News and papers on surrogate models and physical AI.surrogate models.physical AI.neural simulation.
AI is learning to simulate the physical world: surrogate models, neural solvers and physical AI predict flows, stresses and heat in milliseconds instead of hours. A curated feed of the news and papers moving the field, each with what it does and why it matters, and a course to learn it from the ground up.
Latest news
All newsIndustryVinci raises $250M at a $1.5B valuation for physics AI in chip designA physics foundation model for hardware, starting with chip thermals and moving to full systems.$250MSeries B, led by Advent, Temasek and Xora Innovation$1.5Bvaluation15B+degrees of freedom per design, as reported by the companyVinci
ResearchCANTO: aerodynamic fields straight from parametric CADA transformer neural operator that predicts aerodynamic fields directly from parametric CAD, and gives gradients to improve the design.NVIDIA Research · AI-Aided EngineeringIndustryPhysical Superintelligence launches with a $58M seed roundA new "AI-native physics lab" building virtual physicists, starting with data-centre design.Physical Superintelligence
Highlighted papers
All papers- CANTOCAD-Native Transformer Operators for AI-Aided EngineeringTokenizes native NURBS CAD patches and predicts continuous surface and volume fields at any query point.arXiv 2026Leibovici et al.
- AB-UPTScaling Neural CFD Surrogates for High-Fidelity Automotive AerodynamicsA few anchor points carry full attention; millions of query points read only the anchors, with separate surface and volume branches.TMLR 2025Alkin et al. (Emmi AI)
- AeroJEPASemantic Latent Representations for 3D Aerodynamic FieldsPredicts the latent representation of the flow from geometry and conditions, and decodes a field only when one is needed.arXiv 2026Giral et al.
NewsletterNew surrogate models and papers, in your inbox.The latest news and highlighted papers, every two weeks. No tracking, unsubscribe anytime.