Editorial contact for comments and feedback is [email protected].

Roche is expanding its AI infrastructure with 2,176 NVIDIA Blackwell GPUs, bringing its total on-premise and cloud GPU footprint to over 3,500—the largest announced in pharma.

Roche’s leadership emphasizes that time is the critical variable in healthcare, underscoring the need to accelerate research and development through AI.

The build-out began in 2023 as part of a broader Roche-NVIDIA collaboration aimed at shortening development timelines and reducing drug development costs.

Genentech’s Lab-in-the-Loop strategy already blends AI with experiments and data, delivering faster outcomes such as a degrader molecule for oncology designed 25% quicker and a backup molecule developed in seven months to mitigate risk.

Industry context notes AI’s potential to tackle undruggable targets and shorten costly clinical trials, potentially compressing timelines from about a decade and $1 billion per drug.

The surrounding page content is promotional, linking to sponsor materials and podcasts rather than independent reporting.

The investment aims to accelerate target identification, molecule design, manufacturing optimization, and therapy advancement in collaboration with Nvidia and partners.

Key figures include Genentech R&D head Aviv Regev and Nvidia CEO Jensen Huang, underscoring AI’s strategic importance and transformative potential in drug research.

Digital twins via NVIDIA Omniverse are used to simulate and optimize manufacturing processes, including a GLP-1 facility in North Carolina, supporting regulatory, quality, and scheduling activities.

Roche’s push follows similar AI initiatives, such as Lilly’s $1 billion AI lab with Nvidia, reflecting a broader industry trend toward AI-assisted drug discovery.

Key applications include Lab-in-the-Loop with BioNeMo, digital twins with Omniverse, accelerated genomics with Parabricks, digital pathology image analysis, and safe healthcare AI with NeMo Guardrails.