When Prith Banerjee walked on stage at Simulation World, he was not just representing Synopsys. He was speaking for a newly combined force: Synopsys and ANSYS, a merger that has redefined the landscape of simulation and EDA (electronic design automation).
Banerjee, Senior Vice President of Simulation and Analysis Incubation at Synopsys. He used to be the CTO of Ansys. At the conference, Banerjee outlines nothing less than a manifesto for the future of engineering: five pillars of technology that will allow industries to move from costly prototypes to fully virtualized design.
From Wind Tunnels to Code
Banerjee began with a reminder of the old ways. “Thirty years ago, when an airplane manufacturer like Boeing designed an airplane, they said, ‘I’m going to make a scale model of the airplane, put it in a wind tunnel and see if it flies.’
The same held true in automotive crash tests. Engineers built cars, crashed them with dummies, and hoped for the best.
“Oops. The dummy died. You just spent $10 million on the dummy,” he quips. “That was the world of physical prototyping.”
Now, he argued, products can be simulated with accuracy rivaling real-world tests. “The vision is zero physical prototyping—powered by HPC, powered by AI, powered by the cloud.”
The Five Pillars
Banerjee organized his keynote around five technology pillars:
- Numerical Methods – Solving partial differential equations (PDEs) like Navier-Stokes and Maxwell’s equations with advanced techniques such as mesh fusion. These improve accuracy without exponential increases in compute cost.
- High-Performance Computing (HPC) – Parallelization across GPUs and CPUs, scaling from single devices to 1,000+ GPUs. “Something that took months to run now runs in a matter of minutes,” Banerjee said, citing examples using ANSYS Fluent running on multi-GPU systems.
- Artificial Intelligence / Machine Learning (AI/ML) – Using AI to train models on simulation outputs, then generating near-instant results. “With AI you can get super-linear speed-up,” he said. Products like Ansys SimAI act as wrappers around solvers, while foundation models under development could make simulation as fast as inference.
- Cloud Platforms – Extending simulation beyond local machines. “You just can’t go to Best Buy and buy yourself a Blackwell machine,” Banerjee says, referring to NVIDIAs latest GPU-intense, AI-enabled computing platform. With partners like AWS, Google, and Microsoft Azure, Synopsys is offering “cloud bursting”—moving from desktop to up to 10,000 GPUs at will.
- Digital Engineering – A model-based approach that unifies mechanical, electrical, and software design. “In a large company like Boeing or Airbus, you still have engineers exchanging Excel sheets. That does not scale,” Banerjee said. The solution: a single source of truth linking requirements to subsystem design to digital twin.
Industries in Transition
These pillars, Banerjee argued, are not abstractions—they are reshaping industries.
- Automotive: The shift to software-defined vehicles requires simulating everything from battery performance to autonomous driving simulations in software. A Tesla today may carry 100 million lines of code, all of it running on chips that must themselves be simulated.
- Aerospace: Digital twins are replacing physical wind tunnels, cutting costs and timelines.
- Energy: Renewable systems such as wind turbines demand multiphysics modeling.
- Healthcare: Perhaps the most striking example, Banerjee said, is the human heart. ANSYS has built “the most accurate model of a heart,” which can be used for in-silico trials. Instead of waiting years for physical clinical trials, companies could test drugs or pacemakers on digital replicas.
“This is literally the future,” he said, showing a demo built with NVIDIA’s Omniverse where doctors could query a simulated heart in natural language.
The Quantum Horizon
Banerjee looks beyond the current success of GPUs to a future with quantum computing, where qubits can represent exponentially more states than classical bits. “It’s not going to happen today,” he admitted. “But in five years, you will see massive amounts of solvers and EDA tools running on quantum.” Synopsys is already experimenting with platforms like IBM Qiskit and NVIDIA CUDA-Q.
AI Agents and Design Exploration
Beyond raw compute, AI is reshaping design workflows. Banerjee described a future where generative AI agents propose design alternatives automatically, rather than waiting for human engineers to tweak parameters.
“AI agents doing the work—this is literally the future,” he said, noting Synopsys’s internal work on engineering copilots.
Cloud Burst and Democratization
One challenge of simulation has always been access. High-end computing clusters cost millions. By pushing workloads into the cloud, Synopsys hopes to democratize simulation. Engineers can start on desktops and, when needed, access thousands of GPUs on the cloud.
Banerjee likened it to renting compute by the hour rather than buying a machine. “As you are about to run a very high-profile simulation, you don’t have to buy a $10 million supercomputer. That’s the power of the cloud.”
Toward Digital Engineering
Banerjee returned to the idea of digital engineering as a unifying thread. Complex products—from cars, planes down to semiconductor chips—are now combinations of hardware and software, with interactions too complicated for siloed teams.
The future, he argued, is a model-centric process where specifications flow into subsystems, then into component design, all validated virtually.
No longer do you have to kill the dummy in the car, he said.
A Personal Note
Banerjee closed with a personal story. His father suffered from a heart condition known as a branch block. He imagined a future where doctors could simulate such conditions on digital twins, test drugs virtually, and recommend treatments without trial and error.
“What if you could do virtual patient designs? That’s the world of in-silico trials,” he said.
It was indeed a heartfelt appeal for a true digital twin, one that truly represents all aspects of a human body, a system of systems, not the partial digital representation that most engineering and design software companies try to pass off as a twin when it is but a shadow of the real thing.