Business
Data Center Market Size and Outlook: Key Trends Through 2030
Today, it seems almost impossible to imagine a world without data centers. These gigantic high-tech buildings host IT companies and their software, including large language models and artificial intelligence, providing the environment necessary for their operation.
While data centers face criticism for their high energy and water consumption and environmental impact, they are at the core of the modern digital infrastructure, drawing billions - and even trillions - of dollars in investment.
- From now until the end of the decade, new data center capacity is anticipated to go up by around 97 gigawatts, hitting a total of around 200 gigawatts by 2030.
- The growth, according to experts, is attributed to the AI sector, which could be responsible for around 50% of all data center workloads.
- Currently, we are witnessing an infrastructure investment within the AI sector that surpasses any previous estimates, requiring up to US$3 trillion by 2030.
The graph below illustrates the global forecast by region in the capacity of data centers between 2025 and 2030:
So how big is this market, exactly? It depends on who you ask.
Market size: what the analysts say
If we take a look at the data center market information provided by research firms, each will have a different figure. This shouldn’t be a surprise, considering how fragmented the market is.
But let’s lay the numbers from several reputable data providers side by side to get a clearer picture.
2025 market size | 2026 projection | 2030-2035 projection | Compound Annual Growth Rate (CAGR) | |
$269B | $300B | $699B | 11.1% | |
$386B | $430B | $1.1T | 11.06% | |
$383B | $425B | $902B | 11.3% |
Although Fortune Business Insights provides lower estimates than Precedence Research and Grand View Research, all three seem to agree on a stable growth rate of approximately 11% per year.
The main driver of the boom seems to be the rapid growth of investor interest in artificial intelligence, which requires tremendous space for hosting in data centers.
What's driving the boom: AI infrastructure buildout
If we turn to the spending side, according to BloombergNEF, the 14 largest public data center operators are on pace for US$750 billion in capital expenditures (CAPEX) in 2026, which is a dramatic increase from the previous year, when the figure stood at around US$450 billion.
From the same research provider, we learn that 23.1GW of data center IT capacity is currently under construction around the globe, with 15.9GW of that being in the U.S.
In 2026, the overall capex of the nine largest cloud service providers (CSP) could surpass US$886.7 billion, say experts from TrendForce. These nine CSPs include the US-based tech giants Google, Amazon, Meta, Microsoft, Oracle, and China-based tech titans ByteDance, Tencent, Alibaba, and Baidu.
In the end, we have the following picture: investors are putting money into AI much faster than the physical structures and power systems can be erected.
That’s why people in the industry stopped talking solely about the number of servers required and started thinking about the amount of power needed.
Data center types, segments, and operations
According to Cisco (one of the world's leading information technology and networking companies, according to Investopedia.com), there are six types of data centers, and each type differs based on who the facility’s de facto owner is, who operates it on a regular basis, and where it is located (i.e., how close to those who use it).
1. Enterprise data centers
These are owned and run by one company, and usually on its own campus, serving internal users only.
2. Managed services data centers
Company A rents (leases) the computers, servers, storage, and building from company B (called a managed services provider).
Company B takes care of everything (keeping equipment running, fixing issues, updating software, and managing security) so company A can just use the data center without owning or managing it itself.
3. Colocation data centers
This is like renting a ready-made room in someone else’s secure building for your computer equipment. So, Company A rents space (cabinets, racks, etc.) within Company B’s data center. The latter provides the building itself, cooling, power, internet bandwidth, security, etc.
4. Cloud data centers
Company A rents space and power from Company B - a big cloud provider (it doesn’t keep its own servers). Cloud data centers are gigantic computer facilities owned and managed by another company (a cloud company).
This means that all of Company A’s information and applications “live” on someone else’s computers, and usually it’s the hyperscalers like Amazon (AWS), Google (GCP), Microsoft (Azure), or IBM.
5. Edge data centers
Facilities like these sit at the other end of the spectrum, being small and found near the people or devices that use them.
They handle information right where it’s created (on phones, cars, etc.) without having to transmit the data to a server hundreds or thousands of miles away. Thus, you benefit from faster response (lower latency). It’s useful for smart devices and sensors (IoT), streaming videos or games, and more.
Finally, we have a category that today drives most of the capex numbers:
6. AI data centers
From the name, you can guess that this is a type of data center developed specifically for the work that AI does - tasks that are heavy and power-hungry. So these data centers’ goal is to process gigantic amounts of data in a fraction of the time, and to do that, they need advanced infrastructure.
