At NVIDIA GTC Berlin, Synopsys exhibited more than 20 GPU-enabled EDA and multiphysics solutions and customer results ranging from 12x to 40x across lithography, power devices, automotive lighting, and photonics. Futurum’s view is that the Ansys portfolio is converting into NVIDIA GPU consumption and the customer evidence is now strongest in photonics, where design iteration speed decides who supplies the optical layer of the AI buildout.
What Is Covered in This Article:
- More than 20 GPU-enabled EDA and multiphysics solutions and support for 50 NVIDIA CUDA-X libraries at NVIDIA GTC Berlin
- imec reporting up to 40x faster GPU-accelerated S-Litho computational lithography across DUV, EUV, and High-NA EUV processes
- Infineon reporting a 12x speedup in 3D power-device breakdown simulation using Sentaurus Device with NVIDIA cuDSS
- Synopsys joining the NVIDIA Halos safety ecosystem, with a FORVIA HELLA lighting workload on T-Systems’ sovereign cloud cut from over 4.5 hours to about 19 minutes
- iPronics running Ansys Lumerical FDTD on NVIDIA H100 GPUs in AWS with a 13x solver speedup and a nine-day edge-coupler sweep cut to 18 hours
The News: Synopsys (Nasdaq: SNPS) showcased more than 20 GPU-enabled EDA and multiphysics solutions at NVIDIA GTC Berlin, with support for 50 NVIDIA CUDA-X libraries and frameworks across its portfolio. The customer results span the design stack. imec reports up to 40x faster computational lithography with GPU-accelerated S-Litho across DUV, EUV, and High-NA EUV, and Infineon reports a 12x speedup in 3D breakdown simulation of next-generation power devices using Sentaurus Device with NVIDIA cuDSS. A FORVIA HELLA automotive exterior lighting workload run with Ansys Speos in T-Systems’ sovereign Industrial AI Cloud dropped from over 4.5 hours on CPUs to about 19 minutes, and photonics startup iPronics reports a 13x solver speedup running Ansys Lumerical FDTD on NVIDIA H100 GPUs in an AWS environment, with a 20-point edge-coupler parameter sweep falling from about nine days to 18 hours. Synopsys is also joining the NVIDIA Halos safety ecosystem, connecting Ansys AVxcelerate sensor simulation and Ansys medini safety analysis to a traceable workflow for autonomous vehicle development.
Synopsys GPU-Accelerated EDA Shows up to 40x Gains at GTC Berlin
Analyst Take: GPU-accelerated EDA has moved from a benchmark exercise to a customer evidence contest and Synopsys brought the longest list to Berlin. The portfolio claim matters less than the names vouching for it, with imec on the lithography models that will qualify High-NA EUV production, Infineon on the power devices Europe’s industrial base depends on, and a Valencia photonics startup that closed a $125 million Series B seven weeks ago with NVIDIA on the cap table. The bull case is that the Ansys portfolio gives Synopsys the broadest physics surface in EDA to accelerate, and the named results, 12x to 40x, are large enough to change how engineering teams buy compute. The bear case is that every number in the release is a vendor-selected best case on a single workload with no revenue attached, and Cadence used the same event to announce sovereign AI infrastructure of its own in Munich. The implication is that the GPU direction in design software is settled, so the competition is now over whose customers prove it fastest in production.
The Ansys Portfolio Is Becoming an NVIDIA Distribution Agreement in Practice
Count the tools behind the Berlin proof points and the Ansys acquisition logic surfaces. Lumerical, Speos, AVxcelerate, and medini are Ansys products, while Sentaurus Device and S-Litho come from classic Synopsys TCAD and mask synthesis. The $35 billion Ansys acquisition gave Synopsys the multiphysics surface area, and the NVIDIA partnership converts it into GPU consumption across sensor simulation, optics, device physics, and lithography in one announcement. The cuDSS sparse direct solver behind Infineon’s 12x gain reached general availability at GTC Berlin, which opens the same acceleration path to every Sentaurus customer. The financial stakes trace to the September 30 Investor Day, where management told investors Ansys-fused products would exit fiscal 2027 above a $100 million revenue synergy run rate on the way to $400 million by fiscal 2029, and GPU acceleration makes those numbers plausible because a portfolio that runs 12x to 40x faster invites customers to run more physics per design rather than the same physics cheaper. A note of caution is that acceleration that collapses runtime can also collapse the core-hour licensing that priced the CPU era.
iPronics Gives GPU-Accelerated EDA Its Clearest Economic Story
The iPronics result deserves the spotlight Synopsys gave it because the company is building the hardware category that decides how AI clusters scale. iPronics spun out of Universitat Politècnica de València in 2019, commercializing programmable photonics research dating to around 2014, and redirected that reconfigurable fabric toward optical circuit switching for AI data centers. Its first product, the ONE-32, became available in May 2025 as the first optical circuit switch built on silicon photonics, packing 32 ports into 1U with under 30 ns of latency and reconfiguration in under 300 microseconds, and the company says it cuts switch power by up to 50% by roughly halving the transceivers a cluster requires.
