85% of the organizations that answered our survey, launched on June 10th, 2026 during a NAFEMS and Teratec event in Paris, “AI & Numerical Simulation”, have begun testing or deploying AI in their simulation workflows.

Only 26.2% report being operationally satisfied with what those initiatives have delivered. If your organization has a pilot that shows promise but has never touched a real design decision, it has plenty of company.

Between 2021 and 2026, as CAE teams were testing convolutional networks, graph networks and physics-informed models, transformers became the default architecture in nearly every other domain of applied AI. And yet, according to our survey, only 5% of the engineers surveyed have experimented with transformer architectures in their simulation workflows. Simulation appears to be one of the last fields where transformers remain a curiosity. Is that a blind spot, or justified caution? Transformers did, after all, arrive late to simulation and for real technical reasons.

This article lays out what the survey respondents highlighted as the main adoption obstacles, what a transformer-based simulation model actually offers on those specific points, and what happened when the question was put to an open industrial benchmark.

This article appeared in the October 2026 Issue of BENCHMARK.