AI and the Future of Colocation

AI and the Future of Colocation: Protecting Public Data & Empowering Communities

As artificial intelligence (AI) continues to evolve, public sector agencies across the U.S. are exploring how this powerful technology can enhance services, streamline operations, and support community outcomes. At SCITDA 2025, DartPoints' VP of Solution Architects, Matthew McKee, outlined how AI is reshaping the future of colocation and why a secure, localized approach is essential for state and local governments (SLED).

AI in the Public Sector: Already Transforming Services

AI isn't a futuristic promise; it's already revolutionizing how state and local governments operate. From smart city infrastructure and digital resident portals to predictive analytics for public safety, artificial intelligence is being woven into the fabric of public services.

Some of the most common use cases include:

• Smart Cities: Using AI to manage traffic, optimize utilities, and monitor infrastructure health.
• Resident Services: Chatbots and self-service portals that reduce wait times and improve accessibility.
• Public Safety: Real-time analytics that aid in emergency response and crime prediction.
• Education & Research: Supporting both university research and K-12 learning environments.
• Healthcare: Improving diagnostics, patient care, and public health monitoring.

According to Deloitte, 70% of state and local governments are already using AI-enabled services. Statista predicts that 92% of global cities will invest in smart city technologies by 2025.

The Data Sovereignty Challenge

But as governments embrace AI, a critical concern arises: where should public data live? Offloading workloads to hyperscalers or public LLMs can create serious vulnerabilities, from vendor lock-in to privacy risks and compliance gaps.

Consider the risks:

• Public trust can erode if data is processed outside state jurisdiction.
• FOIA and data privacy laws may be harder to enforce.
• Costs may escalate over time with opaque pricing models.
• Sensitive datasets risk being absorbed into global models with unclear governance.

Gartner reports that only 18% of government data is currently self-contained. Meanwhile, IDC notes that 42% of public-sector CIOs identify data sovereignty as a leading concern.

Building or Hosting AI Yourself: Not Always Feasible

Training your own language model sounds ideal but it's rarely practical. Training a large language model (LLM) from scratch can cost more than $10 million, and even then, public LLMs often outperform custom-built ones. GPT-4, for example, scored over 84% on the U.S. medical licensing exam without any domain-specific tuning.

Hosting these workloads on-premises comes with its own challenges. GPU-powered AI requires up to 15 times more power than traditional workloads. Facilities need advanced cooling systems, high-density power availability, and 24/7 expert support, resources most public agencies can't easily divert from their core missions.

What AI Infrastructure Really Requires

To responsibly scale AI initiatives, government agencies need access to purpose-built environments that deliver:

• High-performance GPUs without hyperscaler markups
• Dense, reliable power and cooling infrastructure
• Low-latency regional network connections
• Compliance with HIPAA, FERPA, CJIS, and other mandates
• Predictable, transparent pricing

Three Models for Responsible, Sovereign AI

There isn't a one-size-fits-all solution, but here are three infrastructure models DartPoints often recommends for SLED agencies:

1. Local In-State RAG / Vectorization. Sensitive data remains local, while AI models retrieve only what's needed via secure queries. This model can support up to 80% of government AI applications, according to McKinsey.
2. AI-Enabled Private Cloud. Here, GPU infrastructure is provided and managed by a colocation partner, while the agency maintains full control over the data and AI stack. Ideal for balancing control, security, and scalability.
3. Running Local LLMs In-Region. Deploy state-specific language models within a secure, regional data center. Especially valuable for public safety, healthcare, and citizen services requiring tight control over data and latency. IDC estimates that 44% of government workloads require in-region processing.

Why Local Colocation Is the Right Fit

Colocation allows agencies to retain sovereignty over their data, avoid runaway costs, and keep focus on their core missions. With purpose-built facilities designed for compliance, security, and high-density workloads, colocation is emerging as the preferred AI infrastructure model for the public sector.

The benefits are clear:

• Data Sovereignty: Ensure sensitive data remains in-state.
• Predictable Cost: Transparent pricing with no surprise markups.
• Mission Focus: Offload IT infrastructure without compromising control.
• Built-in Security: Infrastructure aligned to government compliance standards.
• Local Impact: Support regional economic development through in-state hosting.

Use Case: South Carolina's AI Strategy - Leading with "Protect, Promote, Pursue"

In June 2024, South Carolina launched a comprehensive AI strategy for its state agencies, guided by three core pillars:

• Protect: Focus on governance, risk management, and data sovereignty
• Promote: Encourage innovation through shared infrastructure and collaboration
• Pursue: Invest in workforce development and long-term research

Key initiatives include:

• An AI Center of Excellence (launched in January 2025)
• 29 active use cases across agencies
• The ADAPT program supporting medical AI and workforce innovation
• Active R&D across USC, Clemson, and MUSC in healthcare, manufacturing, and energy
• Palmetto AI Pathways launching in 2025-26 to expand K-12 AI education

How DartPoints Supports South Carolina

DartPoints operates three strategically located data centers in Charleston, Columbia, and Greenville. Each site is designed for high-availability, AI-ready workloads, and serves state agencies, universities, and critical infrastructure providers across the region. Our proximity to policymakers and research institutions helps accelerate AI adoption safely and locally.

How We Can Help You: Let's Collaborate on Local AI

What AI initiatives are you exploring? Let's design secure, sovereign, and scalable infrastructure tailored to your public sector agency mission. This blog reflects insights from Matthew McKee's SCITDA 2025 presentation. To learn how DartPoints can support your public agency's AI strategy, connect with our AI infrastructure team today.