NVIDIA: Deploying AI Engineers to Transform Chip Design

In July 2026, Autonomous AI engineers have sped up complex system design as NVIDIA’s Timothy Costa highlights a major inflection point in simulation technology and engineering. NVIDIA has deployed autonomous AI engineers to assist with complex design cycles for next-generation chips and systems. Supported by the expanded NVIDIA Agent Toolkit, which is an open foundation that includes models, tools, and skills, the technology helps build advanced systems. The updated toolkit features NVIDIA PhysicsNeMo, which is a suite of agent-friendly libraries, along with upgraded CUDA-X libraries for solvers and quantum chemistry capabilities to directly support engineering workflows. NVIDIA introduced several technical breakthroughs to engineers and software developers, starting with bridging AI and fundamental physics through PhysicsNeMo libraries.

NVIDIA has also introduced several technical breakthroughs to engineers and software developers, starting with bridging AI and fundamental physics through PhysicsNeMo libraries. Rather than relying on traditional physics calculations, this technology helps developers to teach AI models how physical forces work, such as airflow around an aircraft wing or heat moving through an engine block. Engineers can integrate these customizable AI models directly into everyday design tools. This provides fast, reliable feedback without waiting hours for simulations to render.

To streamline hardware development, NVIDIA created Nemotron 3 Ultra, an AI model tuned specifically to help write and test chip blueprints. Working alongside ACE-RTLE, an AI system developed by NVIDIA Research, it understands chip design problems and writes hardware code better than other open-source AI models available today. Technology companies can take Nemotron 3 Ultra, run it on private, on-premises servers, and train it using confidential data. This approach gives companies complete privacy, customization, control, and efficiency.

Impact on the ICT Market

Physical models rely heavily on sparse linear systems, which are gigantic mathematical grids where most entries are zero, but non-zero parts remain difficult to solve. To address this, NVIDIA introduced cuISS, a library of iterative sparse solvers that makes an educated guess at complex calculations and refines them rapidly into an exact solution. Engineered to run across multiple GPUs, cuISS lets developers build fast, scalable simulation engines for automated workflows. For problems requiring numerical accuracy rather than smart guesses, NVIDIA offers cuDSS, a library focused on direct sparse solvers.

Unlike other traditional methods, direct solvers work straight through every step of an equation to find an exact answer without trial and error. This is important in Electronic Design Automation, where circuit layouts leave zero margin for error. This gives developers the high-powered math muscle needed to simulate complex microchips and system architectures seamlessly across massive computing clusters. Simulating how electrons interact within molecules traditionally demands vast computational power, limiting accurate tests to tiny clusters of atoms. The cuEST library changes this by accelerating complex quantum physics formulas on GPUs. This allows researchers to model much larger, real-world molecular structures, such as next-generation semiconductors, battery materials, or pharmaceuticals, while maintaining sub-atomic accuracy to discover breakthroughs.

Impact on the Edge Artificial Intelligence Chips Market

The global edge artificial intelligence chips market size accounted for USD 8.30 billion in 2025 and is predicted to increase from USD  billion in 2026 to approximately USD 36.12 billion by 2035, expanding at a CAGR of 17.75% from 2025 to 2035.

According to Precedence Research, edge AI chips have gained increased demand because of the fast penetration of IoT devices, smart cameras, wearables, industrial control systems, and autonomous technology. The development of these chips helps support real-time data analysis and processing without relying on the central computing infrastructure. Improvement in semiconductor technology has also contributed to better efficiency.

The integration of artificial intelligence is also accelerating the adoption of edge computing chips by enabling real-time data processing and decision-making directly on devices, reducing the need for constant cloud connectivity. This minimizes latency, enhances privacy, and improves efficiency for applications like autonomous vehicles, smart cameras, and industrial automation. As AI algorithms become more sophisticated, the demand for powerful edge AI chips that can handle complex computations locally continues to rise, driving widespread adoption across multiple industries.

Impact on the Data Center AI Chips Market

The global data center AI chips market features industry growth projections, key chip types, competitive landscape, and the role of cutting-edge processors in powering AI-driven digital transformation. The data center AI chips market is witnessing strong overall growth, driven by rapid adoption of generative AI workloads, rising cloud investments, and increasing demand for high-performance, energy-efficient processing solutions.

According to Precedence Research, as hyperscale are currently transitioning to more efficient, better performing architectures that can accommodate the rapid growth of generative AI LLMs, and expansion of their cloud-based automation needs, the industry is accelerating the introduction of new GPU, ASIC, and AI accelerator products that address the need to increase both performance and efficiency, both from an energy consumption perspective and from increasing chip density. The emergence and continuing growth of liquid cooling technologies, chiplet-based designs, and the use of advanced packaging technologies are all driving the next generation of AI Infrastructure.

The ever-increasing demand for greater computational speed and efficiency, combined with the mounting complexity of Artificial Intelligence models, made way for the regenerative proliferation of next-generation architectures enabled by advanced heterogeneous computing capabilities such as CPU/GPU combinations, specialized processors, memory-centric design elements, etc., which allow the processing of parallelized workloads with the least amount of latency possible.

Expert Opinion

Timothy Costa, Vice President and General Manager of Computational Engineering at NVIDIA, says: “Engineering has reached an inflection point. AI can now work with tools of physics, simulation, and design.

“With NVIDIA Agent Toolkit, developers can build agentic engineers that reason using physics, run complex simulations, and generate high-fidelity data to become a new engine for innovation in chip and system design.”