Digital Infrastructure Investing

Last updated: August 20, 2026

Digital infrastructure is the newest large category within infrastructure and the one that fits the label least comfortably. A regulated water utility will be needed in fifty years in recognisably its current form. A data centre built for one generation of computing may or may not suit the next.

That tension — infrastructure-like contracted revenue attached to technology-exposed physical assets — defines the opportunity and the risk.

The three asset types

Data centres

Facilities housing computing equipment, providing power, cooling, connectivity, and physical security.

Hyperscale facilities are built for large technology companies and typically leased to a small number of counterparties on long contracts. The revenue is contracted, the counterparties are usually strong credits, and the exposure is concentrated.

Colocation facilities serve many tenants on shorter terms with more operational service. More diversified, more management-intensive, and more exposed to competitive pricing.

Enterprise and edge facilities serve specific corporate or latency-driven requirements.

Pricing has shifted increasingly toward power capacity rather than floor space, which reflects the underlying constraint on what these facilities can actually deliver.

Fibre networks

Long-lived physical cable plant producing revenue from long-term capacity agreements with carriers, enterprises, and content providers.

The economics are front-loaded and unforgiving: the cost is in construction, and returns depend on utilisation of what was built. A network built ahead of demand that does not arrive is a stranded asset. A network built where demand materialises has very high incremental margins, because additional capacity on existing fibre costs comparatively little.

Towers

Structures leased to wireless carriers for antenna placement.

Tower economics have a distinctive feature: incremental tenants on an existing structure add revenue at very low incremental cost. The structure is already built and already maintained. This makes tenancy ratio the central operating metric and makes the asset behave differently from most real estate, where additional revenue generally requires additional space.

The offsetting exposure is carrier concentration. There are few tenants, they consolidate, and consolidation can eliminate a tenant from a site.

What makes it infrastructure-like

Long-term contracted revenue, frequently with escalators.

High barriers to entry — power access, permitting, land, and capital.

Essential service characteristics. Data transport and processing are not discretionary for the economy that depends on them.

Long asset lives for the physical plant, particularly fibre and towers.

What makes it not

Technology risk. This is the genuine differentiator and it deserves plain treatment.

A road built to a specification remains a road. A data centre built for a given power density, cooling approach, and connectivity profile may not suit computing requirements a decade later. Retrofitting is possible and expensive, and there is a point beyond which a facility is better replaced than upgraded.

The current demand environment — driven substantially by growth in computationally intensive workloads including artificial intelligence — has pushed power and cooling requirements upward within a short period. That has been favourable for owners of suitable capacity and unfavourable for owners of facilities designed for an earlier profile.

The honest reading is that this cuts both ways. Rapid demand growth supports development economics and existing asset values. Rapid change in requirements is exactly what shortens the useful life of a purpose-built asset. An investor being shown the first argument should ask about the second.

Demand concentration. A small number of very large customers drive a large share of demand for hyperscale capacity. Their decisions about building versus leasing, and about where to locate, move the market.

Power as a hard constraint. Facilities need large, reliable electricity supply, and grid capacity in desirable locations is genuinely scarce. This has become a primary determinant of where anything can be built and a substantial component of the value of sites that already have secured power.

What to examine

The contracts. Term, escalators, renewal options and at what price, and — critically — weighted average lease term at the expected exit, not at acquisition. A facility with a much shorter remaining term at sale will price worse.

Tenant credit and concentration. A hyperscale facility leased to one counterparty is a credit exposure. Whether the lease is guaranteed by a parent matters.

Power. Contracted capacity, cost, contract term, and whether expansion capacity is secured. This is frequently the most valuable and least visible attribute of a site.

Technical specification relative to current and plausible future requirements — power density, cooling approach, and the cost of upgrading.

Location for latency, connectivity, and power availability. If the tenant leaves, the location is what remains.

Development versus operating exposure. A fund building facilities is taking construction, leasing, and demand risk. One buying leased operating assets is taking credit and residual value risk. These are different businesses.

Capital intensity and ongoing requirements. These assets consume capital continuously; a projection that treats maintenance capex lightly is understating cost.

Leverage and its maturity relative to lease terms.

Where it sits in a portfolio

Digital infrastructure is frequently presented as combining infrastructure stability with technology growth. The more accurate framing is that it combines infrastructure-like revenue contracts with technology-like asset risk.

The revenue is genuinely contracted, and the contracts are only as durable as the tenant’s continuing need for that specific facility in that specific configuration. Where the two align, these are excellent assets. Where the tenant’s requirements move faster than the building can, the contract runs to term and is not renewed — and the investor discovers that the residual value assumption was doing more work than the rent roll.

For the equivalent analysis in single-tenant real estate more broadly, see net lease investment, which shares much of the same structure.

This guide is educational and general; it is not investment advice. Asset characteristics and contract terms vary materially; review the specific documents.

Frequently Asked Questions

What is digital infrastructure?

The physical assets that carry and process data — data centres, fibre optic networks, and wireless towers, together with related assets such as subsea cables and edge computing facilities. It is generally classified within infrastructure because the assets are long-lived, capital-intensive, and typically produce contracted revenue.

How do data centre investments make money?

Through long-term leases or service agreements with tenants, priced by space and, increasingly, by power capacity. Hyperscale facilities are typically leased to a small number of large technology companies on long contracts, while colocation facilities serve many smaller tenants on shorter terms.

What is the main risk in data centres?

Depending on the facility, either tenant concentration or obsolescence. A hyperscale facility leased to one or two counterparties carries concentrated credit and renewal risk. All facilities face the possibility that power density, cooling requirements, or location preferences shift faster than a long-lived building can adapt.

Why is power a constraint on data centres?

Because facilities require large, reliable electricity supply, and grid capacity in desirable locations is limited. Access to power has become a primary determinant of where facilities can be built and of the value of sites that already have it, which affects both development economics and existing asset values.

Are towers and fibre similar investments?

They share long-lived physical assets and contracted revenue, and they differ in structure. Tower economics improve substantially when additional tenants are added to an existing structure. Fibre is capital-intensive to build and produces revenue from long-term capacity agreements, with returns depending heavily on utilisation of what was built.

Sources

  • ASC 842, Leases
  • Internal Revenue Code Sections 856-860 (real estate investment trusts)

Similar articles

Featured Partner Video Spotlight

Inland: Built for the long game

Discover Inland

Subscribe to our Newsletter