Should Your Digital Twin Be On-Premises or in the Cloud?
A digital twin is virtual and can reside on-premises or in the cloud. While the Department of Defense and intelligence community may prefer to keep their digital twins within their own secure data centers, civilian agencies may be comfortable with theirs being FedRAMP-authorized and cloud-based.
Cloud infrastructure allows for faster scaling of a digital twin at lower cost, while agencies need their own server resources for one on-premises. Because digital twins run on compute, the size of the desired simulation dictates the requisite server infrastructure.
The government can use digital twins to simulate security threats posed by hackers, estimating the size of the network’s attack surface and identifying potential system vulnerabilities.
An important lesson learned from early digital-twin deployments is to have a plan and a goal in mind, because it isn’t always necessary for an agency to replicate its entire network, device for device — which can be expensive to scale and manage. Have use cases prepared; that will save time on experimentation later.
DISCOVER: Digital twins are on the rise in government.
Choosing the Right Digital-Twin Use Case
Not all digital-twin vendors are the same. Some agencies use different vendors for different use cases, and different digital twins often have independent owners within the same agency.
Proofs of concept can take anywhere from a few weeks to many months to set up, including decisions on the requirements, data gathering, designing and implementing. Full digital-twin projects are phased endeavors that take a year or longer.
UP NEXT: CDOs want clarity around their AI responsibilities.
Consider a digital twin of the earth for analyzing global climate change. Predicting weather patterns and the trajectory of severe storms requires a lot of physics, math and computing power for the artificial intelligence model developed. Climate scientists, machine learning engineers and other experts must validate first the data and ultimately the simulation, but once that’s done, it can be used to model a variety of scenarios and can help alert citizens to prepare before natural disasters occur.
The possibility that Chinese AI firm DeepSeek’s latest large language model truly cost under $6 million to train has many experts optimistic over the economies of scale. It would mean, sooner than later, digital twins could be applied to far more use cases more cost-effectively and, in the government’s case, could exponentially improve citizen-facing applications.
