Global infrastructure fundraising hit nearly $200 billion in 2025 — a record, up roughly 60% on 2024. North American fundraising rose about 285%. A further $474 billion is currently being sought by closed-end infrastructure funds, with $143 billion of that concentrated in the top ten managers. Against that, deal count fell about 24% in 2025 while total deal value rose 23%. The capital is up. The deal count is down. The cohort below the top ten managers is about to discover what sub-deployment risk feels like.

The number behind the number

Infrastructure dry powder as a share of AUM has fallen from 35% at end-2020 to 24% at end-2024 to roughly 23% by mid-2025. Read one way, that is a market clearing — capital is being deployed. Read another way, and this is the more useful reading: that drop happened while fundraising was at record highs. The numerator (deployed capital) rose; the denominator (total committed capital) rose faster. The ratio improved because deployment kept pace, but a growing absolute pool of dry powder is now chasing a shrinking pool of deals.

$335 to $400 billion of infrastructure dry powder. 24% fewer deals than in 2024. 23% more capital per deal that does happen. The math is straightforward and the consequence is structural: GPs that can credibly compete for the top ten deals are deploying. GPs that cannot are chasing what's left, and what's left has worse selection, slower close cycles, and higher attrition.

This is the second-order problem most LP-facing pitch decks haven't caught up to yet. The fundraising story is "data centres, energy transition, record allocations." The deployment story is "the top ten are eating the headline transactions, the next forty are running deal-flow processes that were built for a slower market, and the rest will write 'difficult deployment environment' into their 2027 LP letters."

What changed

Three forces compressed deal flow in 2025-26.

The first is hyperscaler bilateralism. AWS, Microsoft, Google, Meta, and Oracle increasingly transact directly with technology and developer counterparties, bypassing the financier intermediation layer for projects of strategic importance. AWS's 17-year Talen Susquehanna PPA is the canonical example. Oracle's 2.3 GW gas BTM deal with VoltaGrid and Energy Transfer is another. These are not deals that ever appeared in an infrastructure fund pipeline; they are off-grid balance-sheet plays at hyperscaler scale. The volume they represent — call it 15-25 GW of US deployment in 2025-26 — is volume that infrastructure funds did not get to compete for.

The second is concentration at the strategic capital layer. Blackstone's $1 billion strategic equity investment into VoltaGrid alongside Halliburton in early 2026 is a marker. The largest pools of capital are not competing in the conventional energy-infrastructure auction process; they are taking strategic positions in technology-developer hybrids that will then deploy into off-grid hyperscaler load. This pulls equity out of the conventional pipeline — and pulls debt with it, because conventional infrastructure debt follows conventional infrastructure equity.

The third is regulatory reset. The One Big Beautiful Bill Act of July 2025 phased out the wind and solar ITC/PTC for projects not in construction by 5 July 2026, retained geothermal/nuclear/hydro/battery storage, and imposed new Foreign Entity restrictions for the 2026 tax year. The composite effect was a sharp pruning of the developer pipeline. The deals that used to be in the queue are not all in the queue anymore. The deals that remain are concentrated in narrower technology categories, narrower geographies, and narrower counterparty profiles — easier to identify, harder to win.

These three forces don't reduce the addressable market for infrastructure capital. They reshape it. The total tonnage of energy infrastructure that needs financing over the next five years remains roughly $1.5-2 trillion globally on most credible estimates. The change is in where that capital flows through and who structurally has access to it.

What deal flow velocity actually measures

Deal flow velocity is a composite metric — the variable a sophisticated LP should be asking about and which most GP reporting doesn't isolate. Three components:

Sourcing rate. New qualified deals reaching the fund per quarter, where "qualified" means the deal could plausibly close under the fund's mandate. Most GPs report deal flow as raw count of memos received; the relevant number is the share of those memos that pass the first qualification gate. In current market conditions, that share is commonly 5-15% for mid-market infrastructure funds; the rest are mandate-misaligned, geography-misaligned, technology-misaligned, or counterparty-deficient.

Conversion rate. Qualified deals reaching financial close, as a share of qualified deals entering due diligence. The market average across structured energy infrastructure deals in 2025-26 is sitting at 30-40% — meaning 60-70% of deals that pass first qualification still fall over before close. Term sheet attrition is concentrated in the qualification work that wasn't done at the front end.

Time to close. Average elapsed time from first qualified meeting to financial close. The market median across non-hyperscale energy infrastructure deals is sitting at 9-14 months. Hyperscale bilaterals close in 4-8 months. Funds that close in 6-9 months on quality deals deploy three to four times the capital per quarter that funds running 12-18 month cycles do.

Velocity = sourcing rate × conversion rate × (1 / time to close). It is not a marketing metric. It is the variable that determines whether your fund deploys $400 million in three years or $100 million in three years against the same dry powder. The funds doing the first will return capital and re-fundraise; the funds doing the second will not raise their next vintage.