The latter encompasses high-performance GPUs (the same kind of chips that power graphics in games but used here for AI math) and TPUs (chips made especially for AI tasks). In addition, they use strong built-in security.
Here, speed matters a lot, and that’s because many AI applications need answers almost instantly, for example, predicting something in real time or understanding and responding to language right away.
Suggested reading: The Future of China's Semiconductor Industry: 2030 Market Predictions
Data centers’ power and capacity constraints
You can invest huge sums of money in locations, designs, and the like, but you can’t change the underlying physics, and this is the point where ambitious plans start colliding with reality.
The figures presented by the International Energy Agency in its annual 2025 report show a 17% year-on-year (Y-o-Y) increase in energy usage for global data centers (around 485 terawatt-hours).
This number is expected to grow further, with experts anticipating it to reach 950 TWh by 2030.
In 2025, the power used solely for running AI systems was about 155 terawatt-hours, which is about one-third (32%) of all the electricity used by data centers in the same period, and the growth is expected to continue.
The rising demand for electricity puts pressure on the companies and people operating the power grid. One of the effects of this pressure is the growing interest in nuclear power.
Suggested reading: Semiconductor Market Outlook: Key Trends and Challenges in 2026
That being said, two real-world examples of deals are worth noting:
In 2024, Microsoft reached a 20-year agreement with Constellation Energy to acquire 835 megawatts of power from the restart of Unit 1 at the Three Mile Island nuclear plant.
The plant is expected to resume operations in 2028 to supply electricity to Microsoft's data centers, which power the company's cloud computing and AI initiatives.
During the same year, Google reached a deal with Kairos Power to create advanced small modular reactors (SMRs) aiming for a capacity of 500 megawatts by 2035.
The deployment of the first reactor is expected in 2030. The two companies look forward to obtaining reliable, carbon-free power for the tech giant’s rising data center demand while promoting the commercial use of next-gen nuclear power.
Obtaining valuable insights about data centers and projects in the sector
The public data available online can draw a general picture of data center market growth, but to be able to answer more niche, more specific questions like “Will it get its grid connection when expected?” or “Could permitting or power shortages delay it?” requires insights from people in the field.
These people can provide a lot of details that public data often misses.
If you need deeper insights into data centers, AI infrastructure, power supply, or related markets, the Expert Network Calls (ENC) platform (a marketplace of 20+ trusted expert networks) can help you reach people with direct industry experience and get a more detailed view of the market.
Frequently asked questions
What type of industry is a data center?
Data centers can be considered part of the Technology, Media, and Telecommunications (TMT) sector, and more specifically of the digital infrastructure and IT services, as they represent physical facilities that need to store, process, and transmit data. A data center is a building equipped with servers, storage systems, networking equipment, power systems, and cooling equipment.
What are the “big 3” data centers?
Data centers can be measured by IT load capacity and floor size. By IT load capacity (MW) the top three are:
- China Telecom-Inner Mongolia Information Park (IT capacity: ~3,000+ MW).
- The Citadel Campus (IT capacity: ~650 MW).
- Maiden NC Cluster (IT capacity ~400–500 MW).
By floor space, the top three data centers are:
- China Telecom - Inner Mongolia Information Park (994,000 m² (10.7 million ft²).)
- Switch - The Citadel (669,000 m² (7.2 million ft²)).
- Harbin Data Centre (660,000 m² (7.1 million ft²)).
If we talk about hyperscale cloud infrastructure, the list of the top three hyperscalers includes Amazon Web Services (AWS), Microsoft Azure, and Google Cloud - these operate the largest global data center networks.
Who is the biggest data center company?
Amazon Web Services is generally believed to be the largest data center and cloud infrastructure operator, with the top three list also featuring Microsoft and Google - this is if we analyze companies by the number of data centers and their total capacity.
What is a data center, and why is it bad?
A data center is a facility that was built to host computing and networking equipment where data is stored, processed, and distributed. There are three key things that data centers are criticized for: 1. High electricity consumption. 2. Large volume of water used for cooling in some facility types. 3. Land-use concerns for communities residing near large campuses.
Final word
When AI just started becoming a widely used tool worldwide, there were discussions about increasing the number of data centers. Now, the focus shifted more towards getting more energy for them. Data center capacity will continue growing, based on the picture we are witnessing at the moment, and AI will be the major driver of this trend.
To better understand the trends and get a more detailed picture with niche insights of what is going on in this field, connecting with experts is an investment you might consider. By registering on the ENC platform, you gain access to over 20 expert networks that will find the best subject-matter professionals for your project.
Suggested reading: How to Choose Experts Wisely: Vetting Criteria for High-Stakes Calls