Nvidia CEO Jensen Huang declared AI PCs as a strategic inflection point comparable to the smartphone revolution.

Nvidia CEO Jensen Huang delivered the keynote address on Monday at GTC Taipei 2026, outlining the next evolution of AI compute. The presentation included updates on the Vera Rubin platform, a new lineup of Windows PCs developed with Microsoft for AI workflows, the launch of an enterprise agent toolkit, and next-generation AI infrastructure systems to accelerate data-center and agentic AI adoption. A team of Goldman Sachs analysts, led by James Schneider, attended the conference and identified three key investment themes: Nvidia's aggressive pursuit of the traditional PC TAM to drive momentum for Windows on ARM, its maintained advantage in datacenter-level performance and cost leadership relative to competitors, and its aggressive investment in driving agentic AI adoption across developers and ecosystem partners, with Vera Rubin revenue ramp remaining on track.

Nvidia announced that it is now ramping full production of its Vera Rubin platform, with multiple rack-scale systems—NVL72 GPUs, Vera CPUs, Groq 3 LPUs, BlueField storage, and Spectrum-X networking—contributing to AI factory designs. The company highlighted that Vera GPUs are purpose-built for agentic AI use cases, with up to 1.8X the performance of X86 systems and 10X agent throughput versus Blackwell. The company also highlighted its DSX AI Factory reference platform, which helps customers optimize their AI datacenters to bring operations up faster while optimizing power consumption and system uptime. Goldman Sachs expects a materially steeper revenue ramp for Rubin beginning in 3Q relative to Blackwell given meaningful manufacturing efficiencies and greater total capacity.

In collaboration with Microsoft and Mediatek, Nvidia launched a new Windows-based PC platform targeting AI workflows. The RTX Spark product combines a Blackwell RTX GPU with a 20-core Grace GPU co-designed with Mediatek using NVLink to deliver a high-performance PC experience optimized for AI applications, targeted at the premium segment of the market. OEM partners will launch laptop, desktop, and workstation systems beginning this fall, with launch partners including ASUS, Dell, HP, Lenovo, Microsoft, MSI, Acer, and Gigabyte. Huang stated: "The PC is being reinvented. For forty years, you launched apps. Click. Type. With RTX Spark and Microsoft Windows, you ask—and the PC does the work. RTX Spark brings everything NVIDIA has built—CUDA, RTX, our AI platform—into a single superchip. Local agents. Frontier models. Creative workflows. RTX games. All on a laptop. This is the new PC. The personal AI computer." Jensen Huang characterized the AI PC reinvention as "a new line" and "a new beginning," placing it on par with the smartphone shift.

Nvidia also announced a series of new software releases targeting agentic AI use cases in the enterprise, including NemoClaw, Nemotron 3 Ultra, OpenShell, and CUDA-X Agent Skills. Additionally, the company launched new versions of its open Cosmos (v3) frontier model targeting multi-modal reasoning, and Alpamayo (v2) targeted as a reference platform for self-driving cars, alongside its first open reference design for humanoid robots based on its Isaac Gr00t and Jetson Thor hardware platform.

Market reaction to the announcements was swift: Nvidia shares rose 2.5% in premarket trading in New York, while Arm ADRs soared 12% as traders viewed Nvidia's PC push as supportive of the Arm ecosystem. However, the announcement pressured incumbent processor stocks, with Intel sliding 6%, Qualcomm down 9.5%, and AMD falling 3.5%. Goldman Sachs' Schneider is "Buy" rated on NVDA with a 12-month price target of $285, based on a 30X P/E multiple applied to his team's normalized EPS estimate of $9.50. Separately, Intel announced a new AI chip, code-named Crescent Island, expected to hit the consumer market by the end of the year, with Kevork Kechichian, who leads Intel's data center group, stating: "We decided to start rebuilding our muscles in AI . . . [but] we are not particularly aiming for [the training market] based on past experience."