Incumbent OCS products steer light with MEMS mirrors that reconfigure in tens of milliseconds, so a solid-state switch that reconfigures in microseconds can act as a scale-up fabric and a dynamic traffic optimizer rather than a slow topology-change tool. Google settled the category by deploying OCS to reconfigure TPU topologies, and the $125 million Series B that closed on September 2, with NVIDIA backing the second-generation OCS technology, funded iPronics’ merchant answer.
The GTC Berlin result shows how a startup outpaces deeper-pocketed rivals. A 20-point edge-coupler sweep that took about 9 days in a 96-core CPU environment now completes in 18 hours on NVIDIA H100 GPUs in AWS. Coupler loss and polarization behavior determine whether a photonic switch can scale beyond 32 ports, so simulation velocity is the binding constraint on the roadmap. NVIDIA now sits on every side of the startup’s business, as investor, design flow compute supplier, and networking partner through Spectrum-X. GPU-accelerated photonics simulation is becoming the entry ticket for the optical layer of the AI factory and Synopsys sells the ticket.
Cadence Answered at the Same Event With Infrastructure Synopsys Chose Not to Build
The competitive split at GTC Berlin was unusually clean. On the same day, Cadence announced a sovereign AI data center in Munich combining OnCloud, Millennium M2000 systems on NVIDIA Blackwell, and its Reality Digital Twin Platform, selling jurisdiction and owned capacity as the product. Synopsys showed the opposite posture. Its sovereign proof point runs on T-Systems’ Industrial AI Cloud, its photonics customers run on AWS, and its capital stays out of data centers. Cadence bets that sovereignty-constrained customers will pay for EDA compute inside a boundary it operates, and Synopsys bets that partner clouds will supply that boundary without the fixed cost. Siemens EDA, the vendor with a German parent and a natural sovereignty claim, has yet to respond to either move. NVIDIA profits from every outcome, since Halos, cuDSS, CUDA-X, and Blackwell capacity are present in each version of the story. The open question is whether capital-light reach or owned sovereign capacity wins the renewal cycle, and fiscal 2027, the year Synopsys has promised 15% growth and Cadence must fill Munich, will answer it.
What to Watch:
- Whether Synopsys discloses GPU-related consumption or cloud revenue detail on its Q4 FY 2026 earnings call in early December
- Whether cuDSS general availability pulls Sentaurus and multiphysics workloads onto GPUs beyond the named early customers
- Whether the 13x to 40x speedups are reproduced on customer production workloads outside vendor-selected showcases
- Whether iPronics names a tier-one hyperscaler deployment for its Optical Networking Engine following the $125 million Series B
- Whether Synopsys answers Cadence’s Munich sovereign data center with owned European infrastructure rather than partner clouds
The announcements are detailed in the Synopsys GTC Berlin news release on the Synopsys newsroom.
Sources
Declaration of generative AI and AI-assisted technologies in the writing process: This content has been generated with the support of artificial intelligence technologies. Due to the fast pace of content creation and the continuous evolution of data and information, The Futurum Group and its analysts strive to ensure the accuracy and factual integrity of the information presented. However, the opinions and interpretations expressed in this content reflect those of the individual author/analyst. The Futurum Group makes no guarantees regarding the completeness, accuracy, or reliability of any information contained herein. Readers are encouraged to verify facts independently and consult relevant sources for further clarification.
Disclosure: Futurum is a research and advisory firm that engages or has engaged in research, analysis, and advisory services with many technology companies, including those mentioned in this article. The author does not hold any equity positions with any company mentioned in this article.
Analysis and opinions expressed herein are specific to the analyst individually and data and other information that might have been provided for validation, not those of Futurum as a whole.
Read the full Futurum Group Disclosure.
Other Insights From Futurum:
Synopsys Investor Day 2026 Turns EDA Into a Royalty and AI Model Revenue Share Business
Synopsys Autopilot Aims to Consolidate Chip Design Around One Stack
TSMC’s OIP Forum Turns Agentic Chip Design Demos Into Production Flows
Author Information
Brendan is Research Director, Semiconductors, Supply Chain, and Emerging Tech. He advises clients on strategic initiatives and leads the Futurum Semiconductors Practice. He is an experienced tech industry analyst who has guided tech leaders in identifying market opportunities spanning edge processors, generative AI applications, and hyperscale data centers.
Before joining Futurum, Brendan consulted with global AI leaders and served as a Senior Analyst in Emerging Technology Research at PitchBook. At PitchBook, he developed market intelligence tools for AI, highlighted by one of the industry’s most comprehensive AI semiconductor market landscapes encompassing both public and private companies. He has advised Fortune 100 tech giants, growth-stage innovators, global investors, and leading market research firms. Before PitchBook, he led research teams in tech investment banking and market research.
Brendan is based in Seattle, Washington. He has a Bachelor of Arts Degree from Amherst College.
