The landscape of global digital infrastructure is undergoing a seismic shift. For those monitoring the pulse of the data centre market, the trend is impossible to ignore: project announcements are ballooning in scale, transforming from modular regional facilities into sprawling, multi-billion-dollar AI-training campuses. This is not merely an incremental increase in activity; it is a fundamental restructuring of the industry’s DNA.
New data from GlobalData reveals that projects announced in 2025 carry an average value of roughly $1.9 billion—a fourfold increase from the $460 million average observed just three years earlier in 2022. As the industry pivots toward these "AI-scale" behemoths, analysts are warning that the traditional metrics used to forecast market health may be dangerously outdated. The pipeline is no longer just growing; it is being assembled from significantly larger, more complex, and riskier components.
The Chronology of Expansion: From Regional Hubs to Hyperscale Titans
To understand the current state of the market, one must look at the rapid trajectory of the last three years. In 2022, the data centre sector was largely characterized by a high volume of mid-sized regional facilities. These projects served localized enterprise needs and edge computing demands, typically requiring moderate capital expenditure and standard power grid connections.
By 2023, the emergence of generative AI began to rewrite the playbook. The demand for compute-heavy workloads necessitated massive, centralized clusters capable of hosting thousands of high-density GPUs. The "average" project size began its upward climb, signaling a transition from incremental capacity additions to large-scale infrastructure deployments.
By 2025, the shift reached a critical mass. The median project announcement has tripled, rising from approximately $200 million in 2022 to roughly $640 million today. This confirms that the market’s growth is not being driven by a few statistical outliers or vanity projects, but is a systemic evolution. The "standard" data centre is no longer a regional office support facility; it is a gargantuan, power-hungry campus that consumes land and capital on an unprecedented scale.
Supporting Data: By the Numbers
The disparity between project value and project count is the most telling metric in the current pipeline. While headline figures suggest a booming market, the underlying reality is one of concentration.
- Average Project Value: In 2022, the mean value per announced project sat at $460M. By 2025, that figure reached $1.9B.
- Median Project Value: Perhaps more indicative of the broader shift is the median value, which surged from $200M to $640M in the same timeframe.
- The Volume Paradox: While the total dollar value of the global pipeline has surged, the number of new individual project schemes has grown at a significantly slower rate.
This data highlights a critical warning for investors and stakeholders: headline growth in dollar terms should not be conflated with market breadth. If an analyst measures market health solely by the total capital expenditure projected for the coming decade, they are likely overstating the true demand, as the industry is funneling its resources into fewer, higher-risk ventures.
Growth in Scale is Not Growth in Breadth
A common trap for market observers is interpreting a larger pipeline as evidence of a more diverse and resilient market. However, the current data suggests the opposite.
The move toward these massive campuses means that the industry’s success is now tethered to a smaller number of high-stakes, high-risk projects. When the majority of a market’s value is concentrated in a handful of "gigaprojects," the loss or delay of a single facility can disrupt entire regional forecasts.
Furthermore, these projects do not follow the same development cycle as their smaller predecessors. They require intensive environmental impact assessments, massive transmission-level grid connections, and complex, multi-year construction timelines. Consequently, the "deliverability" of these projects is significantly lower than that of the smaller, more agile facilities that defined the market in 2022.
The Implications of "Lumpy" Demand
For contractors, equipment suppliers, and energy providers, this shift introduces the phenomenon of "lumpy demand." In the past, a portfolio could be balanced by winning contracts for several mid-sized facilities across different geographies. Today, the competitive landscape is binary: companies either secure a massive, multi-billion-dollar contract or they don’t.
1. Concentration of Financial Exposure
Financing has become increasingly rigid. What was once spread across ten mid-sized facilities—providing a natural hedge against project failure or regulatory hurdles—is now frequently consolidated into a single campus. This creates a "single point of failure" scenario where a delay in grid interconnection or a land-use dispute can freeze billions of dollars in capital. Diversification, once an inherent feature of the data centre market, must now be built intentionally by developers.
2. The Procurement Challenge
Tendering for these AI-scale campuses requires an entirely different level of corporate resources. The sheer scale of these projects necessitates massive pre-construction commitments, long-term procurement of power equipment, and specialized construction management teams. Only a limited number of players possess the balance sheet and technical expertise to compete for these projects, which could lead to a consolidation of the supply chain and reduced bargaining power for developers.
3. The "Deliverability Gap"
The most pressing implication is the disconnect between announcement and realization. Because these projects are so massive, they are inherently more sensitive to macroeconomic shocks, interest rate fluctuations, and energy infrastructure bottlenecks. A pipeline dominated by 2024 and 2025 announcements is, by definition, a pipeline of early-stage, long-term, and potentially fragile schemes. It will convert into revenue—and operational capacity—far more slowly than the industry is accustomed to.
Expert Perspectives and Risk Management
Industry observers are increasingly calling for more rigorous due diligence. The traditional approach of tracking a project from announcement to "go-live" is insufficient when the projects themselves carry such high degrees of systemic risk.
"Assessing the delivery risk on a $2 billion campus requires a forensic approach," says one lead analyst. "You cannot treat these like standard commercial real estate developments. Each one is a bespoke industrial project that relies on the stability of the local power grid and the long-term commitment of the cloud hyperscaler."
GlobalData’s comprehensive Data Centre Projects Database now tracks approximately 2,500 individual developments, segmenting them by stage, owner, and real-time activity. This granular level of tracking is essential for understanding which projects are actually poised for construction and which are merely "paper" announcements.
The consensus among experts is clear: the data centre market is no longer in a phase of simple expansion. It is in a phase of intense, high-stakes infrastructure development. As the industry looks toward 2030, the ability to discern which of these giant schemes will reach the finish line will be the primary differentiator between successful players and those who overextend their capacity.
Conclusion: Navigating the Future
The shift toward AI-scale data centres is an inevitable consequence of the global AI boom, but it brings with it a complex set of challenges. As projects grow in size and complexity, the "vintage" of a pipeline becomes a critical indicator of its reliability.
Stakeholders must look beyond the eye-watering headline numbers and focus on the underlying project-level evidence. With the "Deliverability Gap" becoming a central concern for the next half-decade, the industry must pivot toward more nuanced risk assessments, deliberate diversification, and a deeper understanding of the infrastructure constraints that will ultimately dictate the success of the AI era.
For those navigating this new environment, the message is simple: size matters, but it is the quality and the deliverability of the individual project that will define the next chapter of the digital revolution.
