Global data center spending is expected to reach $1 trillion in 2026 and more than double to $2.1 trillion by 2028, according to Bank of America Global Research estimates reported by Reuters on October 23, 2026. The forecast reflects rising demand for artificial intelligence services, which require vast amounts of computing capacity, advanced servers, high-speed networking and reliable power infrastructure.
The projected investment boom is creating opportunities across the technology and industrial sectors. Chipmakers and server manufacturers are among the main beneficiaries, but companies supplying generators, electrical equipment, cooling systems and construction machinery are also positioned to gain. BofA estimates that the global data center market could reach $2.93 trillion by 2030, although delivering this expansion will depend on access to electricity, equipment, financing and construction capacity.
Key Takeaways
- $2.1 trillion forecast: Global data center spending could exceed $2.1 trillion by 2028, more than double the estimated $1 trillion in 2026.
- AI drives demand: Training and running AI models is increasing demand for computing capacity and supporting infrastructure.
- IT equipment dominates: BofA estimates IT-related expenses account for 78% of total data center spending.
- Servers lead costs: Servers are estimated to cost $31.4 million per megawatt of data center capacity, making them the largest component in BofA’s cost breakdown.
- Neoclouds cost more per megawatt: AI-focused data centers are estimated to cost around $57 million per megawatt, compared with $49 million for data centers overall.
- Market expansion: BofA projects the global data center market could reach $2.93 trillion by 2030.
- Execution risks remain: Electricity availability, equipment supply, construction constraints, financing and community opposition could slow projects.
Why Data Center Spending Is Expected to Surge
The rapid adoption of generative AI and other computing-intensive services is changing the economics of digital infrastructure. AI models need powerful processors, large amounts of memory and storage, high-speed connections between servers and substantial electricity to operate.
As businesses integrate AI into search, software development, customer service and enterprise workflows, infrastructure providers are investing in additional computing capacity. Cloud companies and specialized AI infrastructure providers need more data centers to host the servers that process these workloads.
According to BofA Global Research, the global data center market was worth approximately $275 billion in 2022. The bank expects the market to grow substantially as AI services expand, with spending potentially reaching $2.93 trillion by 2030.
BofA estimates that the market could achieve a compound annual growth rate of 37% from 2025 to 2030. This is a forecast rather than a guaranteed outcome, and the actual growth rate will depend on AI adoption, infrastructure utilization and the economics of building new facilities.
How Much Will the Data Center Industry Spend?
The scale of the forecast highlights how quickly data centers are becoming a major category of global infrastructure investment.
| Metric | BofA estimate |
|---|---|
| Global data center market in 2022 | $275 billion |
| Expected spending in 2026 | $1 trillion |
| Expected spending by 2028 | $2.1 trillion |
| Potential market size by 2030 | $2.93 trillion |
| Estimated market CAGR, 2025–2030 | 37% |
Source: Reuters, October 23, 2026.
These figures represent BofA’s estimates and projections. The $1 trillion and $2.1 trillion figures describe expected spending at different points in the forecast, while the $2.93 trillion figure represents the bank’s estimate of the market’s potential size by 2030. They should not be interpreted as guaranteed expenditure or as the revenue of any single company.
Servers and IT Equipment Account for Most of the Cost
Building a data center involves much more than constructing a building. Operators must purchase computing hardware, connect servers through high-speed networks, distribute electricity and maintain temperatures within acceptable limits.
BofA estimates that IT-related expenses account for 78% of total data center spending. Servers are the most expensive component in its per-megawatt cost breakdown, followed by networking equipment.
| Component | Estimated cost per megawatt |
|---|---|
| Total data center construction and equipment | $49 million |
| Servers | $31.4 million |
| Networking | $4.8 million |
| Other individual infrastructure functions | Approximately $0.1 million to $1.1 million |
Source: BofA Global Research estimates reported by Reuters.
The component figures are estimates and should not be treated as a complete itemized bill for every project. Costs vary depending on the type of facility, computing equipment, power density, cooling requirements and location.
The figures nevertheless show why the AI infrastructure boom is creating demand for more than semiconductor products. Data center operators also need networking systems, electrical distribution equipment, backup generators and specialized cooling technology.
Why AI-Focused Data Centers Cost More
AI-focused data centers, often called neoclouds in the industry, are designed to support intensive computing workloads. These facilities may use high-density accelerator systems that require substantial electricity and specialized cooling.
BofA estimates that neocloud facilities cost approximately $57 million per megawatt, compared with $49 million per megawatt for data centers overall. Although AI-focused facilities may require less physical space for a given amount of computing capacity, their equipment and power requirements can increase the cost per megawatt.
