thousand times—without denting a single panel—in the Abaqus simulator. Martian habitat? In 2021, Interstellar Lab sealed BioPod, a bubble that produces its own air and food. The train? Since 2019, Alstom has been designing Trenitalia trains in a virtual world before driving the first bolt in the real one. And the tides? On February 9, 2024, at dawn, Minesto’s Dragon 12 fed the energy of the ebb and flow into the Faroe Islands’ grid.
All of this is real. Delivered. And already so commonplace that you’ve forgotten to marvel at it.
And they all share the same secret…
None of this was achieved through a series of failed trials in the real world, where mistakes come at the cost of lives, fortunes, and lost years. All of this first came to life within a virtual twin: a digital replica so faithful to reality that the heart was operated on a thousand times, the car crashed a thousand times over, the train assembled, the building sealed and tested, the tide tamed—until only success remained. And only then was all of this built in reality.
All these virtual twins bear the same signature. One company has been making them possible for over forty years. It is called Dassault Systèmes. Twelve industries, more than 370,000 customers, 140 countries. Dassault Systèmes gives you not only the tools to bring a product to life, but also the means to manage its entire lifecycle—and, before a single brick is laid, to build its factory in the virtual world and watch the entire production line run. With these solutions, you can place a technician’s hands on an aircraft, replicate the exact steps of a repair, and prove that the mechanic will be able to do the job before a single bolt is tightened. From the first sketch of an object to its final hour of service, Dassault Systèmes has quietly become one of the few platforms capable of supporting a product throughout its entire lifecycle. And one of the foundations upon which the world’s most demanding industries rest: aerospace, automotive, and life sciences.
So here’s the only question that matters: if this were just the past, what would the future look like?
The Cost of the Trial
Behind every major technological leap lies the same relentless arithmetic: to find the one answer that works, you have to rule out the dozens that fail. It’s impossible to avoid making mistakes on the path to success. The only real choice is where you’re allowed to make those mistakes.
Make a mistake in reality, and the price is paid in lives: nine out of ten new drugs fail in human trials, and the cruelest cost is neither the years nor the billions, but the patient who dies while waiting—failure after failure—for the cure that is still being missed in a laboratory.
But reality presents a second bill, one paid in the planet. And here, the real waste isn’t the prototype we sacrifice: it’s everything we mass-produce afterward without having optimized it. Because 80% of a product’s environmental footprint is determined right from the sketch—in the choice of a material, the design of a part, the layout of a supply chain. A poorly shaped wing means tons of fuel burned on every flight, over the entire lifespan of a fleet. A part that’s too heavy means raw materials extracted millions of times over. We’re already consuming the Earth as if we had 1.7 of them; every ton of material, every superfluous gram, every ill-conceived logistics kilometer is taken from a world already running on empty.
This is where it all comes down to. In the digital twin, we don’t just avoid mistakes: we streamline before we produce, we refine the engine before we build it, we optimize the supply chain before we stock it. We don’t correct waste—we prevent it from happening in the first place.
The cruelest cost is neither the years nor the billions, but the patient who dies while waiting
Move all of this into the virtual world, and the math flips. The rocket can explode as many times as it wants. The bridge can collapse a hundred times without killing anyone. The cure can fail a million times in a single night without a single real body being harmed. Dassault Systèmes has built the only place where humanity has the right to make mistakes without paying for them in lives and planets—so that reality never has to bear the burden of what doesn’t work.
That’s how they’ve always managed to stay ahead of the curve. And that’s about to matter more than ever.
The Era of Industry World Models
This is precisely why the next chapter belongs to artificial intelligence, and why Dassault Systèmes is better positioned than anyone else to write it. Because AI is changing the very scale of what it means to predict the future. But not on its own. Its power does not come from within: it comes from a structure that no one else can build. This structure has a name: the 3DEXPERIENCE. And you can explore it from its very foundations.
Underground, first, what we never see and what makes everything possible: a sovereign infrastructure. OUTSCALE’s AI factories, built with NVIDIA, where each customer’s knowledge remains their own—protected, exclusive, never mixed with anyone else’s. Sovereignty isn’t a line at the bottom of a contract; it’s the condition that allows a company to dare to entrust its memory.
