Over the past decade, data centers have shifted from being an IT concern to a core part of global infrastructure. AI models, cloud platforms, streaming, e-commerce, fintech, and IoT all depend on one thing: dense clusters of compute, storage, and network capacity plugged into reliable power.
The result is a construction boom with real estate, infrastructure, and technology converging in a single asset class. Let's take a look at where the data center market is now, who is driving it, where growth is headed, and how generative design and real-time feasibility tools can help teams keep up with the pace and complexity of demand.
Market Size: A Hundreds-of-Billions-Per-Year Asset Class
Different analysts break the market down slightly differently (hardware vs. services, colocation vs. hyperscale), but they all tell the same story: fast growth from a very large base.
• One global outlook estimates the data center market at around USD 347.6 billion in 2024, reaching approximately USD 652.0 billion by 2030, implying roughly 11% CAGR between 2025 and 2030 (Grand View Research).
• Another forecast puts the sector at about USD 386.7 billion in 2025, growing to USD 627.4 billion by 2030 at roughly 10.2% CAGR over that period (Mordor Intelligence).
• Looking further out, one long-term projection sees the global market rising from roughly USD 319.5 billion in 2024 to almost USD 987.7 billion by 2035, with a ~10.8% CAGR from 2025 to 2035 (Spherical Insights).
Even if you take the conservative end of those ranges, the picture is clear: data centers are set to nearly double in size this decade, with continued growth beyond 2030.
On the services side (managed, colocation, hosting, etc.), one recent report values the global data center services market at USD 23.1 billion in 2024, projecting it to reach USD 38.7 billion by 2030 at about 9% CAGR (GlobeNewswire).
The real constraint is no longer demand—it is power, land, and the ability to deliver highly engineered facilities fast enough.
Why Growth Is Accelerating: AI, Cloud, and Power Demand
Several overlapping drivers are pushing demand for new capacity:
• AI workloads: Training and serving large models requires orders of magnitude more compute than traditional enterprise workloads. McKinsey notes that AI is now a primary driver of incremental demand for AI-ready data centers, with hyperscale cloud providers (AWS, Google Cloud, Microsoft Azure, Baidu, etc.) (DGT Infra) fueling much of the growth (McKinsey & Company).
• Cloud and SaaS adoption: Enterprises continue to shift workloads off-premise, increasing demand for both hyperscale and colocation capacity (JLL).
• Data center construction boom: A recent insurance and risk analysis report describes a "great data center goldrush" driven by AI's heavy computing and cooling needs, with record levels of construction underway worldwide (Allianz Commercial).
This growth has a direct impact on infrastructure:
• Goldman Sachs Research forecasts that global power demand from data centers could rise by about 50% by 2027 and by as much as 165% by 2030 compared to 2023 (Goldman Sachs).
• A McKinsey-highlighted analysis suggests data center electricity demand could grow at roughly 17% per year globally between 2022 and 2030, with U.S. data center power consumption potentially exceeding 14% of total national electricity use by 2030, roughly triple its share in 2023 (Business Insider).
• The International Energy Agency has similarly warned that data center power consumption could more than double by 2030, with AI-focused centers consuming as much electricity as some countries (The Wall Street Journal).
In parallel, the U.S. hyperscale data center market alone is forecast to reach around $290 billion by 2030, growing at roughly 7.5% CAGR from 2024 to 2030, driven primarily by AI and cloud demand (Business Wire).
The implication for architects, developers, and investors: demand is strong, but project viability is increasingly constrained by power availability, grid capacity, cooling strategies, and permitting.
Who's Involved: Hyperscalers, Colocation REITs, and Infra Capital
The ecosystem around data centers spans several layers.
Hyperscale Cloud and Tech Platforms
At the top end, cloud and internet platforms build and operate their own campuses:
• Amazon Web Services, Microsoft Azure, Google Cloud Platform, Meta Platforms and others are listed among the leading global data center companies by installed capacity and scale (DGTL Infra).
