The New Cloud Calculations: How AI is Reshaping Infrastructure Decision-Making

If you’ve been around the tech world for long enough, you’ll remember the dawn of the Information Age. It was a time of boundless enthusiasm: The birth of the internet felt like something truly new and exciting that was going to fundamentally change every aspect of our lives.

Now, we’re moving into a new era, one defined by knowledge, not just information. We call this era the Intelligent Age. If the previous age was about democratizing access to information, this new one is all about the things we can accomplish with that information by processing and analyzing it. As a result, IT is now moving from being a cost center to being a true driver of business value and differentiation for organizations.

In the past few years, we’ve seen new generative AI models that promise to extract more value from business data. This means it’s now more important than ever for enterprises to build an AI-ready data pipeline. In particular, organizations need an interconnected hybrid multicloud architecture to optimize performance, privacy and cost-efficiency in their AI strategies. This may require them to fundamentally rethink the way they use cloud services.

Navigating the AI hype cycle

In some ways, it feels like we’re in a hype cycle for AI, similar to the one we experienced during the early days of cloud computing. Like cloud, AI is a very powerful technology that’s going to create a lot of business value in the long run. However, it’s still important for CIOs to proceed with caution. If they’re too desperate to chase the hype, it could lead them to rush into poor infrastructure decisions that will cause a lot of headaches down the road.

I have a personal motto that defines the way I see the enterprise IT space, and it’s more pertinent now than ever before: Data is forever.

The lasting nature of data is a fundamental truth of the internet, and of human society in general. Data, once captured, will always continue to have value. No matter what line of business they’re in, enterprises need to capitalize on that value in order to thrive.

However, we must remember that data may be forever, but what we do with it is not forever. Who owns it, how they extract value from it and what infrastructure they use to manage it will all change over time. When it comes to modern AI applications, we’re still in the early days. We can safely assume that we’ll see more powerful hardware, better algorithms and more efficient infrastructure in the years to come, which is why it’s not ideal for businesses to lock themselves into a certain way of doing things. What works well today may no longer work 10 years down the road. It’s not enough to say that you need flexibility in your AI strategy; flexibility is the strategy.

Making the AI pivot

According to IDC, many businesses are currently making the AI pivot. While they may have spent 2023 and 2024 scrambling to pursue generative AI use cases in an ad hoc manner, they’re now focused on setting priorities, addressing barriers and planning for long-term AI success.[1]

As the graphic below shows, most businesses are still early in their AI maturity, and very few have progressed all the way to become true AI-fueled organizations.[2] This demonstrates that despite the hype, there’s no pressure to get AI right immediately. What matters is having infrastructure that’s able to change as you progress through the phases of your AI adoption.

Planning for cloud flexibility

Cloud services have traditionally been considered perfect for any use case that requires infrastructure flexibility. Businesses can stand up cloud infrastructure quickly, while removing the cost and complexity of deploying physical infrastructure. As they pursue AI, IT leaders will need to add compute capacity very quickly. They often see the cloud as a cost-effective way to do that. They may find it especially helpful to work with a GPU as a Service provider to deploy virtual GPU capacity on demand, instead of going through the challenges of buying and deploying physical GPUs for themselves.

However, the view that cloud is fundamentally flexible has changed. Enterprises now understand the risks of vendor lock-in, including higher cloud costs and limited access to best-of-breed solutions. They need to be proactive about ensuring flexibility in their hybrid multicloud environments. This is particularly true for AI, in light of everything they’ll need their AI infrastructure to do:

• Manage large volumes of AI data from many different sources
• Move that data quickly to different processing locations worldwide
• Be aware of their data sovereignty and privacy requirements
• Keep an eye on costs and energy consumption, which could easily spin out of control as data volumes scale up

The key to meeting all these requirements is to work with many clouds, not any one particular cloud. We live in a time when many cloud services have become commoditized. Individual providers may have little things they specialize in, but for the most part, it’s hard to differentiate between their basic offerings of compute, networking and storage. Instead of competing to provide the best capabilities, cloud providers are now looking to maximize the value they provide for customers.

One way they’re working to optimize value is by making it easier for customers to incorporate their services into a hybrid multicloud strategy. It may seem counterintuitive that they’re essentially encouraging customers to diversify their cloud portfolios, but enterprises have come to expect nothing less.

Capitalize on a neutral platform for hybrid multicloud

The idea of flexible hybrid multicloud infrastructure is nothing new to us at Equinix. For more than 25 years, our commitment to vendor neutrality has been one of our guiding principles. It started during the early internet, when we provided a neutral platform for internet service providers to exchange traffic with one another. It remained through the advent of cloud computing and up until today, when we continue to partner with all major cloud providers.

As businesses think about how to incorporate cloud services into their AI strategies, we don’t point them in any particular direction. Instead, we give them the platform on which to find the right mix of cloud services for themselves. We also help them pair those cloud services with private infrastructure in any of our 260+ Equinix IBX® colocation data centers worldwide.

To further optimize flexibility, we give our customers a robust portfolio of interconnection services they can use to connect with cloud providers or any other partners in their AI ecosystems. With Equinix Fabric®, our software-defined interconnection solution, they can create, remove or adapt virtual connections whenever the need arises. Using Equinix Fabric, our customers can access our market-leading portfolio of low-latency cloud on-ramps from many locations throughout the world.

The AI era can be scary for businesses that don’t know where to start. The most valuable thing we offer our customers is the confidence that comes from knowing their AI strategy doesn’t have to be perfect. They have the flexibility to change their minds at any time, because they’re not locked into any particular vendor. They can decide on a new solution on Monday and have it up and running by Tuesday, thanks to our ecosystem of cloud providers and on-demand connections with Equinix Fabric.

As the AI landscape continues to alter the way enterprises approach cloud services, we’ll be there to help them navigate the changes. To learn more about how perspectives are shifting in this new AI era and how businesses that optimize for cloud flexibility are positioning themselves for long-term success, read our white paper Thriving with a hybrid multicloud strategy.

[1] “IDC FutureScape: The AI Pivot Towards Becoming an AI-Fueled Business,” October 21, 2024.

[2] IDC Directions 2025, “Time for the AI Pivot: From Experimentation to Industry Transformation,” April 2025.