$NVDAB *NVIDIA's Current Progress - 2026* NVIDIA is currently in the "Agentic AI" and factory-scale AI infrastructure phase. It is no longer just selling GPUs; it has evolved into a comprehensive AI systems company. Here is the summary: *1. New Hardware: The Vera Rubin Era & Agentic AI* - *Vera Rubin Architecture*: A new architecture succeeding Blackwell. It has entered production and is running more than six months ahead of schedule. - Key advantages: Inference cost per token drops by up to 10x, and GPU requirements for model training decrease 4x compared to Blackwell. It targets "AI factories" that operate non-stop. - *Vera CPU*: NVIDIA's new CPU designed to compete with Intel and AMD. It is specifically engineered for AI agents that work continuously without breaks. Perplexity reports that Vera is 1.5x faster for AI agent coding tasks. NVIDIA targets $20 billion in Vera sales for this fiscal year. - *LPU - Inference Processing Unit*: A new component for AI racks designed for low-latency inference. *2. PCs & Laptops: RTX Spark + N1X* - *RTX Spark*: A new "Superchip" for Windows laptops. It claims 1 Petaflop of AI performance, supports up to 128GB of unified RAM, and can run 120B-parameter LLMs locally. Partners include Microsoft, Lenovo, HP, Dell, and Asus. - *N1X*: An ARM laptop chip featuring 20 cores and a GPU equivalent to the RTX 5070, with 6,144 CUDA cores. It focuses on enabling laptops to run AI models with over 100 billion parameters. - *DLSS 4.5 Ray Reconstruction*: Releasing in August 2026. It utilizes a new transformer model for superior ray tracing and upscaling. It is already supported by 11 new games. - *G-SYNC Pulsar*: A new monitor offering 1000Hz motion clarity for competitive gamers. *3. AI & Software* - *Agentic AI*: The vision is for AI to go beyond merely answering questions and instead perform tasks autonomously—such as writing code, accessing databases, and generating reports. Jensen declared that "the era of clicking and typing is over." - *OpenShell runtime*: Enables the creation of local AI agents on RTX and DGX systems at twice