If you follow technology news, you must have read the headlines about the extraordinary demand for new data centers. Artificial intelligence (AI) is driving billions of dollars in infrastructure investment, with technology companies competing for access to facilities that, in some cases, have not yet been built.
At a recent industry conference, a breakout session with industry engineers and datacenter operators discussed the challenges of meeting this demand. The scale of investment was interesting, but what caught the audience’s attention was how different the requirements of these AI facilities are from those of a typical business data center.
It got us thinking about the businesses Morefield works with across Central Pennsylvania.
For a business considering whether to move its servers out of an office building or replace aging equipment with hosted infrastructure, the headlines might suggest that data center capacity is becoming difficult to obtain. The reality is more nuanced. While AI is placing enormous pressure on certain segments of the market, most businesses have different computing requirements and different hosting options.
The demand for AI computing infrastructure is real, and the numbers are remarkable.
According to CBRE’s North America Data Center Trends report, published in August 2026, data center capacity across the eight primary North American wholesale markets increased 33.7% year-over-year during the first half of 2026. Despite that growth, vacancy declined to just 1.4%.
80.4% of the capacity currently under construction in those markets is already committed to customers.
Much of this demand is being driven by hyperscale computing and AI applications that require enormous amounts of processing power. Companies are securing large blocks of electrical capacity, often measured in megawatts, to support specialized computing environments.
These are not typical business technology deployments. Training sophisticated AI models can require thousands of specialized processors operating simultaneously, creating infrastructure requirements far beyond those of conventional business applications.
The industry is responding with new construction, substantial capital investment, and long-term 10+ year commitments for power and facility capacity. But the important distinction is understanding what makes these facilities different.
A traditional business server might support an accounting application, database, virtual desktop environment, or specialized line-of-business software. These systems require reliable computing resources, but their electrical power and cooling requirements are relatively modest compared with high-density AI infrastructure.
AI computing is different.
Research published by JLL in October 2026 illustrates the difference. Traditional data center racks commonly operate around 10 kilowatts (kW) of electrical power, while some high-performance AI configurations require as much as 100 kW per rack. That concentration of computing equipment creates several challenges:
These requirements explain why so much new data center construction is focused on facilities specifically engineered for AI. Retrofitting an existing facility is possible, but depending on its original design, the cost and complexity makes new construction more practical.
Here is the important takeaway: A data center that cannot accommodate high-density AI computing is capable of supporting conventional business workloads.
Most small and midsized businesses are not building AI models or operating thousands of specialized processors. Their computing requirements are more familiar: business applications, databases, file storage, virtual servers, backup systems, and disaster recovery environments.
These workloads do not require the specialized facilities driving national construction headlines. Research from the Uptime Institute’s 2026 Global Data Center Survey confirms that while high-density computing is growing, conventional infrastructure remains an important part of the data center landscape.
Existing Datacenter facilities are equipped to support traditional business requirements.
Morefield works with established data center operators, who have sufficient capacity within their Pennsylvania facilities to support traditional business demand. That is an important distinction from the national headlines.
While companies requiring enormous amounts of electrical power may be competing for capacity in major data center markets, businesses with traditional workloads can still explore existing regional facilities and hosting services.
Of course, availability, pricing, and technical suitability depend on the specific requirements. But the AI infrastructure boom does not mean that a Pennsylvania business must postpone its own data center strategy.
Consider a scenario familiar to many businesses. A company owns or leases an office building that historically accommodated most of its employees. Today, a significant portion of the workforce operates remotely or on a hybrid schedule.
The building may have more office space than the business needs, yet somewhere within the building in a closet or a remote room in the basement, sits the server room supporting applications and infrastructure essential to daily operations.
Those servers continue to require electrical power, cooling, physical security, maintenance, and periodic replacement. As the equipment approaches the end of its useful life, the organization faces an important decision: Should it purchase another generation of servers and continue hosting them on-site? Or is this an opportunity to reconsider how those workloads are delivered?
For some organizations, maintaining servers within their own facilities remains the right choice. Specific applications, performance requirements, regulatory considerations, or business needs may favor local infrastructure. However, this approach means continuing to invest in hardware and maintaining the environment necessary to operate it reliably.
Colocation, often called colo, allows a business to place its own servers within a professionally operated data center. The organization retains ownership of its equipment while the facility operator provides the physical environment, including power, cooling, connectivity options, and physical security.
This can reduce risks associated with maintaining infrastructure in a traditional office building. Colocation is particularly worth considering when existing equipment still has useful life or when the organization wants to retain direct control over its hardware.
Infrastructure as a Service (IaaS) offers another alternative. Rather than purchasing replacement servers, a business obtains computing, storage, and networking resources from a service provider.
The provider is responsible for maintaining the underlying physical infrastructure, while the business continues managing its applications, data, access, and other responsibilities as defined by the service agreement.
This approach eliminates the capital investment associated with a hardware refresh and shifts infrastructure costs toward an ongoing operating expense. For businesses who currently Work With an MSP or IT departments with limited personnel, reducing the burden of hardware management can also free resources for projects with a greater impact on the business, including cybersecurity improvements, data governance, automation, and AI initiatives.
There is no universal answer to whether a business should retain its servers, move them into colocation, or transition workloads to IaaS.
The decision requires understanding the applications being supported, their performance and availability requirements, connectivity needs, security obligations, recovery objectives, and total cost of ownership.
For example, moving a server into a data center does not automatically improve Disaster Recovery. Organizations still need to evaluate backups, geographic redundancy, recovery procedures, and the consequences of a network outage.
Similarly, IaaS may reduce hardware responsibilities, but it does not eliminate the need for IT administration or cybersecurity oversight. These are decisions best made through an assessment of business requirements rather than a reaction to technology headlines.
At Morefield, our role is to help your business evaluate these options, understand the tradeoffs, and develop an infrastructure strategy that is best for your operations.
AI is transforming the data center industry, creating significant demand for facilities capable of supporting a new generation of computing. But not every data center serves the same purpose, and not every business competes for the same infrastructure.
For Central Pennsylvania organizations considering a server refresh, colocation, or hosted infrastructure, existing data center facilities remain a viable option worth evaluating. The question isn’t whether the industry has enough capacity to support AI.
It’s whether your business has the right infrastructure strategy to support its operations today and adapt to what comes next.