The definition of infrastructure is expanding rapidly as digitalisation and the energy transition create new investment opportunities. PIC’s Radim Radkovsky explains why investors increasingly need to look beyond the physical asset to the cash flows and underlying risks.
Airports, wind farms and data centres are readily recognisable as infrastructure investments. But what about graphics processing units (GPUs) housed inside a data centre, a sports stadium, a crematorium – or even a salmon farm?
For Radim Radkovsky, senior credit analyst at Pension Insurance Corporation (PIC), the boundaries of infrastructure are increasingly being determined by the underlying economic characteristics of an investment than the physical asset underlying it.
“Twenty years ago, infrastructure was a niche market, now it’s becoming an established asset class,” he says. “I saw it growing from classic core infrastructure assets – airports, ports, railroads, bridges, utilities – into renewables, energy transition assets and digital infrastructure. The definition of infrastructure has been broadening as new asset classes are discovered.”
PIC has invested more than £15 billion in UK housing and infrastructure to date, predominantly through debt. For a leading DB pensions buyout insurer, the value of its characteristics are obvious: long-term cash flows that provide a close match for liabilities extending decades into the future.
“The definition is less about the physical assets than the economic terms of these transactions,” he says. “What matters is the ability to generate stable cash flows, typically over long periods and often linked to inflation, alongside high barriers to entry and the recovery value provided by real assets.”
This approach potentially opens infrastructure portfolios to a much wider range of investments.
Core Infrastructure Summit 2026 | London | 11 November 2026
Four forces
Radkovsky identifies four structural forces – the “four Ds” of decarbonisation, demographics, digitalisation and deconsolidation – that have shaped infrastructure over the past decade and which he expects to continue driving its development.
Decarbonisation has created opportunities spanning wind and solar generation, battery storage, electricity networks and potentially hydrogen. Demographic changes support demand for social housing, healthcare, transport and other essential services, while digitalisation has brought data centres, fibre networks, telecoms infrastructure and smart meters firmly into the infrastructure universe.
Deconsolidation is somewhat different. Rather than reflecting demand for a particular service, it describes companies carving infrastructure-like assets out of their balance sheets and finding long-term investors able to provide a lower cost of capital.
Together, these trends are expanding the investable universe. But they are also changing the nature of infrastructure investment.
Renewables provide one example. Wind and solar assets were relatively novel investments two decades ago, but have become increasingly standardised and well understood. Greater competition has consequently compressed yields.
“Infrastructure is no longer a boring asset class where you can just sit on an asset and hope cash flows result,” says Radkovsky. Investors increasingly need to generate value through operational improvements rather than simply acquiring an asset and collecting a predictable yield.
The broader shift is from construction and counterparty risks towards more complicated revenue risks. Newer infrastructure can contain merchant exposure, requiring investors to make judgements about demand, market dynamics and downside resilience.
“The core challenge has shifted from construction risk to revenue risk, but understanding both risks remains critical to successful infrastructure investing,” Radkovsky says.
Institutional investment Conferences & Summits from Longview Networks
Testing the boundaries
Nowhere are the changing boundaries more apparent than in digital infrastructure. While data centres have become an established part of the infrastructure universe over the past decade, supported by the growth of cloud computing, the boom in artificial intelligence is rapidly transforming the sector.
PIC continues to invest in data centre transactions and overcomes the market shift by sticking to its principles – investing only where revenues are contracted with counterparties it considers credible. “We are cash flow investors, we don’t take technology or obsolesce risks. We do our homework,” says Radkovsky.
But innovation is already pushing beyond financing the buildings themselves. Radkovsky points to GPU financing as one of the more unusual developments to have appeared in the market in recent years.
While a data centre has many of the characteristics traditionally associated with infrastructure, including a long physical life and substantial barriers to entry, GPUs have much shorter economic lives. Radkovsky nevertheless argues that financing them can have “short to medium term” infrastructure characteristics where there is a contracted revenue stream and sufficient visibility over those revenues.
Other investors have pushed the definition in different directions, with assets ranging from stadiums to fisheries being considered infrastructure investments.
Such examples demonstrate why the infrastructure label alone increasingly says relatively little about the risk an investor is taking. The common thread is instead the underlying economics – and particularly the predictability and durability of cash flows.
Risk and resilience
Changing infrastructure also requires more sophisticated financing. “In the past, investors had been effectively taking revenue risk and counterparty risk – almost like a bond investment,” says Radkovsky.
But as revenue structures become more diverse, transactions require increasingly bespoke financing solutions; greater innovation should not come at the expense of investor protection.
The same principle applies to construction risk. PIC is prepared to accept it, but only where the risk can be appropriately mitigated through experienced and financially strong contractors, appropriate contractual arrangements and robust oversight.
He emphasises that the objective is to place individual risks with the parties best equipped to manage them.
This is illustrated by the Haweswater Aqueduct Resilience Programme, which PIC has highlighted as an example of private capital financing essential infrastructure. Radkovsky points to its clearly defined regulatory framework and allocation of project risks.
Once built, the asset itself should carry relatively limited operational risk; the risk lies in getting there. The lesson, Radkovsky argues, is broader than Haweswater: “Good infrastructure finance starts with good project design.”
The investment requirement is also evolving as governments and businesses place greater emphasis on security. “Before Covid-19, we invested a lot in efficiency. Now we need to invest more in resilience,” Radkovsky says.
Energy provides an obvious example. Increasing renewable generation can create greater volatility in electricity supply, requiring investment both in transmission and distribution networks and in storage capable of balancing the system.
Asked where he sees opportunities within the energy transition, Radkovsky is emphatic: “Batteries, definitely.”
He also expects further investment in transmission networks, while established wind and solar will remain important. The next generation of renewable investment will include repowering older sites as early assets reach the end of their original lives and are replaced by more efficient technology.
Financeable assets
For all the demand for new infrastructure, Radkovsky does not see capital itself as the principal constraint. “There is a shortage of financeable projects. Not every infrastructure project which is desirable is financeable.”
The distinction goes to the heart of how PIC approaches the asset class. Governments and societies may want a new piece of infrastructure, but that does not automatically produce an investment capable of supporting institutional capital.
Asked what ultimately makes a project financeable, Radkovsky’s sticks to his core principle: predictability of cash flows.
PIC’s internal framework formalises this in practical terms. Investors need to understand who ultimately pays for a service, whether revenues are contracted, regulated or exposed to demand, and whether risks have been allocated to those best able to manage them.
Government therefore has an important role, but Radkovsky sees this principally as providing the conditions within which private capital can operate.
“The best work any government, whether the UK government or any Western government, can do is build a stable regulatory and legal frameworks to support the inflow of private capital into the infrastructure,” he says.
Political and regulatory risk cannot be removed entirely from an asset class so closely connected to essential services. What matters is whether investors can understand and price those risks.
This becomes more important as the infrastructure universe expands. The physical assets may be changing and financing structures becoming more innovative, but PIC’s underlying requirements remain relatively conservative: durable cash flows, appropriate returns and risks that can be identified, allocated and mitigated.
For Radkovsky, that combination of continuity and change is what makes today’s infrastructure market interesting. “Infrastructure is evolving – but it will stay a major part of investors’ allocations for many years to come.”
Radim Radkovsky will be speaking at PMP’s Core Infrastructure Summit on 11 November