data Archives - | Carahsoft| Carahsoft

To maintain its place as a leader in AI advancements, and to comply with the latest White House guidance, Government agencies must harness AI capabilities, such as secure cloud computing platforms, high-performance data processing systems and scalable machine learning frameworks, for critical functions such as cybersecurity, predictive analytics and economic competitiveness. As with any new technology, AI requires updated infrastructure to power these advanced capabilities.

The Capabilities of AI Infrastructure

AI infrastructure refers to the hardware and software needed to create and deploy AI-powered applications and solutions. It enables both AI, the technology that simulates the way people think, and machine learning (ML), a focus area of AI that utilizes data and algorithms to imitate the way humans learn, increasing the accuracy of its results the more data you input. AI infrastructure enables users to create and deploy AI and ML apps, such as chatbots, facial and speech recognition and computer vision.

Building the infrastructure for AI requires data storage and processing, compute resources, ML frameworks and MLOps platforms to acquire the processing capabilities needed for AI, and also to train ML models.

AI Infrastructure Deep Dive

Below are the six pillars that define a strong AI foundation, each continuously evolving to keep pace with the next generation of AI capabilities.

Specialized Compute
In 2025, AI solutions rely on more than GPUs, they use a mix of processors designed for different types of AI tasks. This makes it faster and more cost-effective to train, update and run today’s complex models. As AI systems are becoming more advanced, many models are becoming larger and require HPC solutions. On the other hand, smaller models can run on cloud-based architecture for lower compute needs.

Data Preparation

The success of an AI solution can tie back to how well the data is prepared before it’s used. Modern AI infrastructure now includes built-in tools to clean, label and organize data at scale, sometimes using AI itself to automate the work. This ensures models are trained on accurate, relevant information, while also tagging and tracking data to meet security, compliance and transparency requirements.

Data Storage
Because today’s AI solutions are becoming more and more advanced, additional data is required to train the models. AI now depends on lightning-fast data storage that can easily grow alongside datasets. New tools also make it possible to keep sensitive data in specific locations or environments, meeting strict privacy and Government requirements without slowing down AI workflows.

Networking
As AI models get bigger, the speed of moving information between systems is critical. New high-speed networks reduce delays so AI can process and deliver results in near real-time, even across large environments.

Software & Orchestration
Managing AI today requires controlling the entire process from development to deployment. Modern platforms help teams easily update models, track their history and ensure they run efficiently whether in the cloud, on-premises, or in secure Government networks.

Security & Governance
AI infrastructure in 2025 is built with security at its core. It goes through rigorous testing to ensure it meets Government compliance standards and protects sensitive information. It is important to choose solutions from providers that continuously monitor their models, ensuring they’re safe, reliable and ready to be audited at any time.

All these AI Infrastructure features will be utilized by Government agencies to enable AI solutions that improve workflows and maintain global competitiveness.

AI Infrastructure: A National Priority

Executive Order 14141 names AI infrastructure, including data centers and compute clusters that are powered by clean energy, as a national priority to upholding U.S. leadership, national security and competition.

The order encourages Government agencies to secure supply chains, integrate clean energy and collaborate with the private sector. It also directs Federal agencies to make Federal lands and sites available for clean power generation and gigawatt-scale AI data centers

In alignment with the Executive Order, the Department of Energy (DOE) has released a Request for Information (RFI) to use its territories to build AI infrastructure datacenters, citing that they would enable AI training and inference, scientific research and other essential services.

Most recently, the AI Action Plan outlines recommended policy actions regarding building AI infrastructure such as data centers, semiconductor manufacturing facilities and energy infrastructure. The goal of the AI Action Plan is to streamline AI adoption and, in turn, speed up and scale the development of AI infrastructure on the federal level. National Security, AI incident response, cybersecurity and secure-by-design systems are highlighted as vital pillars of the AI Action Plan’s infrastructure guidance. By sharing specific steps to achieve safe and secure AI infrastructure, such as identifying available federal land, training our workforce, building data centers and keeping security at the backbone, the AI Action Plan outlines clear next steps that agencies need to take in order to push AI adoption forward.

