Nvidia Expands AI Dominance Beyond Chips to Full Ecosystems

Key Takeaways:

• Evolving Beyond Chips: Nvidia is transitioning from a chip supplier to a platform enterprise, aiming to encompass the entire "AI factory" value chain.
• Vera CPU Launch: Introduction of the Vera data center CPU, optimized for agent tasks, offering substantial performance improvements over traditional x86 CPUs.
• DSX Platform for AI Factories: Launch of the DSX platform, providing comprehensive digital simulation, deployment, and management solutions for AI factories.
• Entry into PC Processors: Unveiling of the RTX Spark PC chip, integrating CPU and GPU for Windows for Arm devices, targeting enhanced AI and gaming performance.
• Robotics and Autonomous Driving Push: Introduction of new robotics products, including Cosmos 3 foundational model, Nemotron AI models, and intelligent agent toolkits, alongside advancements in autonomous driving platforms.
• Deepened TSMC Partnership: Strengthening collaboration with TSMC on AI technologies to boost semiconductor design and manufacturing efficiency, leveraging Nvidia's tech in foundries.

In a significant strategic maneuver, Nvidia (NVDA.O) is broadening its operational scope, moving beyond its established position as an AI chip vendor. The company is positioning itself as a comprehensive platform enterprise, aiming to cover the complete value chain of "AI factories." This ambitious plan is accompanied by aggressive advancements in AI infrastructure, intelligent agents, humanoid robots, and autonomous driving technologies. Concurrently, Nvidia is pushing its silicon business into the data center CPU and personal computer markets.

At the GTC Taipei 2026 conference on June 1st, Nvidia unveiled a suite of new products and collaborative initiatives designed to facilitate this strategic evolution. Central to these announcements is the Vera CPU, a new data center processor engineered for the emerging era of intelligent agents. Nvidia states that Vera is purpose-built for CPU-intensive tasks such as agent processing, reinforcement learning, data handling, and task orchestration, boasting an 80% improvement in agent task processing speed compared to traditional x86 architecture CPUs.

The Vera Rubin platform, powered by the Vera CPU, has entered full production, with related systems slated for official release by system manufacturers and cloud service partners in the upcoming autumn. Nvidia also disclosed that OpenAI, Anthropic, and SpaceX will be among the first adopters of Vera chips. Beyond these prominent entities, organizations such as ByteDance, CoreWeave (CRWV.O), and Oracle (ORCL.N) cloud infrastructure are planning Vera platform deployments. Manufacturers including Dell (DELL.N), HPE (HPE.N), Lenovo Group (0992.HK), and Super Micro Computer (SMCI.O) are reportedly scaling up the construction of independent Vera CPU systems.

Jensen Huang, Nvidia's CEO, highlighted that Vera represents Nvidia's first standalone data center microprocessor, directly challenging established server processors like Intel's (INTC.O) Xeon series, AMD's (AMD.O) Epyc products, and Amazon's (AMZN.O) Graviton. He asserted that Vera can achieve up to 1.8 times the speed of Intel x86 architecture products on core AI foundational workloads.

On the infrastructure front, Nvidia is further bolstering its "AI factory" strategy with the introduction of the DSX platform. This platform is designed to cover the entire lifecycle of an AI factory, from design and simulation to deployment and operational management. According to Nvidia, enterprises can conduct complete digital simulations of a factory before physical construction, enabling validation of performance and operational efficiency, thereby mitigating construction costs and deployment risks.

Huang noted that with the rapid proliferation of generative AI, tokens have transformed from mere computational outputs into profit-generating assets, driving continuous expansion in compute infrastructure demand. He believes that the era of intelligent agents and practical AI has arrived, and the global demand for "AI factories" is growing rapidly.

Entry into the PC Processor Market

Beyond its data center initiatives, Nvidia has officially announced its foray into the personal computer processor market. Jensen Huang unveiled the RTX Spark PC chip, a collaborative development with MediaTek. This product features a converged CPU and GPU design, will run Microsoft's (MSFT.O) Windows for Arm operating system, and will debut in laptops and desktops from manufacturers such as Dell and Lenovo.

Details revealed indicate that RTX Spark will be equipped with up to a 20-core CPU and a Blackwell architecture GPU featuring 6144 cores. Both components will share unified memory and work in synergy via NVLink interconnect technology, aiming to enhance performance for large model execution and high-end gaming.

The chip will be manufactured by TSMC using its 3N process technology and is expected to launch in the autumn of this year. Huang stated that Microsoft and Nvidia have entered into a three-year collaboration aimed at redefining the personal computer experience for the AI era. He also disclosed that Nvidia plans to release new generations of PC chips concurrently with each new generation of its AI processors.

Accelerating Robotics and Autonomous Driving

In the burgeoning field of "Physical AI," Nvidia has introduced several robotics products. The company launched the Cosmos 3 open-world foundational model to drive physical AI development. It also released the Nemotron 3 Ultra next-generation AI model and introduced the intelligent agent toolkit, encompassing NemoClaw, Nemotron, OpenShell, and CUDA-X. Furthermore, Nvidia unveiled a large-scale open-source intelligent agent tool and skill suite for physical AI, alongside the Isaac GR00T reference humanoid robot platform for academic research.

Huang mentioned that Nvidia has launched its next-generation robotics AI models. Nvidia executives revealed plans to collaborate with humanoid robot manufacturers in the US, Europe, and South Korea on robotics research projects. For autonomous driving, Nvidia positions DRIVE Hyperion as a global platform for autonomous taxi operations and has released the Alpamayo 2 open inference model specifically for Robotaxi scenarios. Foxconn announced an expanded strategic partnership with Nvidia, intending to develop and deploy a fleet of L4 autonomous driving-capable Robotaxis in Taiwan. Foxconn is also piloting the NemoClaw system to support its Nurabot and CoDoctor platforms.

Deepening Collaboration with TSMC

As one of Nvidia's most critical manufacturing partners, TSMC is further expanding its AI technology collaboration. The two companies are jointly advancing AI technology integration into wafer fabrication production systems to improve semiconductor design and manufacturing efficiency. TSMC has already utilized Nvidia's technologies to enhance turnaround times, energy efficiency, yields, and operational productivity in its advanced fabrication plants.

In specific applications, TSMC is leveraging Nvidia's CUDA-X software libraries and AI models to accelerate GPU workloads and is exploring the construction of FabTwin virtual fab environments using the Omniverse platform. Concurrently, TSMC is advancing its production defect detection capabilities through the Metropolis platform and TAO toolkits.

As AI evolves from model training towards agents, robotics, and industrial applications, Nvidia is extending its technology ecosystem from chips to encompass data centers, industrial manufacturing, autonomous driving, and robotics. This broad expansion aims to solidify its core position within the global AI infrastructure ecosystem.