T hat era has not ended, but it has collided Brookfield describes the convergence of digitalisation, decarbonisation and deglobalisation as an infrastructure supercycle: more than USD 100 trillion of investment required by 2040, with the AI value chain alone requiring some USD 7 trillion over the next decade.¹ The bill for rebuilding the physical world: USD 100 trillion by 2040 with the physical world. Artificial intelligence requires data centres, fibre and vast quantities of reliable electricity. Electrification requires generation, transmission and storage.
Engineers have always understood that optionality, staging and redundancy create value. Today, investors are becoming explicit about recognising those qualities in underwriting, in the leverage an asset can safely carry, the probability that it reaches financial close, the range of customers it can serve and the value it retains at exit. Project drawings are entering the investment memorandum. Forecast error is part of the design brief Traditional project design begins with a specification across expected demand, defined capacity and often, an anchor customer. More resilient underwriting begins by testing each of those assumptions. What happens if demand arrives three years late or utilisation settles below the base case? Does the site retain value because of its land, permits, grid connection or access to water? Under this lens, the second user, the alternative use and the residual value of the site become part of the investment case. Legacy mines illustrate the point. An asset developed for one purpose can acquire a second life as an industrial location, an energy-storage site or a source of recoverable materials. The commodity forecast that justified the mine may have expired. The land, water, grid connection and permissions have not. The question shifts from “will the forecast be right?” to “how much of the A renewable energy project without transmission cannot sell its power. A data centre without reliable electricity cannot process data. An industrial facility without access to water, logistics or permits may never operate at its intended scale. Value concentrates at the constraint. Brookfield cites estimates that annual investment in electricity grids will need to exceed USD 600 billion by 2030, with interconnection queues in major markets already approaching a decade.¹ The bottleneck is the business: where infrastructure returns now concentrate asset's value survives if it is not?” The valuable asset may be the bottleneck
~USD 7tn
>USD 600bn
Total capital required, the AI value chain slice. Over the decade
Grid investment. Annually by 2030
The asset that survives the forecast
Source: Brookfield 2026 outlook.
Capital is returning to pipes, wires, steel and land, but not indiscriminately. Taken together, the perspectives of infrastructure investors including Brookfield, Macquarie and Stonepeak point to a more demanding underwriting question than confidence in the base case: Will the asset still create value if the assumptions used to justify it prove wrong? And can its financing survive the wait? The test The data centre debate provides the clearest example. What makes a data centre investment resilient, as Stonepeak’s Michael Dorrell argues, is the quality of the customer, the strength of the location, the adaptability of the asset and a financing structure capable of absorbing periods of weaker utilisation.² The point is not that demand forecasts will always be correct. It is that the asset and its capital structure should not depend on every forecast being correct from the first day of operation. Brookfield makes a related case through its emphasis on embedded resilience. Macquarie argues that future infrastructure returns will depend more on income, earnings growth and operational improvement than on continued expansion of valuation multiples.¹·³ Investors are interested in the return an asset produces when the forecast is right and increasingly, how much value remains when it is wrong.
As capital returns to the physical economy, investors are paying closer attention to the decisions made at the drawing board. Over the past two decades, the market's most highly valued businesses were those that could scale without proportionate investment in physical assets. Software could be developed once and distributed globally. The prize was scale without steel.
Robert Casamento Global Strategy Executive Across AI, Energy and Climate
10 year interconnection queue
USD 600 bn grid investment gap
18 | GHD | Nexus Magazine
Nexus Magazine | GHD | 19
Made with FlippingBook - professional solution for displaying marketing and sales documents online