The building itself rises from these foundations. The 3DEXPERIENCE is not just another floor: it is the framework, the walls, and the hallways. It is what fuels every level—the tools that AI can activate, the expertise it can mobilize, the context it provides for models to work within, the connectors that link it to the rest of the world. No matter which floor you’re on, you’re on the platform.
On the first level lies the treasure: decades of capitalized reality. Decades during which Dassault Systèmes has tested its models against aircraft in flight, factories in production, and the most complex systems the industry has ever designed—and has codified what manufacturers know without always being able to articulate it: their rules, their standards, their know-how. This knowledge is not a database; it is an industrial memory.
And it is from this memory that the next level emerges: the virtual twin. Capitalized knowledge becomes an executable world—an environment where every idea is tested not against statistics, but against the inviolable laws of physics. It is the test bed of the structure. Imagination proposes; simulation renders its verdict. What is probable is no longer enough; only what works survives.
And from these twins running endlessly arise the Industry World Models. Every simulated aircraft, every modeled factory, every reconstructed organ feeds them; supplemented by thousands of models—general-purpose or specialized—and by NVIDIA’s open models, they distill the dynamics of the physical world: matter, behaviors, consequences. These are not models that have merely read the world; they have experienced it. Validated by science, they give AI what it was missing: an anchor in reality. And in return, they make the twin truly generative—no longer a replica that describes the world, but an engine that proposes new ones.
Above them works a population that is never introduced and never sleeps: the agents. Autonomous and goal-oriented, they are the executors of the edifice. Those who break down, launch, verify, and start over. Every hypothesis they put forward descends one level to be judged; every verdict rises to fuel the next action.
Then come the virtual companions: Aura, Leo, and Marie. Three embodied experts who personify decades of scientific knowledge, industrial experience, and technical know-how. Aura keeps a sprawling project coherent and aligned. Leo, named after Leonardo da Vinci, handles the engineering, from the first sketch to the workshop. Marie, for Marie Curie, delves into the very substance of things—materials, chemistry, medicine—to ask the questions no one has yet thought to ask.
Where an engineer formulates one hypothesis, they devise ten thousand: they explore, prioritize, eliminate, recombine, and entrust agents, all night long, with a testing campaign that no human team could even plan. Their role is not to be right; it is to imagine without limits, and to know what to test next.
And because they rely on all the lower levels, this imagination never runs wild: every “hallucination” is put to the test in simulation before reaching the real world. For a hallucination in an operating room means a child who doesn’t wake up; in an airplane, it means a crash. When the outcome involves a human life, “probably” is not a word we have the right to use. Here, it is never used alone: it is always put to the test.
That leaves the roof—the only level we see. CATIA, SOLIDWORKS, the applications that millions of engineers open every morning. This is where it all begins and where it all comes back: an intention flows down the structure, becomes delegation, then hypotheses; reality’s verdict flows back up, becomes execution, then response. The user, for their part, has seen only a question asked—and a trustworthy answer.
In 2026, with NVIDIA, Dassault Systèmes set the standard for an entire field and raised the bar for the competition. You can train an entire online model in a few months—but you can’t download forty years of know-how, that verified reality that makes this AI trustworthy and that can only be acquired one step at a time. That’s why, where AI must be both infinitely powerful and certain, no one on Earth has a greater head start.
Every simulated aircraft, every modeled factory, every reconstructed component enriches the Industry World Models: real-world models where AI no longer merely predicts—it experiments, fails, learns, and tries again. And the more these worlds expand, the faster each new question finds its answer. It is an acceleration that accelerates itself.
All of this lives in a single place: the 3DEXPERIENCE platform, where thousands of these companions and the people they serve work as a single, choreographed whole. The right expertise arrives for the right person at the right time; a master engineer at every engineer’s side, a great scientist alongside every scientist, the accumulated wisdom of an entire industry placed in the hands of those who need it. A mastery that once belonged only to a select few, now given to all who are building the future—and kept, always, under their own control.
a great scientist alongside every scientist
The continuation of a proven track record
Here’s what that makes possible, in concrete terms.
