New Stage of Data Center Growth: AI-Driven Workloads, Investment Pressure, and Delivery Risks

Industry Background

The rise of AI is fundamentally reshaping the computer age, driving an upgrade in workload demands and redefining data center requirements. Historically, enterprise computing evolved from on-premises mainframes to physically centralized infrastructure, and ultimately to cloud computing dominated by hyperscalers. Today, the proliferation of AI and its penetration into cloud workloads are triggering a transformation: data centers must provide higher compute density, stricter total cost of ownership (TCO), and increasingly latency-sensitive architectures. Consequently, facility scale continues to expand, with next-generation AI campuses reaching hundreds of megawatts or even gigawatt levels. Compared to traditional cloud or colocation workloads, AI's demands for compute and power place new pressures on infrastructure design, site selection, and economics.

Market Impact

Meeting the surging demand for AI requires unprecedented near-term capital expenditure, with global data center investment set to exceed the annual capex trajectory of the hyperscalers themselves. According to FTI Consulting, by 2027-2029, AI-driven capital expenditure needs will surpass hyperscaler capex by 140-160%, indicating the need for new capital pools and accelerating the rise of alternative infrastructure providers.

• Two major trends are emerging:
• Outsourcing at the chip level is driving the growth of "NeoCloud" or GPU-as-a-Service (GPUaaS) platforms—spanning giants, regional emerging players, sovereign clouds, and traditional B2B providers expanding into the GPU space.
• Speed to market is becoming a decisive competitive barrier, as platforms race to lock down land, power, and customer commitments before supply tightens further.

Despite rapid development, structural constraints—utilities and investors requiring customer commitments to enable operators to secure financing and power—limit the risk of oversupply and speculative construction.

Competitive Landscape

The report identifies four key areas that determine the achievability of the "ready for service (RFS)" timeline: planning and permitting, power procurement and delivery, construction execution, and customer contract visibility. Power is often the primary bottleneck, prompting operators to explore alternatives beyond traditional grid connections, including natural gas, nuclear, and batteries, to bypass grid constraints and accelerate deployment.

Meanwhile, the colocation and cloud markets are evolving alongside AI. Hybrid cloud strategies are expanding, as enterprises balance the scalability of public cloud with the need to manage specific workloads on-premises, while sovereign cloud proposals are becoming matters of national strategic importance. Retail colocation is also expected to absorb demand from industries with regulatory, latency, or deployment-specific requirements.

Enterprise ImplicationsIn a rapidly changing market, investors and operators need to ask the right questions to assess platform risks and valuations: - What is the strength of tenant contracts and the counterparty risk assumed by data center operators? - How realistic is the RFS timeline for each site? How does it compare to market construction over the next 24 months? - As chipset technologies evolve, what obsolescence risks do data center infrastructures face? How much upgrade capital expenditure is required? - Where does demand come from? Which locations can most efficiently serve different workloads? - How might prices erode with the release of new-generation hardware? - How will alternative power sources reshape workload siting? When will they begin to affect operators’ purchasing criteria?

Only those investors and operators who deeply consider these questions can accurately price risks, capture emerging value, and remain competitive in the next wave of transformation in the data center landscape.

Future Outlook

Over the next 12-24 months, the speed of data center construction will become a key differentiator, and power bottlenecks may drive more off-grid power solutions. In the 24-month to 3-year horizon, the NeoCloud/GPUaaS model may become mainstream, with hyperscalers either partnering with alternative capital or accelerating their own investments. Sovereign clouds and edge deployments will continue to grow due to regulatory and latency demands. Ultimately, only those participants who precisely manage risks and respond quickly will succeed in the next wave of growth.