Executive Overview
As the global proliferation of artificial intelligence workloads forces data center operators into an unprecedented race for infrastructure capacity, a parallel revolution is unfolding behind the scenes. Operators are no longer just building facilities to house AI; they are actively deploying AI to run those facilities. From sub-second thermal optimization and power distribution management to predictive maintenance and automated incident response, artificial intelligence is rapidly transitioning from experimental pilots into mission-critical production systems.
While full facility autonomy remains a distant horizon, AI is delivering measurable, game-changing gains in energy efficiency, uptime, and operational velocity. Industry giants—including hyperscalers and top-tier colocation providers—are spearheading the movement with bespoke, in-house tools. Meanwhile, midsized operators are urgently piloting targeted applications to counter mounting operational pressures, while smaller legacy environments evaluate their initial steps into digital transformation.
This deep dive examines how leading industry players—specifically Digital Realty, AWS, DPR Construction, and Schneider Electric—are leveraging artificial intelligence today, and explores the technological convergence, architectural shifts, and trust barriers shaping the future of data center operations.
Detailed Chronology: The Evolution of AI in Facility Management
The integration of artificial intelligence into critical infrastructure did not happen overnight. It represents a steady, strategic evolution from basic automated scripting to sophisticated machine learning models and generative AI agents.
- Phase 1: Reactive Automation and Scripting (Pre-2015): Early facility management relied heavily on static rule-based software, building management systems (BMS), and basic data center infrastructure management (DCIM) tools. Systems could flag high temperatures or power spikes, but human intervention was entirely required to diagnose and resolve anomalies.
- Phase 2: Predictive Maintenance and Machine Learning (2015–2020): Forward-thinking operators began deploying foundational machine learning algorithms to analyze historical equipment logs. This era introduced early predictive analytics, allowing teams to anticipate chiller failures or battery degradation before catastrophic outages occurred. Companies like Digital Realty and Schneider Electric laid their foundational data telemetry pipelines during this period.
- Phase 3: Generative AI and Telemetry Integration (2020–2024): The advent of large language models (LLMs) and advanced neural networks shifted the paradigm. AI systems began digesting high-velocity telemetry streams in real time. Vendors introduced generative AI layers to assist human operators with natural language queries, root-cause analysis, and dynamic cooling adjustments.
- Phase 4: Agentic Workflows and IT/OT Convergence (Present Day): Today, the industry is entering the era of "agentic" AI. Autonomous and semi-autonomous software agents are no longer just observing; they are executing discrete tasks, coordinating cross-system workflows, and bridging the historical gap between Information Technology (IT) and Operational Technology (OT).
Industry Case Studies: How the Leaders Are Deploying AI Today
1. Digital Realty: OPDaaS, Cooling Optimization, and Measured Savings
As a global colocation leader operating more than 300 data centers worldwide, Digital Realty has utilized machine learning for over half a decade and generative AI for the past three to four years, according to CTO Chris Sharp.
At the heart of their strategy is the proprietary Operational Data as a Service (OPDaaS) platform. This curated suite of AI tools captures sub-second telemetry from power, cooling, and building infrastructure into a centralized data store. Exposing this information via APIs enables both internal and customer-facing tools to analyze and dynamically tweak environmental parameters.
- Proactive Maintenance: AI models can detect subtle operational variances, such as clogged air-cooled system filters that force fans to run at maximum capacity unnecessarily. Proactively addressing these components yields double-digit percentage efficiency gains.
- Liquid Cooling Integration: At its IAD51 facility in Northern Virginia—a site utilized by NVIDIA—Digital Realty deployed Phaidra’s AI platform. The system grants granular visibility into secondary liquid cooling loops, ensuring cooling matches actual compute demand rather than defaulting to full-capacity running.
