#USA #Seattle #Ravel #Sandia National Laboratories #Project Quicksilver

RAVEL and Sandia National Laboratories Partner on Project Quicksilver

RAVEL has joined forces with Sandia National Laboratories to launch Project Quicksilver, a groundbreaking research initiative aimed at revolutionizing AI infrastructure through power-aware technology. This ambitious project focuses on orchestrating artificial intelligence workloads across distributed computing environments, which holds the potential to redefine how organizations utilize AI capacity while being mindful of energy efficiency and operational costs.

Understanding Project Quicksilver

As organizations rush to expand their AI capabilities, they face a significant challenge: increasing computing capacity does not necessarily equate to enhanced productivity. Project Quicksilver aims to tackle this issue head-on. It explores how factors like power availability, infrastructure costs, and workload demands can influence decisions around scheduling and managing workloads across multiple data centers. By adopting smart orchestration techniques, the initiative seeks to optimize the use of every megawatt of power and every dollar spent on AI infrastructure.

Denise Muyco, the CEO and Co-founder of RAVEL, emphasizes that the key to leadership in AI infrastructure lies not in sheer volume but in efficiency and productivity. She states, “The next phase of AI infrastructure leadership won't be won by simply building more. It will be determined by how efficiently and productively we use every megawatt, every GPU, and every dollar invested.”

Researching Dynamic Power-Aware Orchestration

A core component of Project Quicksilver involves examining how to dynamically schedule and distribute AI workloads in response to changing energy availability and operational conditions. The first phase of the research focuses on comparing air-cooled and immersion-cooled computing environments across three different Sandia locations, all interconnected by a wide-area network. Researchers will manipulate variables such as network bandwidth and speed to investigate three crucial aspects:

1. Energy Efficiency: Assessing the energy demands and costs associated with identical workloads in various cooling settings, including evaluating energy use metrics and the potential benefits of heat reuse.

2. Total Cost of Ownership (TCO): Investigating how the scheduling and placement of workloads across distributed environments affect overall computing costs, including both capital and operational expenditures.

3. Workload Time-to-Completion: Analyzing how these decisions impact the execution time for AI workloads.

Dave Martinez, Engineering Program Manager at Sandia, highlights the importance of efficiently managing workloads across different computing environments, stating, “Under a Cooperative Research and Development Agreement with RAVEL, we'll explore whether dynamic workload placement can make better use of available power and computing resources while balancing performance and cost.”

Federated Orchestration with RAVEL Orchestrate AI

At the heart of this initiative is RAVEL's Orchestrate AI platform, which fosters a federated orchestration layer for managing AI workloads across various computing resources. This platform allows for coordinated workload management, integrating factors such as power availability and cost into the decision-making process.

Philippa Carroll, Chief Product Officer at RAVEL, adds, “As AI infrastructure becomes more distributed, the challenge is no longer just how much computing capacity we have, but how effectively we can use it. Project Quicksilver will investigate the trade-offs between power, cost, and performance, ensuring we make better and more responsible infrastructure decisions.”

Aiming for the Future of AI Infrastructure

The findings from Project Quicksilver could have significant implications for the future design and operation of AI infrastructures, particularly in optimizing workload management across distributed environments. As the project progresses, there are plans to extend the research to additional national laboratories and explore various use cases that emphasize environmental sustainability and performance enhancement.

For more insights into this innovative project, RAVEL and Sandia will present their findings at the SC26 conference from November 15-20, 2026, where they will discuss experimental approaches and anticipate research outputs promoting intelligent AI infrastructure.

In conclusion, Project Quicksilver represents a pivotal move towards not only expanding AI capabilities but doing so in a way that is energy-efficient and responsible, setting a new standard for AI infrastructure investment and management.