These players typically drive demand for large-scale, AI-ready campuses with hundreds of megawatts of capacity, often developed with specialized partners.
Colocation and Wholesale Providers
Below the hyperscalers, a group of global colocation and wholesale operators provide space, power, and connectivity to enterprises and cloud tenants:
• Recurrent names include Equinix, Digital Realty, NTT Global, CyrusOne, QTS, GDS Holdings, Telehouse, Switch, CoreSite, Centersquare, and others (Blackridge Research).
These companies tend to focus on flexible, carrier-rich facilities with strong interconnection and are heavily involved in multi-tenant facilities in key metros.
Investors, Utilities, and Local Stakeholders
There is also a growing role for:
• Infrastructure funds and REITs financing large campus builds
• Utilities and grid operators managing power generation and transmission to data center clusters
• Local governments and communities involved in zoning, environmental approvals, and negotiations around power, land use, and tax incentives
At the same time, some large tech firms are recalibrating specific projects in response to land, power, and strategy considerations. For example, Microsoft has slowed or paused certain AI data center projects in parts of the U.S., even as it continues to invest heavily in AI infrastructure globally (AP News).
What's Coming Next: Trends Shaping Data Center Development
From an architectural and development standpoint, several trends are reshaping how these facilities are sited and designed.
1. Power-First Site Selection
Power has effectively become the primary constraint. Reports on hyperscale development highlight:
• Competition for grid capacity
• Long lead times for new substations and transmission
• Trade-offs between co-locating near existing generation vs. building on-site or behind-the-meter solutions (MMCG Invest).
This is shifting site selection criteria: the "best" site may be the one with available megawatts and upgrade potential, not simply cheap land.
2. Cooling and Density
AI workloads and high-density racks demand:
• Higher power densities per rack and per square meter
• Advanced cooling solutions (rear-door heat exchangers, direct-to-chip liquid cooling, immersion concepts)
• Highly optimized hot-aisle/cold-aisle containment and airflow management (Cove).
Cooling strategy is no longer a late-stage MEP decision; it directly influences massing, floor-to-floor heights, service corridors, and roof/yard allocations.
3. Sustainability and Regulation
Regulators and communities are increasingly focused on:
• Energy efficiency targets (PUE, WUE)
• Renewable energy sourcing and PPAs
• Noise, heat rejection, water use, and visual impact
This pushes designs toward more efficient envelopes, mechanical systems, and sometimes hybridization with district energy, industrial symbiosis, or on-site generation (Cove).
4. Edge and Regional Expansion
As latency-sensitive applications (gaming, AR/VR, autonomous systems) grow, capacity is moving closer to end users via:
• Smaller edge facilities in secondary and tertiary markets
• Hybrid deployments combining central hyperscale campuses with distributed nodes (JLL).
For designers and developers, this creates a spectrum of typologies—from 10+ MW edge sites to 300+ MW hyperscale campuses—each with distinct planning, zoning, and infrastructure challenges.
Why Data Center Design Is a Natural Fit for Generative Design
Generative design is generally defined as an approach where the designer specifies goals, constraints, and inputs (e.g., performance targets, materials, costs), and the software algorithmically generates multiple options that meet or approximate those criteria.
In the context of data centers, the design problem is highly constrained and multi-objective:
• Site and planning constraints: parcel shape, setbacks, height limits, FAR/coverage, easements, buffers, truck access, service yards
• Technical constraints: power density, cooling strategies, equipment layouts, redundancy/availability targets
• Economic constraints: phasing, capex per MW, operational efficiency, scalability
• Sustainability and regulatory constraints: noise, visual impact, water use, emissions, grid impact
This is exactly the type of problem generative design handles well: many variables, clear constraints, and quantifiable performance metrics.
Recent work on generative design in architecture and the built environment notes that these methods are particularly effective when they can combine performance metrics (e.g., energy use, structure, cost) with iterative spatial layouts.
