Infrastructure in Modern Portfolios: The Allocation Case
the hoard grows cold that is not set to flow.
Wise hands lay treasure where the great roads run —
in bridge and wire and water, where it grows.”
Infrastructure has graduated from a niche allocation to a core one because it does something few asset classes can: it produces long-duration, inflation-hedged cash flows from essential, often monopolistic assets, with low correlation to the economic cycle. For an institution matching long-dated liabilities, that combination is not a luxury but a structural fit — and it explains why, through a difficult few years for private markets, infrastructure allocations have kept rising while conviction in the asset class has held.
Why it earns its place
The allocation case rests on a handful of properties that are rare in combination. Infrastructure assets — toll roads, transmission, water systems, contracted power, digital infrastructure — generate predictable, long-dated cash flows, frequently linked to inflation through regulation or contract. They tend to be essential services with high barriers to entry, which insulates revenue from the cycle. And they offer a yield premium over comparable public credit — on the order of 200 to 400 basis points over corporate bonds — compensation for illiquidity and complexity that a long-horizon holder is well placed to harvest.
For a pension fund or insurer running an asset-liability framework, this is close to an ideal building block: a real, cash-yielding, long-duration asset that dampens portfolio volatility and hedges the liability side. That is why infrastructure behaves less like an opportunistic bet and more like a permanent allocation — and why, once an institution establishes a target, it tends to raise it rather than retreat.
Weighted target allocations. Actual allocations sit ~100 bps below target, and 56% of investors report being under-allocated. Source: Hodes Weill / institutional allocations survey, 2025.
The asset class has scaled and matured
The numbers now describe a mature asset class rather than an emerging one. Private-infrastructure assets under management reached roughly $1.3 trillion in 2024, having tripled over the preceding decade. After a rate-driven slowdown in 2023-24, fundraising snapped back to a record: closed-end infrastructure funds raised close to $300 billion in 2025. Dry powder, meanwhile, has fallen to around 23% of AUM from about 35% in 2020 — a sign not of weakness but of deployment, as managers put capital to work.
global private-infrastructure AUM, roughly tripled in a decade. Infrastructure is no longer an alternative at the margin — it is a core institutional allocation with its own fundraising cycle, its own megafunds, and its own concentration dynamics.
Approximate; 2019 interpolated between decade endpoints. Source: Preqin / BCG (2024 AUM ~$1.3tn, ~3x over the decade).
The maturation shows in the allocation data too. Target allocations have climbed — to a weighted average near 5.9%, with private pensions around 7.7% and insurers around 4.8% — and infrastructure has grown faster than any other alternative asset class over 2020-24. Crucially, most investors remain under-allocated relative to their own rising targets, which points to a structural bid that persists regardless of the near-term rate cycle.
The two engines: transition and compute
What is pulling capital in now is not the classic toll-road story but two mega-trends that have proven largely immune to the wider private-markets slowdown. The first is the energy transition: energy and environment already account for roughly half of infrastructure portfolio assets, and the capital needs of decarbonisation are structural and policy-backed. The second, and newer, is digital — specifically the physical build-out for artificial intelligence.
The scale of the compute build is reordering the asset class. BlackRock acquired Global Infrastructure Partners for $12.5 billion in 2024 and, through the AI Infrastructure Partnership with Microsoft, Nvidia, and MGX, has been assembling capital to fund data centres at a scale measured in tens of billions — the partnership aims to mobilise $30 billion of equity and up to $100 billion including debt. Its acquisition of Aligned Data Centers at roughly $40 billion was the largest digital-infrastructure deal on record, and total AI- and data-infrastructure M&A exceeded $70 billion in 2025, with participants expecting that to double in 2026. Data centres are being repriced from a specialist niche into core infrastructure, precisely because they combine power, land, and long-dated contracted demand.
The fee-bearing-capital lens
Seen from the manager’s side rather than the allocator’s, the same growth is a story about fee-bearing, often permanent, capital — the most valuable kind an alternative-asset manager can hold. The consolidation is telling: traditional asset managers and private-equity firms have been buying infrastructure GPs outright (BlackRock/GIP the largest), converting one-off fund economics into durable, scaled platforms. The result is concentration — the top 10 managers took 44% of 2025 commitments — and a widening moat for the megafunds that can write multi-billion-dollar equity cheques into power and compute.
This is where the asset class meets the manager economics that define modern alternatives: the prize is not a single fund’s carry but a compounding base of management-fee-bearing AUM, increasingly backed by perpetual-capital vehicles and, as the next note argues, by insurance balance sheets. Infrastructure is attractive to allocators for its cash flows; it is attractive to managers for the durability of the fee stream those cash flows support.
The investment read
The bid is real and under-filled
Rising targets, 56% of investors under-allocated, and two policy- and demand-backed mega-trends (transition and compute) point to a durable capital inflow that outlasts any single rate cycle. This is the sturdiest part of the case.
Crowding at the top
The top 10 managers take nearly half of commitments, and mega-deals compress entry multiples at the large end. Scale advantages are real, but so is the crowding — differentiated returns are harder to find where the most capital is pointed.
Rates, repricing, and the AI question
The asset class repriced with rates after 2022, and the denominator effect still lingers. The newest risk is the AI build itself: digital-infrastructure capex is running ahead of demonstrated, durable demand, and a compute air-pocket would land first on the newest core allocation.
The Insurance–Infrastructure Convergence (C2) — the largest single pool of long-duration, liability-matched capital feeding this asset class.
The Cost-of-Capital Gap (G10) — where this capital is scarcest and most expensive, and where the allocation case meets its hardest test.
The Net-Zero Arithmetic (G18) — the transition capex this capital is meant, in large part, to fund.
The Power–Compute Nexus — the demand side of the digital-infrastructure bid now reordering the asset class.
Infrastructure earns its growing place in institutional portfolios on merit: long-duration, inflation-hedged, essential-service cash flows that few other assets can match, at a yield premium a patient holder is built to capture. The asset class has scaled threefold in a decade, its allocations keep rising off under-filled targets, and it now sits at the centre of the two largest capital stories of the moment — the energy transition and the compute build.
For the allocator, the discipline is to separate the structural bid from the cyclical noise: the case for a permanent, rising infrastructure allocation is sound, but the entry point, the manager, and the sector matter more than ever now that capital is crowding into the same megafunds and the same AI theme. For the manager, the prize is the durable, fee-bearing capital those cash flows support — which is why the next and largest source of it is the insurance balance sheet.
but whether it will stand when winters come;
and lays a part of every hoard in things
that do not move with the market’s fever-drum.”
Leave a Reply