Why the matching problem is the velocity problem

Across all three components — sourcing, conversion, time — the binding constraint is the same: the qualification gap between what comes through the broker network and what would actually close.

A typical mid-market infrastructure fund in 2026 sees 200-400 deal memos per year. Of those, maybe 40-60 pass first internal qualification. Of those 40-60, perhaps 15-20 enter formal diligence. Of those 15-20, 4-7 close. The funnel narrows by a factor of 50-100 from memo to close. Most of that attrition happens because the matching between deal characteristics and fund mandate was never structured at the front end — it was assumed.

This is the financier-side mirror of the matching problem that tech companies and site developers face. A tech company with a working megawatt fails distribution because the developers it meets are not pre-qualified. A site developer with a controlled site fails offtake because the counterparties it meets are not pre-qualified. A financier with a defined mandate fails deployment because the memos it sees are not pre-qualified against that mandate.

The traditional response is to hire more deal team. More analysts, more associates, more time spent qualifying memos that should have been qualified before they arrived. This is linear scaling against a non-linear problem. The funds that have built — or rented — a matching capability are sourcing already-qualified deals; the funds that have not are paying their deal team to do the matching work the market should have done upstream.

The cost of sub-deployment

A $400 million infrastructure debt fund with a 5-year deployment window that deploys $80 million in years one and two is sitting on $320 million of opportunity cost against its target return. At a 10% target IRR, the opportunity cost on the un-deployed portion is roughly $32 million per year of foregone return — net of management fee drag, the realised LP-level return on the vintage will land somewhere between 4% and 6% if the deployment pace doesn't recover.

That fund will not raise its next vintage. The LPs running its 2027 re-up review will see the deployment graph and pass. The general partners will tell themselves the deals weren't there, but the deals were there — they went to other funds.

This is the consequential fact. Record fundraising plus declining deal count plus concentration at the top of the market means a meaningful share of 2024-26 vintages will end up with deployment pace problems. The funds that survive will be the ones that built sourcing infrastructure to compete; the funds that don't will end the cycle with one published vintage, no follow-on, and a partners table that finds new positions.

Mandate clarity as a competitive moat

A mandate that is articulated only at the LP-disclosure level — "infrastructure debt focused on energy transition in North America" — is not a mandate; it is a category. Mandates that drive deployment velocity are operational, not categorical:

Geographic specificity. Not "North America" but "PJM, ERCOT, CAISO, and emerging Australasian markets with sovereign offtake support."

Technology specificity. Not "energy transition" but "firm-power technologies with bankable performance data: geothermal, advanced storage, behind-the-meter gas with hydrogen-ready conversion paths."

Structural specificity. Not "infrastructure debt" but "senior secured, 7-12 year tenor, 6-9% coupon, 60-70% advance rates against contracted offtake revenue."

Counterparty specificity. Not "investment-grade counterparties" but "named counterparty profiles by sector, including non-rated counterparties with credit support structures we have pre-cleared."

A fund whose mandate is articulated at this level can be filtered for by a matching layer. A fund whose mandate is articulated at the category level cannot — and will be sent everything, including the 85-90% that doesn't fit. The work of articulating the mandate operationally is the work of being eligible for high-quality deal flow under the current market structure.

What this means for your fund in 2026

Three operational changes worth considering this quarter:

Audit your deal flow funnel quantitatively, not anecdotally. Pull the actual numbers — memos received, qualified, in diligence, closed — by quarter for the last six quarters. Most fund management teams operate on a felt sense of deal flow; the actual numbers are usually worse than the felt sense, and the trend is usually worse than the level. The audit will tell you whether the gap between dry powder and deployment is sourcing, conversion, or time — they require different fixes.

Re-articulate the mandate operationally. Write the geography, technology, structure, and counterparty specificity in a single document and circulate it to your active deal-flow channels. The cost is low. The result is that the channels that were sending you mandate-misaligned memos either start filtering or stop sending; either is an improvement.

Decide build vs. rent on the matching capability. Building it internally requires a sustained deal team investment, a multi-year sourcing flywheel, and the willingness to absorb the cost in the years before it pays back. Renting it requires identifying counterparties who have already built it and structuring the access. Both are valid. Doing neither is the option that produces the sub-deployment outcome the rest of this piece described.

What changes for the financier business model

The structural change underway is that the infrastructure financier earning excess return in the late-2020s will be the one with privileged access to qualified deal flow, not the one with the largest balance sheet. Balance sheet matters at the top of the market, where the top ten managers are bidding into auction processes for the largest deals. Below that — where the bulk of the deployable capital actually sits — the differentiator is whether the deal flow is pre-qualified or not.

The matching layer is the infrastructure that provides that pre-qualification at scale. It is what allows a mid-market infrastructure fund with a $400 million mandate to deploy that mandate inside its window, against the right risk profile, at the right velocity, without burning down the deal team on memos that were never going to close. It is the operational answer to the sub-deployment problem that the next eighteen months will surface.

That answer is being built. The funds that engage with it during construction will have access on terms that the funds that arrive in 2028 will not.