Nvidia has launched the BioNeMo Agent Toolkit, a domain-specific platform designed to bring agentic life sciences workflows to researchers and scientists. The toolkit provides accelerated life sciences tools spanning biology, chemistry, genomics and drug discovery. Industry and research leaders, including Dassault Systèmes, Databricks, Lilly, OpenAI, Schrödinger, Snowflake and the UW Medicine Institute for Protein Design, are adopting the technology, while Anthropic and OpenAI are integrating it.

The BioNeMo Agent Toolkit comprises several components, including NVIDIA Nemotron, NemoClaw, OpenShell and BioNeMo. These tools give agents the context and know-how to execute scientific computing. The platform aims to improve accuracy, task completion and token efficiency in scientific workflows.

Key Components and Capabilities

The toolkit integrates various NVIDIA technologies to support complex scientific tasks:

• NVIDIA Nemotron: Part of the core toolkit infrastructure.
• NemoClaw: A tool designed for specific agent interactions.
• OpenShell: Facilitates access to scientific computing environments.
• BioNeMo: The foundational platform for biological and chemical data processing.

Industry Adoption

Several prominent organizations are leveraging the new toolkit:

Nvidia Corp. has challenged concerns regarding the water consumption of artificial intelligence data centers, stating that such facilities account for a surprisingly small share of America's total water usage. The chipmaker cited findings from the Manhattan Institute indicating that data centers utilize only 0.2% of daily water usage in the U.S. This figure has reportedly decreased in recent years due to the adoption of advanced cooling technologies. Elon Musk, CEO of Tesla Inc. and Space Exploration Technologies Corp., publicly endorsed Nvidia's argument with a one-word affirmation of "True" on social media platform X. However, critics note that the data does not address water usage outside the data center, such as electricity generation and chip manufacturing, which increases a facility's total water footprint.

Cooling Technology Reduces Water Footprint

Nvidia explained that newer AI facilities are increasingly adopting liquid-cooling systems, which can significantly reduce or eliminate the need for traditional, water-intensive cooling towers. The company noted that AI facilities utilizing 45-degree Celsius liquid cooling can rely on dry coolers. This technological shift reduces cooling-related water consumption from approximately 2.6 million gallons per megawatt annually to near zero in favorable climates.

Rubin AI Systems Target Zero Water Use

In a blog post, Nvidia disclosed that its upcoming Rubin-generation AI infrastructure will be the company's first platform to feature a fully liquid-cooled architecture. Ali Heydari, Nvidia's director of data center cooling and infrastructure, stated that the NVIDIA DSX reference design for AI factories targets zero water consumption. Heydari added that dry-cooler-based systems operate as closed loops, eliminating almost all water usage except in limited circumstances. The design also aims to lower energy demand by capturing heat directly at the chip level, addressing the fact that cooling has historically represented up to 40% of a data center's electricity consumption.

Financial and Operational Benefits

The potential benefits extend beyond water conservation. Nvidia estimates a 50-megawatt hyperscale data center could save more than $4 million annually in combined cooling, energy, and water expenses by adopting liquid-cooled infrastructure. Cooling has historically represented one of the largest operating expenses for data centers, accounting for as much as 40% of total electricity consumption in some facilities. By capturing heat directly at the chip level, Nvidia says operators can significantly reduce both energy use and physical infrastructure requirements.

Industry Context and Resource Scrutiny

The debate over resource usage occurs as investors and policymakers scrutinize the infrastructure required for AI expansion. Amazon.com, Inc. disclosed this week that its data centers consumed approximately 2.5 billion gallons of water in 2025, though the company claimed its efficiency exceeds industry averages. Local opposition to new data centers has intensified across the U.S., with critics raising concerns about water availability, energy demand, noise, and infrastructure strain. Nvidia's latest claims suggest at least one of those concerns may be less severe than many critics believe, provided the liquid-cooling approach delivers the efficiency gains described.