- Tangible ROI: Deployed in over 10 facilities with plans to scale to 40 by the end of 2026, OPDaaS has fueled remarkable portfolio growth. According to the company’s 2025 Impact Report, Digital Realty expanded its portfolio by 34% while increasing water usage by a mere 3%. Sharp credits this efficiency directly to AI monitoring, noting the company saved an impressive 17,800 MWh of energy in 2025.
2. AWS: AI Agents for Network Operations and Stranded Power Reduction
Amazon Web Services (AWS) leverages generative AI-powered software to optimize server rack placement, directly combating "stranded power"—usable electrical capacity that goes unutilized due to thermal or spatial inefficiencies. Furthermore, AWS utilizes agentic workflows to manage its massive global network infrastructure at scale.
According to Zak Islam, AWS’s director of network observability and automation, network automation previously hit roadblocks when complex anomalies defied rule-based scripts, forcing costly escalations to human engineers. Today, AI agents autonomously investigate, collect, and correlate data across disparate monitoring sources.
- Rapid Root-Cause Analysis: When network incidents occur, AI agents cross-reference telemetry from multiple systems to isolate root causes in seconds rather than minutes, sharply reducing diagnostic times.
- Ticket Remediation: For routine operational tickets spanning compute, storage, and networking, AI evaluates incoming issues against historical patterns and frequently resolves them without human intervention, freeing engineers to focus on novel architectural challenges.
- Fiber and Buildout Oversight: AWS employs AI to continuously monitor fiber-optic infrastructure, automatically notifying vendors to accelerate repairs. During greenfield deployments, AI flags configuration errors and routes them to specialized installation teams.
3. DPR Construction: AI-Driven Pre-Construction and Site Robotics
Artificial intelligence is not limited to operational facilities; it is transforming how data centers are built. Santa Clara-based DPR Construction applies AI to streamline pre-construction planning and site execution.
- Predictive Pre-Construction: For the past three years, DPR has utilized AI data analytics during early planning phases. Drawing on multi-project benchmarking data—including site conditions, historical cost models, and construction sequencing—the company accurately forecasts staffing needs, project timelines, and sequencing strategies for new builds.
- Simulated Sequencing: Project teams run complex simulations mapping out foundations, structural steel erection, and MEP (mechanical, electrical, plumbing) systems. This allows builders to determine optimal construction paths, such as whether to build outward from a central core or linearly across a site.
- Autonomous Site Robotics: DPR pilots autonomous robots that traverse active job sites overnight. These machines capture high-definition imagery when construction personnel are off-site, giving clients unobstructed visual progress reports without interrupting daytime construction workflows.
4. Schneider Electric: Digital Twins, DCIM, and Governed AI at Scale
As a premier supplier of critical power and cooling infrastructure, Schneider Electric integrates AI into both its commercial offerings and internal corporate workflows.
- Digital Twins and Predictive Modeling: Operators leverage Schneider’s ETAP electrical power system modeling software and AVEVA industrial operations software—powered by NVIDIA technology—to build digital twins of critical infrastructure. These models allow facility managers to safely run stress-test simulations before executing physical changes.
- EcoStruxure IT and EcoCare: Through EcoStruxure IT, operators gain continuous infrastructure monitoring bolstered by AI-driven remote diagnostics and predictive maintenance recommendations. Its EcoCare service employs AI to analyze equipment faults, recommend precise replacement parts, and assign optimal technician urgency levels.
- Internal Operational Efficiency: On the enterprise side, Schneider Electric launched an agentic system to streamline customer quote requests. The platform slashed the quoting process from 62 distinct steps down to 14, accelerating turnaround times by 95%. All internal AI tools operate under a centrally governed cloud platform with rigorous oversight from the chief AI officer.
Supporting Context & Metrics
The quantitative impact of artificial intelligence on the data center ecosystem is becoming impossible to ignore. Industry data highlights several key trajectories:
- Energy Conservation: Leading operators are shaving thousands of megawatt-hours off annual consumption through targeted thermal and airflow adjustments. Digital Realty’s documented savings of 17,800 MWh in 2025 illustrate the massive financial and environmental stakes.
