5 reasons the enterprise data center will never die

In 2019, Gartner analyst Dave Cappuccio issued the headline-grabbing prediction that by 2025, 80% of enterprises will have shut down their traditional data centers and moved everything to the cloud.

A lot has gone down since 2019, and Gartner’s latest guidance on the topic comes from John-David Lovelock, vice president analyst,who says, “It’s not as though the data center is going away. The enterprise data center is here to stay. There’s still enough spending by enterprises on servers, licensed software, and the skill sets they need to maintain and operate the environment that currently exists.”

The overall trend toward placing new enterprise workloads in the cloud remains firmly in place, Lovelock says, but it’s not so overwhelming as to cannibalize current levels of enterprise data center deployments. In fact, Gartner says that 2024 global enterprise spending on data centers was a healthy $66 billion.

Synergy Research comes to a similar conclusion. Six years ago, nearly 60% of data center capacity was on-premises; that’s down to 37% in 2024. By 2029, the hyperscalers will account for more than 60% of total data center capacity, while on-premises capacity will sink to only 20%. (Colocation facilities make up the remaining 20%.)

However, Synergy points out that these relative percentages exist in a rapidly expanding universe. The hyperscalers are building out new data center capacity as fast as they can to accommodate the explosion of interest in AI and gen AI. In terms of raw server and storage capacity, enterprise data centers will not be shrinking at all. They will remain relatively stable, says Synergy.

So, Cappuccio wasn’t totally wrong; cloud is growing fast, but on-prem isn’t declining as precipitously as anticipated. And to be fair to the now-retired Cappuccio, no one could have predicted game-changing events like a global pandemic in 2020 or the release of ChatGPT in 2022.

As we enter 2025, here are the key trends shaping enterprise data centers.

1. Many applications are better served on-premises

The notion that eventually all applications should be migrated to the cloud has not proved true. While it’s technically possible to rewrite and re-factor legacy applications for the cloud, such activity is not practical. Many legacy apps are running just fine where they are and there is no business need to disturb them. In addition, as security concerns increase, as data privacy regulations tighten, and requirements for control and visibility grow, enterprises are realizing that some applications should remain in the data center.

The Uptime Institute reports that in 2020, 58% of enterprise IT workloads were hosted in corporate data centers. In 2023, this percentage fell to 48%, and survey respondents forecasted that a stubborn 43% of workloads will still be hosted in corporate data centers in 2025.

“Does this mean that almost all IT workloads will — eventually — end up running in third-party data centers? This is unlikely,” says Max Smolaks, research analyst at Uptime.

2. Repatriation is on the rise

When the pandemic hit and employees fled to home offices, enterprises shifted applications to the cloud in a rapid, unplanned, somewhat chaotic fashion. Then the bills starting coming due.

Cloud repatriation — enterprises pulling applications back from the cloud to the data center — remains a popular option for a variety of reasons. According to a June 2024 IDC survey, about 80% of 2,250 IT decision-maker respondents “expected to see some level of repatriation of compute and storage resources in the next 12 months.”

IDC adds that the six-month period between September 2023 and March 2024 saw increased levels of repatriation plans “across both compute and storage resources for AI lifecycle, business apps (CRM, ERM, and SCM), infrastructure, and database workloads.”

IDC analyst Natalya Yezhkova notes a variety of factors behind the rise in cloud repatriation. In some cases, applications hastily “lifted and shifted” to the cloud during COVID were simply not built to performed efficiently in a cloud environment, from both a performance and a cost perspective. Other considerations include security, privacy, performance, management, and governance, she adds. 

“Repatriation is a good option to keep,” Yezhkova says. “CIOs should be reassessing whether the public cloud is delivering value, because the needs of workloads change, regulations around workloads change, offerings change whether in price or in functionality. So, organizations shouldn’t close the door to either option, public cloud or a dedicated environment.”

3. The return of private cloud as data centers modernize

The cloud may represent the cool new thing compared to the tired, old legacy data center, but data centers are changing with the times. Advances in server power and efficiency, new data storage techniques, liquid cooling, virtualization, containers, software-defined networking, and so on, enable organizations to modernize their data centers and make them more efficient — and more cloud-like.

In addition, companies such as HPE (GreenLake) and Dell (Apex) are offering private cloud implementations as a managed service inside the walls of the enterprise data center. Not to be outdone, AWS has a similar offering with its Outposts service, as do Microsoft and Google.

According to Forrester’s 2023 Infrastructure Cloud Survey, 79% of roughly 1,300 enterprise cloud decision-makers said their firms are implementing internal private clouds, which will use virtualization and private cloud management. Nearly a third (31%) of respondents said they are building internal private clouds using hybrid cloud management solutions such as software-defined storage and API-consistent hardware to make the private cloud more like the public cloud, Forrester adds.

4. The growth of edge computing

The proliferation of IoT devices has generated demand for processing power and data analytics capabilities as close as possible to where that data is created. According to IDC’s Worldwide Edge Spending Guide, global spending on edge computing was estimated to reach $228 billion in 2024, a 14% increase from 2023. The forecast anticipates strong growth through 2028, with spending expected to be near $378 billion, at a double-digit rate.

According to IDC, the edge serves as an intermediary between connected endpoints and the core IT environment. “Edge is a crucial technology infrastructure that extends and innovates on the capabilities found in core datacenters, whether enterprise- or service-provider-oriented,” says IDC.

The rise of edge computing shatters the binary “cloud-or-not-cloud” way of thinking about data centers and ushers in an “everything everywhere all at once” distributed model, where apps and data live in the most appropriate location, on a case-by-case basis.

5. AI generates momentum for on-prem and edge

Enterprises understand that they need to jump on the AI bandwagon to remain competitive, but the roadmap is not clear. Should CIOs leverage the virtually unlimited scalability and processing power of the cloud? Or should they keep AI workloads in-house out of concern over security, data privacy, regulations, latency, and other important factors. The answer: It depends.

Hyperscalers are racing to build out new GPU-based data centers, and each offers its own large language models (LLMs) and AI-as-a-service solutions.

But AI is also driving interest in on-prem private clouds.

“The excitement and related fears surrounding AI only reinforces the need for private clouds,” says Dave McCarthy, research vice president for cloud and edge services at IDC. “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.”

There’s an edge component as well, McCarthy adds.

“As the focus of AI shifts from training to inference, edge computing will be required to address the need for reduced latency and enhanced privacy,” he says. “This trend not only optimizes operation efficiencies but also fosters new business models that were previously not possible with centralized infrastructure. Distributing applications and data to edge locations enables faster decision-making with reduced network congestion.”

According to McKinsey, many companies are taking their first crack at AI with off-the-shelf models that are largely hosted in the cloud. But, the firm’s partners write, “As the technology matures, more enterprises are likely to build and train their own models on their internal data, which could lead to demand for private hosting.” As a result, McKinsey estimates that, by 2030, 60% to 65% percent of European and US AI workloads will be hosted on the cloud, with the remaining 35% to 40% on-prem.

Another wrinkle is the “AI-in-a-box” offering that enables CIOs to deploy turnkey, prebuilt AI packages. Nutanix, Nvidia, Microsoft, and others are creating GPU-based integrated systems designed specifically to run AI workloads in enterprise data centers.