In healthcare. Today, it takes ten years for a drug to become available, and nearly all candidates fail—in real patients, in real trials. Tomorrow, an AI operating within the virtual twin of a living heart will eliminate ten thousand molecules overnight, retaining only three, and clinical trials will begin where, yesterday, they would have failed. The surgeon, meanwhile, will have already operated on your heart—yours, modeled down to the fiber—before making his first incision.
In energy. In the south of France, an artificial sun is being assembled: a fusion reactor in which every part, every tolerance, every plasma scenario is tested virtually before it even exists. Tomorrow, with Industry World Models, AI will explore reactor configurations that decades of human calculation would never have touched—and the promise of nearly limitless clean energy will cease to be a gamble and become a matter of timing.
In industry. An entire factory designed, tested, and optimized before a single wall is built: this is already the present. The future is an AI that, in this virtual world, reinvents the production line every night, anticipates breakdowns before they occur, and tracks down waste before it exists—the lightening of a part, the right material, the waste-free process. Products that are right the first time, in a world that can no longer afford to produce just to throw away.
And those who learn. There are professions where one cannot practice: you don’t rehearse open-heart surgery on a patient, you don’t run through your scales on a reactor core, you don’t learn the right move where a mistake costs a life. For centuries, the transmission of knowledge in these fields has followed a cruel rule—see much, do little, and do it late. Tomorrow, that rule will fall. The surgeon will have rehearsed the entire procedure on the virtual twin of the heart he is about to open—your heart, with your specific condition—before the first incision. The reactor operator will have run his facility a thousand times, triggered every alarm, weathered every incident, without any of them ever being real. The technician will have laid his hands on the plane, rehearsed every step of the repair, and proven he knows how to do it—before a single bolt is turned. Theory no longer precedes practice: it merges with it. And at the side of every learner now stands a companion who never sleeps, who knows the rules of the trade and reminds them at the right moment—a master whom one no longer has to wait years to earn. What used to take a career to acquire is now acquired in a season; and experience, once rare and dearly paid for, becomes something that can be given to everyone, right away, without risk.
Beyond. A lunar habitat where humans will have lived a thousand days virtually before the first one spends a single night there. And Mars—no longer a dream, but an engineering problem: solved where failure costs nothing, to be attempted only where it would cost everything. What was yesterday the backdrop of our fiction becomes a series of plans validated, one by one.
And underlying it all, the most discreet of promises: machines absorb the trial and error—slow, costly, sometimes deadly—and give back to humans the only raw material of which the future is truly made: the freedom to imagine. None of this is science fiction. It is the story that Dassault Systèmes has been writing since 1981. Industry World Models are simply what now makes it possible to write that story at the speed of the world that needs it.
And our factories, tomorrow. Imagine a country determined to rebuild what it had let slip away. In a single generation, its industrial base had shrunk by nearly half; its neighbors retained twice as much. It’s not just a few small workshops that need to be revived, but hundreds of factories within a decade. Yesterday, each one was a gamble worth hundreds of millions, played just once, with no room for error: a poorly chosen site, a poorly designed production line, and years of work and jobs were lost. Tomorrow, not a single factory will break ground without having already lived a thousand lives in the virtual world. We will build it there in its entirety, run the entire production line through it, and specialized engineers will explore ten thousand layouts in a single night to keep only the one that produces just right—on the first try. We will no longer relocate blindly: we will aim precisely. And since this knowledge will remain sovereign—kept in-house and never relinquished—it will be a reconquest that does not trade one dependency for another. The old dream of a nation rebuilding its factories then ceases to be a mere incantation: it becomes a plan, already validated before it is built.
Dassault Systèmes has lived by the same ambition from the start: show that a dream is possible, and you give people the desire to create it. For forty years, this company has been turning the future you dream of into a past you take for granted. It is about to do it again. And this time, the future is vaster than any of us has yet dared to imagine.
They don’t predict the future. They prove it… then they build it with you.