Nvidia deploys Vera CPUs in its own chip design pipeline
Nvidia said it is using its Vera CPUs across electronic design automation (EDA) workflows that develop its next-generation processors, creating what the company called a "continuous feedback loop" between silicon design and engineering. The announcement came at the 2026 Design Automation Conference in Long Beach, California. Nvidia is collaborating with Cadence and Synopsys, the two largest EDA software providers, to optimize their platforms for Vera. Early testing showed up to 1.5x higher performance on selected production-class workloads when running Cadence's Jasper formal verification platform and Synopsys' VCS logical simulation tool on Vera. Vera combines 88 custom Nvidia Olympus CPU cores with LPDDR5X memory and the second-generation Nvidia Scalable Coherent Fabric, and is positioned to shorten individual verification runs while increasing throughput across large regression testing farms.
Rosa CPU and Rigel core on the roadmap
Nvidia said it plans to follow Vera with a successor called the Rosa CPU, powered by a new Nvidia Rigel core. The company framed the move as an effort to assign the most appropriate compute architecture to each stage of chip development, with GPUs and AI accelerating parallel tasks while Vera and its successors target latency-sensitive simulation, verification and implementation work. Nvidia said engineers will be able to evaluate more design alternatives and complete more validation within the same development window once Vera is deployed across its EDA flows.
AI agents and physics libraries join the chipmaking workflow
Nvidia said it is integrating its PhysicsNeMo physics-AI libraries and a set of GPU math libraries into the Nvidia Agent Toolkit, enabling AI agents to invoke accelerated solvers during chip design. The company introduced new CUDA-X libraries: cuISS for iterative sparse solvers, cuDSS for direct sparse solvers used in circuit and device simulation, and cuEST for quantum-chemistry simulations. According to Nvidia, the update makes PhysicsNeMo and CUDA-X "agent-ready" tools, letting autonomous agents run simulations to generate high-fidelity data more quickly than human engineers and participate directly in iterative engineering workflows.
Safe Superintelligence taps Vera Rubin GPUs in multi-billion deal
Safe Superintelligence (SSI), the AI lab founded by former OpenAI co-founder Ilya Sutskever, announced a long-term partnership with Nvidia, ending two years of relative stealth. The deal, which includes an undisclosed investment from Nvidia, gives SSI access to Nvidia's Vera Rubin GPU platform, with the chipmaker saying the startup's compute resources would grow "by an order of magnitude." A source familiar with the deal told TechCrunch that the investment runs into multiple billions of dollars; Bloomberg reported the deal size at $5 billion. Nvidia said it signed the compute partnership to "accelerate SSI's next stage of growth after obtaining rare access into the company's closely guarded research." Sutskever said SSI has "research that is worthy of scaling up" and is making "a big bet on the Vera Rubin platform." TechCrunch, citing PitchBook data, reported that SSI has raised $7 billion to date and is valued at $32 billion post-money, with backers including Andreessen Horowitz, Alphabet, Lightspeed Venture Partners, GV and Sequoia Capital Partners.
Siemens extends agentic EDA work with Nvidia
Siemens announced an expansion of its partnership with Nvidia to deliver self-verifying agentic AI workflows for semiconductor and printed circuit board design. The work builds on Siemens' Fuse EDA AI Agent system, now integrated into the company's Intelligence Center X environment. Siemens said it is achieving 10x to 15x speedups with its GPU-accelerated tools, including Solido Spice for circuit simulation. The expanded collaboration uses Nvidia's NeMo Gym open library, Nemotron reasoning models, Switchyard and the OpenShell secure runtime to let long-running AI agents reason about engineering trade-offs and continuously validate decisions against physics-based EDA engines. Nvidia vice president of computational engineering Timothy Costa and Siemens chief AI strategy officer for Solido custom IC Amit Gupta described the effort as a shift from autonomous task orchestration toward continuously validated engineering outcomes.
What to watch next
Nvidia and its partners are presenting further details on the Vera CPU deployment and EDA workflows at the 2026 Design Automation Conference. Nvidia's stated near-term milestones are the rollout of Rosa CPU development on Rigel cores and broader optimization of EDA software across its CPU roadmap, while SSI's compute footprint is expected to scale around the Vera Rubin platform in the coming quarters.
---
https://siliconangle.com/2026/07/26/nvidia-putting-vera-cpus-work-alongside-ai-agents-speed-chip-design/ https://blogs.nvidia.com/blog/vera-cpu-eda/ https://videocardz.com/newz/nvidia-uses-its-own-vera-cpu-to-build-future-cpus-and-gpus https://techcrunch.com/2026/07/27/ilya-sutskevers-safe-superintelligence-partners-with-nvidia-to-scale-its-ai-research/ https://gamesbeat.com/siemens-and-nvidia-launch-self-verifying-agentic-ai-for-chip-and-circuit-board-design/ https://blockchain.news/news/nvidia-vera-cpu-chip-design
Share this article
