Nvidia bets big, Vera Rubin targets $500b
Nvidia has accelerated the launch of its Vera Rubin compute platform, now scheduled for the third quarter of 2026. The platform integrates several new components, including a dual-chiplet Rubin GPU with 288 GB of HBM4 memory, a Rubin CPX accelerator using GDDR7, a Vera CPU, and a BlueField-4 DPU. These elements are connected by a sixth-generation NVLink switch and next-generation networking adapters.
The company is projecting approximately 500 billion US dollars in GPU-compute sales by the end of 2026, a forecast that deliberately excludes the Chinese market. This ambitious target highlights Nvidia’s strategic pivot toward comprehensive artificial intelligence and high-performance computing systems. Chief Executive Officer Jensen Huang confirmed the company continues discussions with American and Chinese regulators, though it is currently proceeding without that market.
This shift in focus signifies a move away from traditional gaming GPUs toward integrated systems. Future performance evaluations are expected to center on the combined throughput of the GPU, DPU, and CPU alongside memory and interconnect capabilities.