In an increasingly technology-driven landscape, AI infrastructure allows Government agencies to modernize their operations and deliver more efficient, responsive services. Strategic investment in AI infrastructure enables agencies to enhance decision-making processes, reduce operational costs, protect national security interest and fulfill their core mandate of serving citizens. Once this foundation is in place, agencies can begin to build and deploy solutions that directly support their missions. The next blog in our series will explore how this infrastructure enables Generative AI and its potential for transforming Government workflows.

Carahsoft’s ecosystem of hardware and software vendors are equipped to connect agencies with the latest technology for AI, including the infrastructure needed to run it. To learn more about AI infrastructure solutions that are tailored for the Public Sector, visit Carahsoft’s Page on AI Solutions.

The Accelerating Pace of AI Evolution

The GenAI movement that emerged around 2017 added a new dimension, enabling AI to create content. However, it was the 2022 release of ChatGPT that democratized access to these capabilities, creating a “Wright Brothers moment,” springboarding the industry of AI. Suddenly, everyone from children to healthcare professionals began experimenting with these systems, often finding productivity gains despite the limitations of early versions of the technology.

Just as organizations were adapting to this new reality, reasoning models emerged in late 2024. These systems do not simply generate content, but think through problems step by step, mirroring human cognitive processes. Within months, more efficient, open-source reasoning models followed, making this technology accessible even for regulated industries like healthcare (e.g. Med-R1 8B).

GenAI Reasoning Models in Healthcare

GenAI enables healthcare professionals to work more efficiently, freeing time to engage with patients. Unlike earlier models, recent GenAI reasoning models provide transparency into their decision-making process. These models can now power advanced AI agents using healthcare-specific models like Google AIM, Med-PaLM 2 or Med-R1. This auditability is crucial in healthcare, where understanding why a recommendation was made is often as important as the recommendation itself.

HIMSS25 AI in Healthcare blog graphics_Embedded in Blog 2025

Before implementing AI agents and reasoning, agencies should define clear outcomes and goals. Here are several factors to consider when integrating GenAI into your agency:

Data Strategy: The effectiveness of AI models depends significantly on the quality and privacy of your data. Organizations need clear protocols for creating evaluation datasets and managing sensitive patient information that can be kept sovereign.

Infrastructure Decisions: Healthcare organizations must decide whether to deploy models in the cloud or on-premises, considering regulatory requirements and data sensitivity. A hybrid approach often provides the flexibility needed to address various use cases.

Model Selection: Open-source models now trail proprietary options by only about six months in capabilities while offering cost advantages and greater control. Many organizations are adopting hybrid strategies, using proprietary models for cutting-edge applications and open-source alternatives for routine tasks.

Scale Considerations: Small, specialized language models can be more efficient for specific healthcare tasks, while larger models may be necessary for complex reasoning about treatment options or research questions.

Agencies should prepare robust data governance frameworks and flexible infrastructure that spans cloud and on-premise environments to enable healthcare personnel to use GenAI effectively. Overall, GenAI enables healthcare professionals to work more efficiently, enabling them to connect more with patients.

Your Journey to an AI Future Starts Now

The future of healthcare will be augmented by reasoning models, making healthcare more affordable and accessible for all.

Some new, AI-driven areas to watch for include:

Data Interaction: LLMs will navigate complex healthcare data ecosystems, from electronic health records to genomic data, answering nuanced clinical questions without requiring complex programming.

Planning and Research: By functioning as collaborative partners in research, the models look to help design clinical trials, analyze research literature and develop treatment protocols.

Actionable Workflows: Reasoning models will help automate clinical and administrative processes while incorporating human feedback in a continuous improvement cycle.

AI agents will begin to help address the acute staffing shortages plaguing healthcare systems worldwide. These digital assistants can handle routine documentation, answer common patient questions, and provide decision support, allowing clinicians to focus on direct patient care. As AI systems become more affordable and consumption increases, we’re likely to see a revolution in healthcare accessibility, particularly for underserved populations, with AI agents augmenting healthcare workers’ efforts.

The journey toward AI-augmented healthcare is accelerating faster than most experts predicted. For healthcare leaders, the question is no longer whether to embrace these technologies, but how to implement them to improve care while maintaining the human connection that defines healthcare.

The content of this blog was pulled from the Healthcare Information and Management Systems Society (HIMSS) panel, “Accelerating Enterprise GenAI.” To learn more about Nutanix GenAI, visit Nutanix’s AI Solution page.