Good news, everyone! The organizations behind some of the data center industry’s most influential technical guidance have published a new framework to help operators navigate the challenges posed by AI workloads.
The new AI Data Center Framework for Efficiency, Resilience, Safety, and Performance has been released by the American Society of Heating, Refrigerating, and Air-Conditioning Engineers (ASHRAE) in collaboration with Pacific Northwest National Laboratory (PNNL) and the National Electrical Manufacturers Association (NEMA).
Tell me more about ASHRAE…
ASHRAE was formed by the merger in 1959 of the American Society of Heating and Air-Conditioning Engineers (ASHAE), founded in 1894, and the American Society of Refrigerating Engineers (ASRE), founded in 1904.
It is now an international society of more than 50,000 heating, refrigerating, and air-conditioning professionals from over 132 nations dedicated to serving humanity and promoting a sustainable world. Now you know!
Available free through ASHRAE, the framework provides guidance for designing, operating, and upgrading data centers that support increasingly-demanding AI and accelerated computing environments.
For Cloud service providers, hosting companies, colocation operators, and infrastructure leaders, the publication arrives at a critical moment.
As AI deployments accelerate, organizations across the internet infrastructure industry are confronting a new set of operational challenges around power, cooling, reliability, sustainability, and cost management.
Facing the challenge of AI infrastructure
AI is changing the economics and engineering requirements of data centers.
The infrastructure supporting large language models, AI training, inference workloads, and accelerated computing often requires:
- Higher rack densities
- Greater power consumption
- Advanced cooling technologies
- More sophisticated thermal management
- Increased focus on reliability and uptime
- Careful management of energy and water resources
Many facilities currently serving Cloud, hosting, SaaS, and internet infrastructure workloads were not originally designed with these requirements in mind.
As a result, operators are increasingly evaluating how existing facilities can be adapted as they plan for the next generation of AI-ready infrastructure.
The new framework is intended to help organizations make those decisions correctly, the first time.
What is included?
According to ASHRAE, the framework covers the entire data center lifecycle.
Rather than focusing solely on new builds, it includes guidance for both new and existing facilities.
Areas covered include:
- Planning and design
- Commissioning
- Facility retrofits
- Day-to-day operations
- Performance optimization
- Reliability management
- Thermal management strategies
- Energy efficiency
- Water use
- Integrated system performance
The framework also recognizes that there is no single blueprint for AI infrastructure.
Recommendations are designed to account for:
- Different climate conditions
- Varying workload densities
- Different operating environments
- Regional infrastructure constraints
Cooling takes center stage
Cooling has emerged as one of the defining challenges of the AI era, so—perhaps unsurprisingly—thermal management sits at the heart of the framework. As compute densities increase, operators are exploring a range of approaches including:
- Advanced air cooling
- Direct-to-chip liquid cooling
- Immersion cooling
- Hybrid cooling architectures
There are even underwater data centers, such as this one in China.
The framework emphasizes that cooling can no longer be viewed as an isolated facility function. Instead, thermal management must be considered as part of a broader strategy that balances:
- Performance
- Energy efficiency
- Reliability
- Sustainability
- Operating costs
Anyone who has attended CloudFest will already know how important each of these components is, and how their importance is growing.
A focus on operational efficiency
The framework’s publication also reflects growing concerns about energy consumption across the data center industry. As AI drives unprecedented demand for computing capacity, placing additional pressure on power grids and energy infrastructure worldwide.
For Cloud providers, efficiency is becoming more than an environmental objective. It is increasingly a key differentiator and contractual requirement.
Improving efficiency can help operators:
- Reduce operating costs
- Delay expensive infrastructure upgrades
- Support sustainability goals
- Improve facility utilization
- Increase competitiveness
The framework encourages operators to evaluate performance across the entire facility rather than optimizing individual systems in isolation.
Reliability remains a priority
While AI is stoking excitement across the industry, uptime is still the most important metric for many customers. The framework places significant emphasis on resilience and operational continuity.
ASHRAE’s recommendations are designed to help operators maintain high availability while adapting facilities to support increasingly demanding workloads.
This is particularly relevant as AI applications move from experimental environments into production systems that businesses depend on every day.
What it means for Cloud providers
The release highlights a broader shift taking place across the infrastructure sector. The conversation is no longer simply about deploying more GPUs. Instead, operators are increasingly focused on the practical realities of supporting AI at scale.
For Cloud service providers and internet infrastructure companies, key questions include:
- Can existing facilities support future AI workloads?
- What cooling technologies should be adopted?
- How can energy consumption be managed effectively?
- What upgrades are required to maintain reliability?
- How should future facilities be designed?
The new framework will not answer every question. However, it does provide a common reference point for an industry that is moving rapidly and learning in real time.
Subscribe to the CloudFest newsletter for the latest industry insights, plus free quarterly business intelligence reports and a free registration code for CloudFest 2027.
