
Tech • AI • Robotics • Game
NVIDIA unveiled a sweeping shift toward “agentic AI,” introducing new chips, systems, and software to power a global boom in AI infrastructure and autonomous digital agents.
NVIDIA positioned “agentic AI” as the next major paradigm, where software agents can reason, plan, and act using tools. These systems combine large language models, memory, and external tools into coordinated workflows that resemble an operating system for AI. This shift replaces traditional click-and-type computing with intent-driven interactions, fundamentally changing how software is built and used.
The company introduced Vera Rubin, a next-generation AI system designed specifically for agent-based workloads. Unlike earlier GPU-focused systems, Vera Rubin integrates CPUs, GPUs, networking, and memory into a fully coordinated architecture. It is already in full production, with assembly time reduced from hours to minutes, signaling readiness for massive deployment.
NVIDIA highlighted the rise of AI factories, large-scale data centers optimized to generate AI “tokens,” now treated as units of revenue. These facilities can cost up to $100 billion per gigawatt and require tightly integrated design across power, cooling, and compute. The company’s DSX platform provides blueprints and operational tools to maximize efficiency and profitability.
AI-assisted software development is accelerating rapidly. Annual code contributions rose from 300 million in 2023 to 500 million in early 2025, with 2026 trends showing even faster growth. NVIDIA estimates 30–40 million developers generate about $3 trillion in wages but now enable up to $9 trillion in productivity, illustrating AI’s economic impact.
A new Vera CPU was introduced, designed specifically for AI agents rather than human users. It emphasizes ultra-low latency, high single-thread performance, and massive bandwidth. The chip delivers up to 3× faster SQL processing and 6× improvements in real-time data workloads, reflecting a shift toward AI-driven computing demands.
NVIDIA launched Nemotron 3 Ultra, an open AI model that is 5× faster and 30% cheaper to run than comparable systems. The company also released training data and tools, enabling enterprises to build custom AI agents. This supports a broader strategy of making AI development more accessible while strengthening its ecosystem.
A new class of AI-powered personal computers was introduced, including the RTX Spark platform. These systems can run AI agents locally, handle complex workflows, and integrate with cloud models. NVIDIA described this as the biggest transformation of the PC in 40 years, turning computers into always-on personal AI assistants.
The company unveiled Cosmos 3, a model for physical AI that can simulate and understand real-world environments, and Isaac GR00T, a humanoid robot platform. These systems aim to solve data limitations in robotics by generating synthetic training environments, accelerating development across industries.
NVIDIA also introduced Alpamayo, a reasoning-based autonomous driving system capable of real-time decision-making. Combined with its broader AI stack, the company is now connected to around 80% of global automakers and 97% of mobility services, signaling deep integration into transportation systems.
Taiwan was highlighted as a critical hub in NVIDIA’s ecosystem, spanning chip manufacturing, system integration, and supply chains. The region’s infrastructure and partners are essential to scaling AI globally, especially as demand for compute continues to surge.
NVIDIA is repositioning itself from a chipmaker to a full-stack AI infrastructure provider, betting that agentic AI will redefine computing, industry, and global economic growth.
Explain this