Artificial intelligence has completely changed the conversation around data centers. Just a few years ago, talking about infrastructure meant focusing on availability, redundancy, or certifications. Today, the questions are different: Can it support racks of over 100 kW? Is it prepared for liquid cooling? Will it be able to power thousands of GPUs within three years?
However, there is one idea that’s worth clarifying. The advent of AI has not rendered the standards used for decades to evaluate data centers obsolete. What is changing is how they are applied.
Certifications remain the starting point for measuring infrastructure quality. What has changed is that they now must coexist with new technical requirements derived from artificial intelligence.
The keys to AI-ready data centers in 20 seconds
- AI is increasing the demands for power, cooling, and scalability in data centers.
- Traditional certifications continue to be essential for assessing infrastructure quality.
- The difference now is that capacity for adaptation to future hardware generations is also evaluated.
- Flexibility, energy efficiency, and operational maturity carry more weight in purchasing decisions.
- The data center is no longer just a building but a platform designed to evolve.
Large AI deployments have not only changed the type of servers being installed. They are also transforming the way entire buildings are designed, energy strategies, and the criteria companies use to select cloud providers.
Certifications remain the foundation
When a company searches for an infrastructure provider, certifications continue to be among the first evaluation criteria.
Standards such as ISO 27001 for information security, ISO 22301 for business continuity, ISO 50001 for energy management, or national certifications like the National Security Scheme (ENS) still provide an objective guarantee of how an organization operates.
Similarly, international standards for data center design and availability remain benchmarks for assessing resilience, redundancy, and construction quality.
None of that disappears with AI.
On the contrary.
The more critical the infrastructure, the more important it is to have documented processes, external audits, and consistent operational procedures.
AI is changing the questions
If a few years ago it was enough to ask how many megawatts a data center had, now that figure alone says very little.
Infrastructure managers are starting to ask different questions:
- Can it grow without rebuilding the building?
- Is it ready for liquid cooling?
- Can it combine different cooling technologies?
- Is there reserved electrical capacity for future expansions?
- Will the network support thousands of GPUs running simultaneously?
AI has shifted the focus from installed capacity to the capacity to evolve.
It’s not just about supporting current hardware, but about preparing infrastructure for multiple future generations.
Cooling is no longer just an auxiliary system
Few areas better reflect this transformation than climate control.
For decades, nearly all enterprise data centers operated via air cooling.
That model remains perfectly valid for many traditional workloads.
But the latest-generation GPUs have raised energy densities to levels that require rethinking many facilities.
Liquid cooling is no longer just a specialized technology.
It is beginning to become part of strategic planning for numerous AI projects.
The issue is no longer just dissipating heat.
It’s also about how to scale the system, maintain it over years, and adapt it to new accelerators without redesigning the entire building.
Energy becomes a strategic factor
AI is also changing electrical planning.
Traditionally, available power at a given time was analyzed.
Now organizations also inquire about future capacity.
Large campuses are beginning to be designed with progressive expansions, new electrical connections, and land reserves to keep growing over the next decade.
Having power available is not enough.
It’s necessary to demonstrate the ability to continue increasing capacity as customer needs evolve.
Daily operations matter as much as infrastructure
A data center can have the best equipment on the market and still deliver poor service if daily operations are not sufficiently mature.
That’s why AI is emphasizing aspects such as:
- Preventive maintenance;
- Standardized procedures;
- Change management;
- Continuous monitoring;
- Technical staff training;
- Incident response.
Certifications are precisely helpful in demonstrating operational maturity.
They evaluate not just the building, but also how an organization manages infrastructure throughout its entire lifecycle.
Sustainability is no longer measured solely by PUE
For many years, the Power Usage Effectiveness (PUE) was the main indicator of a data center’s efficiency.
Today, it remains important, but it’s not enough anymore.
AI requires considering other parameters:
- Water consumption;
- Heat reuse;
- Hardware lifespan;
- Use of low-carbon energy sources;
- Overall campus efficiency;
- Capacity to grow without proportionally increasing environmental impact.
Sustainability is now viewed as a long-term strategy, not just an energy figure.
Flexibility becomes a new standard
Perhaps the greatest change introduced by AI is the level of uncertainty.
No one can guarantee the densities servers will have in five years or what type of accelerators will dominate the market.
That’s why new data centers are designed with less focus on current hardware and more on adaptability.
Modular infrastructures, scalable electrical distribution, hybrid cooling, and campuses prepared to grow are some features increasingly common in modern projects.
They do not yet constitute a specific certification.
But they are becoming increasingly influential in procurement processes.
What companies should ask today
Organizations seeking infrastructure for AI projects should no longer limit themselves to requesting a list of certifications.
It’s also advisable to consider questions such as:
| Aspect | Recommended question |
|---|---|
| Power | Is there capacity for future expansion? |
| Cooling | Can liquid cooling be incorporated when needed? |
| Scalability | Is it designed to grow without service interruptions? |
| Operation | What certifications support your processes? |
| Security | Does it comply with international standards and national requirements like ENS? |
| Sustainability | How do you manage energy efficiency and future growth? |
Certifications remain essential.
What’s different now is that they are part of a much broader evaluation.
AI does not replace standards; it tests them
Every technological revolution prompts a review of infrastructure construction methods.
AI does not eliminate decades of accumulated experience in the data center sector.
Quite the opposite.
It proves that principles like resilience, security, business continuity, and operational excellence remain just as important.
The difference is that they now have to coexist with new challenges: denser racks, advanced cooling, rising power consumption, and hardware evolution that demands more flexible infrastructure design.
Data centers that best respond to the next decade will not necessarily be the ones with the most experimental technology.
They will be those capable of combining established standards, mature operation, and infrastructure prepared to evolve at the pace of AI.
Frequently Asked Questions
Does AI make current data center standards obsolete?
No. Certifications continue to be fundamental in evaluating security, continuity, operational quality, and resilience. What’s changing is that they are now complemented by new criteria related to AI.
Why is liquid cooling gaining importance?
Because AI GPUs generate thermal densities far exceeding those of traditional servers, making air cooling insufficient in certain deployments.
What are companies looking for when choosing a data center today?
Beyond certifications and availability, they value growth potential, readiness for future hardware generations, energy efficiency, and operational maturity.
What role does flexibility play in new data centers?
It has become a strategic factor. Organizations want infrastructure capable of adapting to new technologies without needing to redesign the entire building.