This difference reflects the importance of power density and hardware configuration. Two facilities with similar electrical capacity can have different construction costs depending on the processors they install, the cooling systems they need and the networking infrastructure connecting their servers.
As AI workloads expand, developers are likely to pay closer attention to how efficiently each facility converts its electricity supply into usable computing capacity.
Which Industries Could Benefit From the Spending Boom?
The expected investment could benefit a broad range of technology and industrial businesses.
Semiconductors and servers: Data centers need processors, memory, storage and complete server systems. These products represent a large share of the infrastructure bill and are central to expanding AI computing capacity.
Networking equipment: As operators connect larger numbers of servers, demand increases for switches, network interfaces and high-speed connectivity systems. Efficient data transfer is particularly important for large AI workloads.
Power infrastructure: Data centers require electrical distribution equipment, switchgear, generators and backup power systems. Equipment availability can affect how quickly a new facility becomes operational.
Cooling systems: High-performance computing hardware generates substantial heat. Cooling equipment and thermal-management systems are essential for maintaining reliable operations.
Construction and engineering: Developers need land, buildings, electrical connections and specialized construction services. Companies supplying construction equipment and industrial systems can benefit from the additional projects.
This broad spending pattern means the data center boom extends into traditional industrial markets. Companies that once served manufacturing, commercial buildings or power infrastructure may find new demand from technology developers.
Power Constraints and Other Risks Could Slow Growth
The spending forecast does not mean every planned data center will be built on schedule. Power availability is one of the most important constraints facing the industry.
Large facilities require reliable electricity connections, and new generation and transmission infrastructure can take years to develop. Delays in grid connections, permits or equipment deliveries can push back project timelines and increase costs.
Financing is another consideration. Data centers require substantial upfront investment, and developers must ensure that expected demand and customer commitments justify the expenditure. If AI demand grows more slowly than expected or computing capacity is underused, some projects could deliver weaker returns.
Local opposition is also becoming a concern in some markets. Residents and policymakers have raised questions about electricity bills, water use, land requirements and the environmental impact of large facilities. These disputes can complicate planning and permitting.
The industry’s growth will therefore depend not only on demand for AI but also on the ability to deliver infrastructure efficiently and secure public, regulatory and utility support.
The Bigger Picture
BofA’s forecast illustrates the scale of the physical investment behind the AI economy. Every new generation of AI services requires computing resources, and those resources depend on data centers, semiconductor hardware, power delivery, networking and cooling. The industry’s expansion is consequently creating opportunities across technology manufacturing and conventional industrial infrastructure.
However, the projected growth also raises questions about capital allocation and long-term returns. The amount spent on infrastructure will not automatically translate into equivalent profits. Developers must balance capacity expansion against customer demand, operating costs and the possibility that technology improvements could change hardware requirements. The companies best positioned to benefit may be those that deliver essential equipment efficiently while helping customers overcome power and construction constraints.
FAQs
1. How much will global data center spending reach by 2028?
BofA Global Research expects global data center spending to exceed $2.1 trillion by 2028, compared with an estimated $1 trillion in 2026.
2. What is driving the increase in data center investment?
Growing demand for AI services is a major driver. AI models require powerful servers, networking, storage, electricity and cooling infrastructure, encouraging cloud providers and specialized operators to expand capacity.
3. Which part of a data center costs the most?
BofA estimates that servers are the most expensive component, at approximately $31.4 million per megawatt. Networking is the second-largest component in its reported breakdown, at $4.8 million per megawatt.
4. What are neocloud data centers?
Neoclouds are AI-focused data center providers and facilities designed around high-performance computing workloads. BofA estimates that these facilities cost around $57 million per megawatt, compared with $49 million for data centers overall.
5. Could data center spending growth slow down?
Yes. Electricity shortages, grid-connection delays, equipment constraints, financing costs, regulatory hurdles and slower-than-expected AI adoption could affect construction schedules and returns on investment.
Looking Ahead
The coming years will test whether the infrastructure industry can keep pace with AI demand. Spending on servers, networking, electrical equipment and cooling systems is expected to remain central to new data center development, while power availability and construction timelines will influence which projects move forward first.
BofA’s projections point to a market with substantial growth potential, but the final outcome will depend on execution as much as demand. Investors and businesses will need to watch infrastructure spending, AI-service adoption, electricity capacity and project economics to determine whether the projected expansion translates into sustainable returns.
Get the day’s top stories in your inbox
One concise email. No spam, unsubscribe anytime.
