Today’s agencies are being asked to serve citizens better in an ever-changing world while making the most of their resources. Most leaders recognize that AI will be part of the answer to this challenge. However, in order to capitalize on the full potential of AI, agencies will likely need to change the way they capture, process, store, transfer and protect their data. For this reason, the chief data officer (CDO) role is becoming increasingly important in the federal government.
CDOs are charged with developing the full lifecycle of their agencies’ analytics strategies, along with choosing the technology to do so. The CDO is typically accompanied by a multidisciplinary support team of subject matter experts specializing in data management, cybersecurity/regulatory compliance, network and cloud architecture, application development, and other fields.
As more emphasis is placed on developing an enterprise-wide AI strategy, a new chief AI officer (CAIO) role has also emerged to oversee critical AI development projects. Over time, the CDO and CAIO roles are likely to merge. Leaders in this new combined position will have a mandate to capitalize on the full power of AI for their organizations. Stay tuned, as this is a continuously evolving discussion.
While the CDO role isn’t new, we at Equinix have seen first-hand how it’s changed in recent years. We increasingly hear from CDOs who are taking on greater responsibility for guiding their agencies’ digital infrastructure priorities.
In many cases, CDOs are looking for assistance with building out their AI data architecture. For instance, they may need help getting the source data to train their AI models or building low-latency connections between edge locations to enable inference and retrieval-augmented generation (RAG). Of course, they’re also concerned with issues of data privacy, sovereignty and security. Meeting the AI imperative wouldn’t count for much if they have to put sensitive data at risk in order to do it.
Agency CDOs have their work cut out for them, but the value that AI presents—the ability to innovate, speed up processes and make better decisions—can certainly make it worth the effort.
What are examples of AI use cases for government?
The list of AI use cases that government agencies can pursue is practically endless. In 2024, the Office of Management and Budget published a consolidated inventory that included more than 1,700 specific AI use cases across all stages of lifecycle development. This is more than double the number from the 2023 inventory, so it’s clear that many agencies are already building out their AI strategies quickly.[1]
Here are just a few examples of the kinds of use cases that were included in the inventory:
- Customs and Border Patrol: CBP is using AI to help identify illicit drugs at ports and border crossings, while avoiding delays for legitimate cargo.
- Transportation Security Administration: TSA is using AI to help screen baggage and passengers at airports quicker and more accurately, reducing wait times for travelers. New threats emerge and evolve continuously; AI-enabled processes can help agents track and identify them more effectively.
- Federal Emergency Management Administration: FEMA is using AI to prepare disaster response strategies in a more holistic manner. This could include proactively identifying existing vulnerabilities in communities that are underserved or threatened by rising sea levels. Additional variables that factor into the recovery equation (including access to basic essentials, power grids and medical resources) can vary greatly across different regions of the country. AI-enabled analysis can help intuitively bridge those knowledge gaps to enable optimal disaster recovery response.
- Center for Disease Control: The CDC is using AI to help predict the spread of future disease outbreaks and perform active disease surveillance. This will help better coordinate responses with state, tribal and local health departments.
The common factor across these and other use cases is that they’re all dependent on data from various sources. To enable them, agencies need to capture, process, transfer and secure very large AI datasets. To do this, they need access to AI-ready data centers and related digital infrastructure.
Where should government agencies deploy their AI workloads?
Different government agencies have different priorities when it comes to AI, and this impacts their choice of AI infrastructure accordingly.
Some agencies want to take advantage of pretrained models from a service provider. These agencies are primarily concerned with how best to fine-tune those models and perform inference and RAG in the right places. Therefore, their priorities are deploying edge infrastructure in proximity to data sources and end users and ensuring low-latency network connectivity. They can still benefit from deploying powerful GPU hardware, but that shouldn’t be their top concern.
For other agencies, using a publicly available model would be out of the question, due to privacy and security concerns. Instead, they need to build their own private models. Last year, the Air Force launched NIPRGPT, a private model intended to help enlisted personnel, civilian employees and contractors to experiment with generative AI while ensuring adequate safeguards.[2] The Army has launched a similar private model called CamoGPT. This model is hosted on both NIPRNet and SIPRNet—the Department of Defense (DoD) networks for unclassified and classified information, respectively—as well as cloud and edge environments. This diversity of hosting environments makes CamoGPT an ideal solution for a variety of Army use cases.[3]
Agencies that train their own models need access to GPU capacity in order to process massive volumes of training data. The question of where to deploy those GPUs is paramount. In order to be AI-ready, a data center must offer:
- Advanced cooling capabilities to support high-density AI hardware
- Energy efficiency and sustainability innovations
- Access to a robust ecosystem of partners and service providers
- Networking solutions that make it easy to connect distributed infrastructure and workloads
Public sector agencies can find all these things inside Equinix IBX® colocation data centers.
Why Equinix for government AI use cases?
Equinix was founded more than 25 years ago, and we’ve been a trusted partner to public sector agencies throughout our history. Agency research and development networks, such as the DoD’s High Performance Computing Modernization Program (HPCMP) and the Department of Energy’s Energy Sciences Network (ESnet), have been peering on Equinix Internet Exchange for almost two decades now.
Various laboratories and research institutes have peered with these agency networks over the years, creating ready-made communities of interest at Equinix. Our customers can tap into these communities on a turnkey basis to help accelerate their AI strategies.
Equinix is also the global leader in cloud on-ramps, including connections to all major cloud providers in locations throughout the world. Agencies can incorporate a dedicated storage environment into their hybrid multicloud architecture at Equinix. This allows them to tap into cloud services on demand to optimize infrastructure flexibility, while also maintaining control over their data and meeting their data privacy and sovereignty requirements.
Building an AI strategy is a lot like cooking a good meal. Equinix is the ideal kitchen in which to prepare your AI recipes, because the tools and ingredients you need—the infrastructure, technology and ecosystem—are already here. To get cooking, all you need to do is bring your data.
As agencies look to get AI-ready and overcome other modern IT challenges, their existing on-premises data centers may hold them back. Equinix high-performance colocation data centers provide a quick, cost-effective way for agencies to modernize their digital infrastructure.
Learn more about how to shape your future with AI-ready infrastructure: Access the infographic.
[1] Madison Adler, Federal government discloses more than 1,700 AI use cases, FedScoop, December 18, 2024.
[2] Department of the Air Force launches NIPRGPT, Secretary of the Air Force Public Affairs, June 10, 2024.
[3] Lori McFate, Operationalizing Science at JMC with Artificial Intelligence and Machine Learning, U.S. Army, October 29, 2024.
Subscribe to the Equinix Blog