Artemis I—the program’s first mission—completed a 25-day uncrewed flight that looped around the Moon, validating the Orion spacecraft and NASA’s Deep Space Exploration Systems.  Artemis II advanced these capabilities with a crewed free-return trajectory reaching up to 230,000 miles beyond Earth.

Designing these trajectories is an extraordinarily computation-intensive endeavor. Workloads spawn tens of thousands of jobs across a set of nominal and off-nominal cases and generate 2-5 terabytes of data per launch period. During Artemis I trajectory development, some calculations were so complex they could consume the entirety of the team’s on-premises HPC environment for months.

To support mission planning timelines, additional scalable compute capacity was required. Achieving this meant extending on-premises capabilities into the cloud—a technically complex effort with no established playbook. This effort required expertise in HPC, cloud engineering, and mission-specific software adaptation.

The Trajectory Team’s on-premises HPC environment—managed through the Johnson Space Center Engineering Directorate—is among the most advanced available. However, the workloads did not always require extremely high-powered machines. Instead, they required more compute nodes operating in parallel. In essence, the team had access to a small fleet of very high-performance nodes when the code could be optimized to run with a larger amount of efficient, lower performance nodes.

Acquiring new on-premises servers was neither practical nor cost-effective. Procurement would be slow, and long-term ownership would introduce significant capital, maintenance, and lifecycle burdens.

Booz Allen implemented a cloud-based HPC expansion using AWS GovCloud (US), deployed within a FedRAMP High / ITAR-compliant environment, to augment existing on-premises resources. This scalable architecture enables the Trajectory Team to elastically add compute nodes as needed—an approach known as cloud bursting.

Through a mirrored environment design, both the cloud and on-premises clusters can operate in concert. With minimal code modifications, the Team can run its orbital design software, Copernicus, seamlessly across both systems—eliminating duplicative codebases and maximizing operational efficiency.

By augmenting existing infrastructure with cloud-based HPC, Booz Allen delivered a powerful, scalable, and cost-efficient environment that supports Artemis trajectory planning. The solution provides the Trajectory Team with:

• Significant increases in available compute capacity
• Faster trajectory analysis and mission planning cycles
• A flexible, cloud-ready architecture aligned to future mission growth
• Increased HPC reliability
• Reduced hardware investment and lifecycle overhead

As NASA prepares to send astronauts deeper into space than ever before, cloud-enabled HPC helps provide the computational strength and agility required to reach—and sustain—a presence on the Moon.