In a strategic move to accelerate customer adoption, NVIDIA confirmed it will not require advance payment for its latest H200 AI accelerators. The decision removes a significant capital barrier for cloud service providers and hyperscalers looking to deploy next-generation inference engines. NVIDIA’s H200 series delivers up to a 2.5× performance boost over its predecessor for large-language-model workloads, and management forecasts that the pay-as-you-go model could boost H200 unit shipments by at least 30% in fiscal year 2026. CEO Jensen Huang announced that NVIDIA’s new Rubin platform—comprising six novel accelerator chips—will enter production in the second half of 2026. Early benchmark results indicate up to a 90% reduction in cost per inference token compared with the current Blackwell series, while delivering a 3× uplift in agentic AI reasoning performance. With a sales backlog that NVIDIA estimates will drive combined Rubin and Blackwell revenue of approximately $500 billion next year, the company projects at least 20% annual growth in data-center sales through 2028. At the 44th Annual J.P. Morgan Healthcare Conference in San Francisco, NVIDIA reiterated its position as the preeminent provider of accelerated computing for AI in healthcare. Vice President Kimberly Powell detailed how the company’s end-to-end hardware and software stack—spanning high-performance GPUs, the CUDA accelerator framework and foundational models such as NeMo and BioNeMo—powers imaging, genomics, life-science research and drug discovery workflows. She cited a tenfold increase in healthcare-focused DGX system deployments over the past two years and noted that NVIDIA’s Clara suite now underpins over 150 commercial collaborations with imaging and genomics partners, driving a 70% reduction in model-training time for new diagnostic algorithms. NVIDIA and Thermo Fisher Scientific launched a strategic collaboration to embed AI into laboratory instrumentation and workflows. Leveraging NVIDIA DGX Spark supercomputers alongside the Clara and BioNeMo model libraries, the partnership aims to reduce manual steps in sample prep, instrument setup and data analysis by 60%. Thermo Fisher, which generates over $40 billion in annual revenue, will integrate NVIDIA’s AI infrastructure into its mass spectrometry, sequencing and chromatography platforms, targeting a 50% increase in throughput for life-science customers over the next 18 months.