The data center industry is undergoing a new set of changes. The main reason for this is the slew of new technologies that are ramping up and taking center stage in the world of IT. As a result, data centers are adapting accordingly. For a very long time data centers were associated with the cloud and related services. Now this is changing, and new technologies and approaches are becoming the main usage for these facilities.
Latest research also shows a change in tides. The cloud is no longer the king of the data center. In July 2023, Dell’Oro Group predicted that the artificial intelligence (AI) infrastructure investments will raise data center capital expenditures to over $500 billion by 2027. But data center operators shouldn’t pop the champagne just yet. The research also says that during the same time, the cloud and enterprise capex growth will decelerate, and the market will go through “digestion”.
In short, cloud services seem to be getting close to market saturation and will be plateauing in the near future. Luckily, the IT industry has a set of new favorites which will continue to require big resources and will drive data center growth. The two main new favorites are hyperscale and colocation. Let’s dive into them a bit more.
What’s a hyperscale data center?
Hyperscale data centers are exactly what they say in the name – facilities which are huge in size. They have a massive footprint and can house many servers and a lot of equipment. A key feature is they are able to scale up quickly to meet the constant demand increase. While there aren’t any official rules to define what makes a data center a hyperscale data center, the industry generally accepts that any such facility with more than 5,000 servers and more than 1,000 square meters of area is hyperscale.
In reality though there are buildings with tens and even hundreds of thousands of servers. Some could even go beyond a million, the Enterprise Storage Forum notes. Why would someone need such a massive data center? Actually a lot of reasons. The main one is to be able to provide “unlimited” storage and processing power to the services they host. Often hyperscalers are used by IT giants like Amazon, Google, Meta, Microsoft, etc., to host their own services and to also provide their public cloud services, too.
Obviously hyperscale data centers have one massive drawback – they are very expensive. That’s why only very big companies are able to afford them. The other option is to build one and sell services from it to third parties as a way to share the costs.
The way hyperscale data centers are set up could vary depending on the goal. Mostly the servers are divided in clusters for a specific workload. There’s a lot of additional equipment and software that goes in to maintaining and organizing said work. Especially considering that most of these clusters also have physical servers which are running a bunch of virtual servers. That’s a lot of servers.
What is a colocation data center?
Have you ever wished you had your own data center for your company? Since you are reading this, chances are you have. Or at the very least have thought about the benefits of using a data center for your digital needs. Well, this is where colocation data centers come in. They allow companies of all sizes to simply rent some of the physical space in the data center and send their own servers there.
This way they get to use their own hardware with all of the benefits of a data center – better overall security, better connections, proper maintenance, etc. It’s a great way to share costs of a data center and still have the freedom of using private hardware that you don’t share with other users.
What usually happens is that a data center could be multiple types at once. It’s often that you’ll see a data center offer both “regular” services like shared hosting and cloud, but also space for colocation. Neterra’s Sofia Data Center (SDC) facilities also offer colocation services for clients. You can even separate your hardware physically from the rest and add additional security with locking cabinets and even cages. All in all, some data centers are entirely for colocation, but it’s possible to find colocation services within a data center, even a hyperscale one.
Why are hyperscale and colocation taking the place of the cloud
It comes down mainly because of the all-new technologies that are gaining traction. Yes, one of the main uses for hyperscale is in fact cloud services. But AI and machine learning, along with streaming are gaining popularity much faster. AI in particular is poised to be the next big technology and it will need a massive amount of hardware resources. So, if AI lives up to the expectations and becomes as ubiquitous as intended, it will demand a lot of hyperscale data centers to power it.
Colocation on the other hand also has its benefits. It’s a more economical way to build and maintain a data center. Enterprise Storage Forum calculates that data center colocation has a 37-52% lower cost than building a traditional raised-floor data center. It also offers more options for control and personalization which isn’t available at typical cloud data centers. And it’s a great way to set up backup systems for your business.
The infrastructure is changing
Endlessly building data centers is not the best way to go about it. There are limitations for power and land availability, rising construction costs and other related challenges that are making data center operators want to fully maximize every square centimeter of space they have.
All while growth in cloud spending is showing signs of slowing down. Synergy Research Group reports that in Q2, enterprise spending on cloud infrastructure services got close to $65 billion globally. That’s $10 billion more than Q2 of last year. It’s also the third consecutive quarter in which the market grew by $10 billion from the same time last year. That seems great, right? Well, the analysts also say that the actual year over year growth is going down. It was 20% in Q4 2022, then it was 19% in Q1 2023 and 18% in Q2 2023. So, the money is definitely more, but the rate at which it increases is slowing down.
The current economy “situation” is only part of the reason, says Synergy. “Above all else, the law of large numbers” is to blame. Synergy also says that Amazon, Google and Microsoft continue to hold 65% of the worldwide cloud computing market. Other key players remain Oracle, Snowflake, MongoDB, VMware, Huawei and China Telecom.
At the same time, Dell’Oro notes that there’s a significant jump in workloads from AI workloads. So, we are slowly heading towards a situation where the money “could” be slowing coming in, but increasingly going out as investments for more hardware resources will increase due to the AI demand.
A particular niche of interest would be Data Center Physical Infrastructure (DCPI), notes Dell’Oro. It forecasts that the DCPI market growth will accelerate at a 10% compound annual rate until 2027. The market will score a double digit growth in 2023, as well, due to the improving supply chain.
“I have been tracking the DCPI market for nearly a decade, observing annual revenue growth rates in the mid-to-low single digits. There is a change unfolding. The excitement Generative AI is creating at industry events and in my discussions with ecosystem vendors is palpable,” says Lucas Beran, Research Director at Dell’Oro Group in a release.
“I am already seeing AI workloads leading to a broad proliferation of accelerated computing infrastructure. This will require investments in next-generation data center physical infrastructure to support new architectures with higher power and thermal management requirements,” he adds.
AI loves colocation, too
Here’s another twist. According to some experts, AI-oriented colocation services are also going to be a popular niche. What would that be like?
It’s basically renting a portion of the data center to collocate your AI needs. “Businesses are increasingly turning to emerging technologies like AI to analyze and process this data more effectively. These new technologies consume large amounts of power, which require specialized infrastructure to support them, providing high-power densities, while safely and efficiently regulating the temperature to ensure performance is not impaired,” says Digital Realty, a data center operator.
The new service uses Air-Assisted Liquid Cooling solutions to provide said temperature regulation and the whole infrastructure is built with the idea of colocation and AI services.
“Undoubtedly AI, HPC and video streaming workloads are driving increased demand for capital-intensive, dense computing solutions. To effectively implement these solutions at scale, customers require facilities and services that not only support seamless deployment but also maximize capital utilization,” says Nicole Enright, President of Avnet Integrated Solutions to Data Center Frontier.
Analysts from IDC also seem to agree. They say that data consumption will increase at a rate of over 20% annually in the years to come. And enterprise applications in great numbers will adopt AI capabilities, driving that rise in data consumption up and up. As a result, colocation will become central to the digital ecosystem, says Courtney Munroe, research vice president of IDC.
All of this will further drive data center innovations, too. This means advancements in cooling, processing, storage, etc is on its way, and all of this will be thanks to AI and the increased need of colocation. The cloud is not going anywhere and will continue to have a big share in data centers, but the main drivers of innovation and growth will be AI and colocation.