AI overload drives data center Innovation

AI is overwhelming today’s data centers, but a comprehensive stack of infrastructure, energy, and software solutions is emerging to meet this challenge. 

The AI infrastructure crisis 

JLL’s 2026 Global Data Center Outlook report projects nearly 100 GW of new capacity will be built between 2026 and 2030, with AI workloads accounting for roughly half of all usage by decade’s end. We anticipate that demand will still outpace this supply. 

This transformation creates four critical pressure points: (1) power availability amid grid constraints, (2) cooling requirements for high-density GPUs, (3) construction considerations to build more efficiently and sustainably, and (4) operational efficiency to manage sprawling and distributed campuses. However, the same forces driving this overload are catalyzing innovation that JLL and JLL Spark are helping bring to market. 

1. Power solutions: From grid-constrained to power-abundant 

With primary markets facing grid connection wait times exceeding four years, operators are increasingly adopting “bring your own power” models and behind-the-meter generation: 

It is becoming critical to leverage renewable energy in fulfilling hyperscalers’ power needs, with long-term Power Purchasing Agreements (PPAs), co-located solar and wind, and battery storage driving increased capacity. 

Microgrids and alternative generation address resilience needs through natural-gas-backed microgrids, fuel cells, and emerging small modular reactors in markets where grid scaling lags demand. 

New-age batteries and underlying software enable dynamic load shifting, including timing intensive AI training runs to times of the day with cleaner, more available grid energy. 

This shift is transforming site selection from simple ‘fiber and land’ decisions to integrated power-portfolio strategies. 

2. Cooling the AI factory 

While cooling accounts for up to 40% of facility energy use, traditional air cooling is hitting physical limits at AI rack densities. The ecosystem is converging around layered cooling solutions: 

Advanced liquid cooling systems improve Power Usage Effectiveness (PUE), cutting cooling energy use by up to one-third. 

Hybrid cooling systems and rear-door cheat exchanges enable conversion of existing assets with traditional air-cooling systems. 

AI-driven cooling optimization uses Data Center Infrastructure Management (DCIM) and AI-native platforms to dynamically adjust airflow, liquid flow, and setpoints based on real-time workload distribution. 

3. Building faster and smarter 

AI demand is reshaping physical data center design and delivery through three key themes: 

Modular and prefabricated infrastructure cuts construction timelines by 30–50% and enables phased capital deployment, crucial where speed-to-power is paramount. 

Edge and distributed capacity support low-latency inference applications through micro data centers and edge colocation, as inference overtakes training. 

Sustainable construction incorporating low-carbon materials, adaptive reuse of industrial assets, and integrated waterless cooling creates competitive differentiation, as regulatory and community scrutiny increases. 

4. Leveraging AI for operational efficiency 

Next-generation management solutions address increased operational complexity of larger-format and sprawling campuses, while managing costs 

AI-driven infrastructure management platforms combine DCIM, telemetry, and machine learning to automate energy optimization, predictive maintenance, and capacity planning. 

Workload-aware optimization software understands GPU cluster behavior to consolidate workloads, tune performance, and route jobs to the most efficient locations. 

Edge fleet orchestration tools manage thousands of distributed sites through remote monitoring and autonomous operations without linear headcount growth. 

Bringing the community along 

While these technologies are critical to supporting this growth, the industry’s ability to meet capacity goals also depends on public sentiment and community acceptance, making early engagement essential to prevent project delays and blockages. Solutions that facilitate transparent community engagement, environmental impact visualization, and stakeholder communication could significantly improve project success rates by addressing public concerns proactively. Today this process is managed directly by data center developers and project owners, using in-house teams; however, given the anticipated demand for new projects, this could be a great opportunity for startup innovation. 

JLL Spark’s strategic position 

The 2026 Global Data Center Outlook frames a $3+ trillion investment supercycle supporting nearly 100 GW of new capacity. JLL’s analysis highlights a $500 billion technology opportunity across power, cooling, infrastructure, and management, with a venture-scalable segment of roughly $50 billion. 

JLL Spark is actively looking for software platforms that create and drive grid resilience, cooling and power innovations that unlock higher densities and more sustainable power footprints, innovative ways to build/expand data centers, and solutions that enable higher-efficiency management of campuses. 

By pairing these startups with JLL’s global advisory, development, and facilities management capabilities, we help clients transition from AI ‘overload’ to AI-ready infrastructure strategies, turning today’s constraints into tomorrow’s competitive advantages. 

Written by Carolyn Trickett and Ajey Kaushal, Growth and Investment Principals at JLL Spark.

Interested in a strategic partnership with JLL Spark? Apply for an investment here.