Two sessions at Datacloud Global Congress 2026 mapped the emerging fault lines in delivery strategy for a sector that needs to build faster, larger, and in more constrained locations than it ever has before. The conclusion was not that modular is always better, or that vertical is the answer, but that the industry is deploying both without sufficient clarity about when each approach actually works.

The data centre development pipeline faces a set of intersecting pressures that are, taken together, unprecedented. Hyperscaler demand is pushing campus sizes into the hundreds of megawatts. Urban land scarcity in key markets is forcing development upward.

Grid connection timelines mean that construction schedules must be compressed to remain commercially viable. And supply chain constraints-particularly for skilled labour, which the AI demand panels identified as the primary delivery bottleneck-mean that the old model of deploying large site teams for multi-year bespoke builds is under structural pressure. Modular construction, broadly defined, has been proposed as the answer. But what “modular” means in practice varies so widely that the term is beginning to obscure more than it illuminates.

The modular promise: speed, predictability, and supply chain control

The modular construction panel at DCGC 2026 set out the case for modularisation clearly. Some organisations are targeting data centre builds of 18 months or less using modular approaches-a timeline that, if achievable, would significantly compress the gap between grid connection and facility readiness. Beyond speed, modularisation reduces strain on local construction infrastructure by minimising the need for large onsite labour forces and construction traffic, improves supply chain management through repeatable processes, and enables more predictable project outcomes. For

European operators navigating complex permitting environments, the ability to separate factory-based construction from site-based installation also reduces planning risk exposure. Modular approaches can also support energy efficiency and sustainability goals, particularly as EU efficiency directives tighten. Early contractor and vendor engagement is a prerequisite for capturing these benefits-the panel was clear that modular done badly, without integrated planning and collaborative vendor relationships, produces the same delays and cost overruns as any other approach.

The risks the modular advocates are underplaying

The panel also surfaced concerns that modular proponents tend to downplay. Over-standardisation was identified as a genuine risk: an industry where technology, regulations, and customer requirements change rapidly can find itself locked into modular architectures that were optimised for last year’s requirements.

Interface management between modular components and existing infrastructure introduces its own complexity and risk, particularly when integrating new modules with utility connections or existing building structures. Supply chain risk is also not eliminated by modularisation-it is transformed.

Reliance on a single modular vendor creates a different kind of concentration risk from traditional bespoke construction, and dual supply chain strategies are essential for resilience and to maintain competitive pressure on suppliers. Regional variations in standards and regulations were flagged as a specific barrier to truly standardised global modular products, particularly in Europe, where upcoming energy efficiency directives and localisation requirements vary across member states.

Vertical construction: the land-constrained market response

The vertical data centre panel considered a related but distinct delivery challenge: what operators do when horizontal sprawl is not an option. In Singapore, Japan, and increasingly in dense European markets, vertical construction-multi-storey data centres with stacked technical floors-is becoming the default.

Microsoft’s multi-storey facility in Singapore was cited as a reference project. The panel noted that use case, facility dependencies, and specific customer requirements are critical in determining whether vertical construction is appropriate: for some latency-sensitive applications, the vertical stacking of compute introduces propagation delays that matter; for others, it does not.

In Japan, vertical builds are often the only option due to land constraints, but come with building code complexity, shadowing issues, and logistics challenges that require early engagement with insurers as well as planners. In Europe, height restrictions and permitting complexity make vertical builds harder than in Asia, necessitating modular and flexible internal layouts to maximise the use of constrained floor plates.

Key Takeaways

• Modular construction offers genuine advantages in speed, predictability, and supply chain management, but the term covers a spectrum from standardised component kits to fully containerised modules-the benefits vary accordingly.
• Over-standardisation is a real risk in a rapidly evolving market: modular architectures optimised for today’s workloads may be difficult to adapt for the density and cooling requirements of AI systems two or three years from now.
• Single-vendor dependency in modular supply chains creates concentration risk; dual supply chain strategies are an operational necessity, not an optional resilience measure.
• Vertical data centre construction is the default response to urban land scarcity in Singapore, Japan, and increasingly in European dense markets, but comes with regulatory, structural, and insurance complexity that must be integrated from the design phase.
• The 18-month build target that modularisation promises is only commercially valuable if grid connections and utility services can match that timeline-compression of the construction schedule without parallel acceleration of power delivery solves the wrong problem.

The delivery debate in data centre construction is not going to resolve into a single winning approach. Different markets, workloads, and site constraints will continue to demand different delivery models.

What DCGC 2026 surfaced is that the industry needs sharper thinking about which model fits which context, rather than treating modular or vertical as universal answers. The operators who do this analysis rigorously-and who build the supply chain relationships and planning expertise to execute both approaches-will be the ones who actually deliver the capacity the market needs on the timelines the market demands.