The death of private cloud has been overstated. While many enterprises have pursued public cloud exclusively in recent years, some have learned the hard way that a simple lift-and-shift approach isn’t right for every workload. That’s why IDC predicts that spending on dedicated cloud infrastructure—a term that includes private cloud—will grow at 12.6% CAGR between 2023 and 2028, ultimately reaching $48.2 billion.[1]
Many enterprises see private cloud as a way to get the best of both worlds:
- The benefits of public cloud, including infrastructure agility and elasticity
- The control of private infrastructure, helping them avoid potential drawbacks such as security concerns and high costs
The potential for private cloud to help enterprises overcome challenges and implement advanced technologies is clear. But as we’ll explore in this post, you need to do private cloud the right way to get the full benefits it promises.
How private cloud can help in the AI era
The advent of generative AI is one factor that’s driving the resurgence of private cloud. Enterprises are learning that they can’t run AI exclusively on public infrastructure without putting themselves at risk of data leakage.
To get the full value of AI while keeping sensitive data protected, many enterprises are turning to a private AI approach. This means hosting models on private infrastructure—which could be a private cloud—and training them on proprietary datasets. By bringing models to their data, and not the other way around, enterprises can ensure private data stays inside their security domain.
The excitement and related fears surrounding AI only reinforces the need for private clouds. Enterprises need to ensure that private corporate data does not find itself inside a public AI model. CIOs are working through how to leverage the most of what LLMs can provide in the public cloud while retaining sensitive data in private clouds that they control.”— Dave McCarthy, IDC, from an interview with CIO.com [2]
One reason we’ve heard “private cloud is dead” so many times is that some businesses blame the technology when they don’t get good results, rather than thinking about how they could have deployed it differently. Many enterprises fail with private cloud because they repeat the same mistakes they make in their on-premises data centers, instead of thinking like a cloud provider.
How to do private cloud right
Compute cycles are a fungible resource. In theory, data centers should automatically push workloads to whichever servers are available to support them. However, we all know this typically doesn’t happen in enterprise data centers.
Instead, budgets are divided up among line-of-business teams that only care about their own SLAs. Compute power is set aside in buckets, each with enough capacity to support applications at peak demand. This is despite the fact that it’s highly unlikely the enterprise would ever need peak capacity for all applications at the same time.
Therefore, the utilization rate of most enterprise data centers is very low. Running a data center this way is like lighting money on fire. Enterprises spend millions on building and maintaining data centers, without considering that they’re paying for servers to sit idle much of the time.
Succeeding with private cloud requires businesses to change this fundamental problem: transitioning from siloed compute resources and budgets to holistic infrastructure that works for the entire organization. In short, it means thinking more like a cloud provider. It’s easy to see the benefits of investing in public cloud, because those services are fungible by default. An external service provider isn’t subject to internal complexity around budgets and SLAs.
To do private cloud right, enterprise IT teams must recognize that they too are a service provider. It’s their job to deliver the best results for line-of-business teams—all of them, not just the ones with the biggest budgets. If they can’t make this fundamental shift in mindset, then their private cloud will become just one more siloed compute cluster.
A more efficient way to do private cloud
Shifting to a service-provider mentality is a dramatic change that won’t happen overnight. In fact, some IT leaders might not have the appetite for this kind of change at all. In these cases, deploying in a hosted or managed environment such as a dedicated bare metal server could allow for a faster, more efficient way to get started with private cloud, while avoiding the problems commonly found in on-premises data centers.
If you break it down, what enterprises want from private cloud is actually quite simple: They want the elasticity and agility of the public cloud, but they also want greater control over security and costs. Once they’ve identified these priorities, it’s easy to see how deploying a hosted private cloud with a digital infrastructure partner like Equinix can help enterprise IT teams achieve them—without having to make a fundamental shift in how they operate.
The value of a hosted private cloud is especially clear given new requirements around AI. To support AI workloads on-premises, enterprises would need powerful compute hardware; in particular, they’d need to overcome the challenges of acquiring GPUs and utilizing them to their full potential. They’d also need to invest in energy and cooling innovations—such as liquid cooling—to support that hardware. Trying to meet all these requirements while shoehorning workloads into individual compute clusters with their own SLAs would make things especially difficult.
Enterprises are in a bit of a pickle with this. Security concerns are what is driving them to private cloud, but the specialized hardware required to do large-scale AI is expensive and requires extensive power and cooling. This is a problem that companies like Equinix believe they can help solve, by allowing enterprises to build a private cloud in Equinix datacenters that are already equipped to handle this type of infrastructure.”— Dave McCarthy, IDC, from an interview with CIO.com [3]
In addition to colocation services in our Equinix IBX® data centers, Equinix customers can also choose to deploy Equinix Metal® for dedicated bare metal servers. With Equinix Metal, enterprises can build the cloud environment they need to succeed with AI and other emerging technologies, while maintaining control over their security and costs.
In addition, customers can take advantage of private cloud solutions from partners they know and trust. For instance, VMware Cloud Foundation on Equinix Metal and Nutanix Cloud Platform on Equinix Metal both support enterprise-grade use cases for unique workloads. Furthermore, Equinix Managed Private Cloud and cloud adjacent storage strengthen our portfolio of hybrid offerings for customers looking for managed solutions.
To learn more about how you can make infrastructure your competitive advantage with Equinix Metal, read the data sheet: Fast, flexible, secure infrastructure when you need it.
[I] IDC Press Release, Shared Cloud Infrastructure Spending Continues to Accelerate, Fueled by AI-Related Spending in the First Quarter of 2024, According to IDC, June 28, 2024.
[2] Paula Rooney, Private cloud makes its comeback, thanks to AI, CIO, May 14, 2024.
[3] Ibid.