For data centers, that can translate into:
• Automatically generating and evaluating alternative site layouts (building footprint, yards, substations, cooling plants, parking, offices) within zoning and FAR limits
• Testing multiple white space configurations (rack rows, aisles, hot/cold containment, support spaces) and routing for power and cooling
• Exploring phasing scenarios: how an initial 20–50 MW build-out can logically expand toward 100–300 MW over time, while maintaining live operations
• Comparing cooling and power distribution schemes—for example, evaluating how different mechanical plant locations impact piping runs, distribution losses, and roof/yard utilization
Instead of manually drawing a few options and stress-testing them one by one, teams can generate and filter dozens of feasible schemes in a much shorter time.
How Real-Time Feasibility and Generative Layout Tools Contribute
Real-time feasibility and generative design tools do not replace the domain expertise of data center architects and engineers. They change the way early-stage decisions are made.
For architects, planners, and developers working on data centers, these tools can:
• Accelerate site vetting: Input zoning parameters, setbacks, coverage, and height limits; quickly see which building footprints and massing options are actually viable before committing to a concept.
• Quantify trade-offs early: Understand how changes in FAR, footprint, or building orientation affect capacity (GFA/MW), yard space for mechanical equipment, and potential phasing.
• Support structured scenario exploration: Instead of one or two manual options, teams can work through a structured set of "what-if" questions: What if we change from air-cooled to liquid-cooled in later phases? What if we reserve roof or yard areas for future mechanical expansion? What if local zoning limits height more than expected?
• Align stakeholders around data, not drawings: Acquisition, design, engineering, and finance teams can review the same live feasibility outputs (capacity, coverage, potential MW capacity) instead of trading static PDFs.
For a generative-design-driven feasibility platform like Zenerate, the opportunity in data centers is on the front end of the pipeline:
• Rapidly testing candidate sites for coverage, massing, setbacks, buffers, and potential capacity
• Generating multiple building configurations that respect local zoning
• Providing instant quantitative feedback on area, constraints, and development intensity as teams adjust assumptions
Specialized data center design tools and MEP models will still be essential deeper in the process. The earlier the team can identify viable land, massing, and maximum build-out potential, the easier it becomes to justify acquisitions, negotiate power, and sequence detailed design.
Looking Ahead: Data Centers as a Design and Feasibility Frontier
The macro picture is straightforward: data center demand is strong, driven by AI and cloud, with global market forecasts pointing to high-single- or low-double-digit annual growth through at least 2030.
The challenge is not whether new facilities will be needed, but how quickly and efficiently they can be delivered under tight constraints on:
• Power availability
• Cooling and density
• Land, zoning, and community impact
• Capital and operational efficiency
That combination makes data centers a natural proving ground for generative design and real-time feasibility.
The teams that can explore more scenarios in less time, validate options quantitatively, and bring stakeholders into a shared, data-driven environment will be better positioned—not just to build more data centers, but to build better, more efficient ones.
Where Zenerate Is Heading
As demand for data centers accelerates, early-stage decisions are becoming more complex, more constrained, and more time-sensitive. Power availability, density strategies, land restrictions, and phased capacity planning all need to be evaluated long before a detailed design takes shape. These challenges align directly with the strengths of real-time feasibility and generative modeling.
Zenerate's next phase is focused on expanding our real-time feasibility capabilities into this sector. As we expand, we aim to support developers, designers, and acquisition teams as they navigate the unique constraints of data center planning. It's an evolution of what Zenerate already does best: accelerating the earliest, most impactful decisions in the development process.
We're actively laying the groundwork for this direction now. Stay tuned for future updates as we move toward supporting data center feasibility and scenario exploration within the Zenerate platform.
Explore How Zenerate can help
Whether your team is actively pursuing data center developments or simply exploring what's possible, we welcome conversations about your workflow, challenges, and long-term needs. Even as we continue expanding our capabilities, these discussions help us better understand how Zenerate can support the industry and shape the direction of future features.
If you'd like to share your goals or see how Zenerate's real-time feasibility tools work today, we'd be happy to connect.