← All sectors / The AI transformation
121 · Simulation & synthetic training data
Teaching machines without the world
Curve position
Binding constraint
Whether behaviour learned in simulation survives contact with reality.
Why this sector sits at Launch Pad
This is real and early. Commercial deployments exist but are small, and we cannot point to a customer at scale with a figure attached.
We place it at Launch Pad because the binding constraint, whether cost, accuracy, or a missing market mechanism, has not eased in a way we can date. The technology improving is not the same as the constraint breaking.
It moves to Liftoff when a specific event changes the economics, and we would expect to be able to name that event and when it happened.
Reviewed on a two month cycle. The position moves only when a dated, verifiable change in the binding constraint justifies it.
Heatmap / How this sector's stocks are moving
| Stock | Price | 1 Day | 1 Week | 1 Month | Year to date |
|---|---|---|---|---|---|
| $PTC | $195.18 | +0.58% | +0.11% | +1.13% | +12.04% |
| $U | $42.46 | +4.89% | +7.49% | −5.54% | −3.87% |
| $NVDA | $200.13 | −3.17% | −7.08% | −16.35% | +7.31% |
| $SNPS | $462.95 | +0.02% | −3.93% | −8.36% | −1.44% |
| Sector average | +0.58% | −0.85% | −7.28% | +3.51% |
Prices as of November 5, 2026. The sector's move is the equal-weighted average of these listed companies, a market proxy for context, not recommendations. See it on the full heatmap →
Sentiment / The mood in this sector's stocks
Since 2018, Simulation & synthetic training data has been in Balance and steady 23 separate times. After those, its stocks beat their usual three months 15 times (65%), by a median of +5.0 points.
Confidence: low, a split like this is common by chance, so no edge is called.
Autonomous systems need enormous amounts of experience, and collecting it in the physical world is slow, expensive, and occasionally dangerous. Simulation generates that experience at a rate reality cannot match, provided what is learned there transfers.
Historically simulation was used for testing rather than training, because the gap between simulated and real physics was too large. Better rendering, better physics, and domain randomisation narrowed that gap enough to make training viable.
The structural driver is robotics and autonomy scaling. Every warehouse robot, autonomous vehicle, and manipulation system needs training data proportional to the variety of situations it will meet.
The technology layer spans physics simulation, photorealistic rendering, domain randomisation that varies conditions to force generalisation, synthetic data generation for perception, scenario libraries covering rare events, and the validation that measures whether transfer worked.
Adoption economics work because rare events dominate safety. The dangerous situations a system must handle are by definition rare in real data, and simulation is the only practical way to generate them in volume.
The beneficiaries include simulation platform vendors, synthetic data specialists, rendering technology firms, the accelerator makers whose hardware runs it, and the robotics companies that build capability in house.
The value chain runs from simulation engine through scenario content to model training and validation. Validated scenario libraries may prove more durable than the engines themselves.
The overlooked layer includes scenario content producers, sensor simulation specialists modelling lidar and radar accurately, validation and testing services, and the compute providers running large simulation workloads.
Competitive dynamics involve platform vendors competing against in house tooling at the largest robotics developers, which limits the addressable market at the top end.
Risks: the transfer gap is real and sometimes fatal to a project, largest customers build rather than buy, robotics deployment timelines keep slipping, and validation standards are immature.
What to watch: robotics deployment volumes, simulation to reality transfer results published, regulatory acceptance of simulation for safety validation, and scenario library licensing deals.