- Water Conservation Paradox: As hyperscale facilities grow denser to support high-performance liquid-cooled AI clusters, water usage efficiency (WUE) is under intense scrutiny. AI-driven cooling optimization is proving to be the primary defense against runaway water consumption.
- Operational Velocity: Administrative bottlenecks are shrinking. Enterprise deployments, such as Schneider Electric’s quotation automation, demonstrate operational speed-ups exceeding 90% when autonomous agents take over structured clerical workflows.
Official Statements & Industry Expert Insights
Industry analysts tracking the intersection of data center engineering and artificial intelligence emphasize that adoption remains a nuanced, evolutionary journey.
-
Roy Illsley, Chief Analyst for IT Operations at Omdia:
"AI is being used in discrete packages to do little bits of the work… We are going to move to a world where a lot of the specific tasks are done by AI agents, and then the humans will be connecting those tasks and overseeing decisions."
Illsley points out that while IT environments have matured in utilizing AIOps platforms to filter operational noise, Operational Technology (OT) environments are only now catching up via advanced DCIM integrations.
-
Chris Sharp, CTO at Digital Realty:
"That’s a huge delta that we couldn’t have achieved without generative AI being inside our systems, monitoring pumps, filters, and the full spectrum of infrastructure… You’ll start to see a little bit of those decisions being made with the oversight of a human. That’ll be our first step in that foray."
-
Zak Islam, Director of Network Observability and Automation at AWS:
"With the proliferation of LLMs, we are now able to use these systems at unprecedented scale across a wide range of technical and business problems."
-
Zoe Roth, Senior Research Analyst at 451 Research (S&P Global):
Highlighting the emergence of "agentic analytics" startups like Phaidra, Roth notes that modern orchestration layers are bridging building management systems and IT telemetry. By feeding reinforcement learning agents with cross-domain data, operators can maintain stable, ultra-efficient equipment temperatures.
Future Outlook: IT/OT Convergence and the Autonomy Barrier
The Convergence of IT and OT
Historically, Information Technology (servers, software, storage) and Operational Technology (power distribution, chillers, generators, pumps) operated in strict silos. IT engineers monitored applications; facilities engineers monitored mechanical plants.
The proliferation of high-density AI infrastructure is shattering this divide. Experts agree that the next major architectural milestone is establishing a shared data layer where IT and OT telemetry converge. When an AI agent can simultaneously view a surge in GPU computing loads and correlate it with real-time chiller performance, the facility can preemptively adjust cooling before thermal thresholds are crossed.
Will Data Centers Become Fully Autonomous?
Despite the staggering technical capabilities of modern artificial intelligence, analysts agree that fully autonomous, unstaffed data centers will remain rare for the foreseeable future. The barrier is not technological capability; it is trust and accountability.
- The Trust Deficit: As Dan Thompson, Research Director at 451 Research, observes, average data center operators primarily want to be armed with high-grade intelligence and recommended courses of action rather than surrendering total control.
- Legal Accountability: As Roy Illsley bluntly notes, human engineers remain legally and professionally accountable for facility uptime:
"You are responsible and accountable for that data center, not the AI agent that’s flying around at 5 million miles an hour."
- Remote Frontiers: The calculus shifts when facilities are deployed in geographically isolated environments where human intervention is physically impractical—such as the North Slope of Alaska or proposed space-based data centers. In these extreme edge scenarios, true autonomy or high-latency remote agentic control becomes a operational necessity rather than a luxury.
Conclusion
Artificial intelligence has permanently altered the DNA of data center operations. What began as reactive dashboard alerts has matured into proactive, self-optimizing ecosystems driven by machine learning, generative models, and agentic workflows. While the human element remains safely anchored at the helm of decision-making, the modern data center can no longer survive—let alone scale—without artificial intelligence keeping its pulse.
