Energy took a quarter of US private equity deal value. Software lost two thirds.

What actually rotated?

Capital moved out of asset-light software and into heavy assets, and it moved fast. Energy accounted for roughly 25% of US private equity deal value through Q1 2026, up from roughly 6% in 2025, while technology and IT fell to roughly 17% from roughly 29% and business-to-business services held at roughly 29% (Valuation Research Corporation, Q2 2026 Equity Markets Report, reporting PitchBook data through 31 March 2026).

PitchBook gave the receiving cohort a name. Heavy Assets, Low Obsolescence, covering energy, materials, industrials and environmental services, took 31.2% of US private equity deal value in Q1 2026, against a steady share of about 14% across 2016 to 2024.

The other side of the trade is a credit event, not a valuation opinion. Software’s share of broadly syndicated loan issuance fell to 8.6% year to date in 2026 from 17.6% in 2025, its lowest since 2013 (PitchBook LCD, as at 30 June 2026). Healthcare became the largest sector for institutional loan issuance for the first time since 2015, at a record 12.4% (Sikich, Q2 2026 Credit Market Update, 13 July 2026). Software private equity deal value fell 65.7% year on year in Q2 2026 to $10.7 billion (PitchBook, Q2 2026 US PE Breakdown, 6 July 2026).

Sikich’s forward read on that is the one to underwrite: the disruption of legacy software business models increasingly looks like a structural credit story rather than a cyclical one. Do not model a software snap-back in 2027.

All of this happened inside a shrinking sponsor market. US private equity deal value fell 37.5% on the quarter to $177.3 billion in Q2 2026, the lowest since Q4 2023, while sponsor deal count rose 11.5% year on year to 2,384 (PitchBook, Q2 2026 US PE Breakdown, 6 July 2026). More deals, smaller checks, concentrated differently than a year ago.

Where are buyers concentrating, and where have they thinned?

Inside business services the spread between sub-sectors is now wider than the spread between sectors, which makes the sector average close to useless for an owner.

Environmental services is the clearest divergence in the market. June 2026 printed 38 transactions, up 23% year on year, against a decline of about 17% in overall US deal count. Across January 2025 to June 2026 there were 732 US environmental-services transactions, 57% led by private equity, and listed peers traded at a median 16.0x trailing EV/EBITDA as at 2 July 2026 (PitchBook data via Blue River Financial Group, July 2026). The demand calendar runs past the outlook period: retained drinking-water standards for PFOA and PFOS carry a 2031 compliance deadline, and landfill leachate guidelines are anticipated in 2027.

Accounting roll-ups have changed hands. Volume rose 14.8% year to date in 2026 to 62 deals, and financial acquirers took 54.8% of accounting-services M&A against 38.9% a year earlier, with sponsor volume up 69.1%, reversing the historic strategic-buyer majority. The next wave is pre-funded: sector-focused private equity capital raised rose 16.1% year on year to $12.7 billion, 87.9% of it buyout capital, with median fund size up 23.6% to $581 million (Capstone Partners, Accounting Services M&A Update, 13 July 2026). Rising median fund size pushes those buyers up-market for platforms into 2027.

Wealth management set a record. There were 120 registered investment adviser transactions in Q2 2026, the most active second quarter in the industry’s history and up 17.6% year on year, on $378 billion of transacted assets, with 80% involving strategic buyers and only 13 deals, 10.8%, coming from buyers with two or fewer prior acquisitions. Echelon projects roughly 500 deals for full-year 2026 (ECHELON Partners RIA M&A Deal Report Q2 2026, via PLANADVISER, 28 July 2026). A seller in that market is negotiating against a counterparty that has done this twenty times.

Not everything in business services participates. HVAC and facilities recorded 92 transactions year to date in 2026, down 4.2%, with multiples settling at 9.5x EV/EBITDA across 2024 to date against 13.3x in 2021 to 2023 (Capstone Partners, HVAC Services M&A Update, 27 July 2026). That is close to a four-turn reset, and it reads as structural rather than as a window that reopens. PwC expects business-services deal volumes to decline in 2026 outright (PwC, Global M&A trends in industrials and services: 2026 mid-year outlook, 23 June 2026).

What is 2027 actually funded to do?

Two things in this outlook are funded rather than forecast, and both land in 2027.

The first is power. US data-centre power demand is projected to climb from 31 GW in 2025 to 41 GW in 2026 and 66 GW in 2027, taking US capacity to roughly 95 GW by the end of 2027. Annual capacity additions are projected at 13.6 GW in 2026 and 36.3 GW in 2027, against 6.4 GW realized in 2024 and 8.5 GW in 2025. Data centres’ share of total US peak summer power demand rises from 4.1% in 2025 to 5.3% in 2026 and 8.5% in 2027 (Goldman Sachs Research, US Data Center Power Demand Projected to Double by 2027, 20 May 2026). The same authors apply their own haircut: historically about 72% of scheduled capacity goes live on time, and they adjust to roughly 60% over the next year and 50% over two.

A 4x step-up in annual additions is the demand curve under every power, grid, cooling, interconnection and site-services thesis in the market. It reaches the middle market as services and equipment rather than as generation: electrical contracting, power equipment, cooling, connectivity, engineering, environmental consulting and industrial supply chain. One HVAC operator reported commercial orders up 35% with data-centre orders up more than 500% in a single quarter (Capstone Partners, 27 July 2026).

The second is defence. The FY2027 US defence budget request totaled roughly $1.5 trillion, a 44% increase year on year, and for FY2026 the Pentagon raised training investment 7.8% to $159.7 billion. Training and simulation M&A has already responded, with volume up 133.3% year to date in 2026 to 14 transactions and sponsor dealmaking up four deals to six after two flat years (Capstone Partners, Training & Simulation Sector Update, 2 June 2026, citing the FY2027 US Government budget request). A budget request is an appropriation path into contractor backlogs, not a sentiment reading.

PwC’s forward volume forecasts line up behind both: aerospace and defence up 14% and manufacturing up 5% in 2026, with engineering and construction leading value growth at 64%, while automotive, engineering and construction and business services volumes decline (PwC, 23 June 2026, on LSEG data through 31 May 2026).

Real assets shows the same capital concentration in its purest form. Infrastructure fundraising hit a record low of $40.8 billion in H1 2026, below the $71.6 billion of H1 2024, previously the worst comparable half on record (Infrastructure Investor, 14 July 2026). Inside that half, single funds set all-time records, including a $19.2 billion flagship infrastructure close in August 2026 (Infrastructure Investor, 3 August 2026). Limited partners have not left the asset class. They have consolidated into a handful of names. For an asset owner that means a small number of very large and very motivated buyers.

Where has middle-market pricing actually moved?

Toward heavy assets, and against asset-light services. That reversal is the clearest evidence that the rotation is money rather than narrative.

The latest published middle-market print is Q1 2026. Manufacturing improved to 7.2x total enterprise value to EBITDA from 6.6x in 2025, while business services eased to 7.0x from 7.3x. In the small-deal cohort of $1m to $25m of transaction value, manufacturing firmed to 5.8x while business services eased to 5.5x (Mercer Capital, Middle Market Transaction Update, Summer 2026, reporting GF Data for Q1 2026).

Sub-sector prints in industrials moved further. Building products rebounded 28.2% year to date in 2026 to 182 transactions, with multiples averaging 10.6x from 2025 to date against 9.4x in 2024, more than a full turn (Capstone Partners, Building Products Market Update, 30 June 2026). Precision manufacturing volume rose 19.6% year on year in Q1 2026 to 61 transactions, with the average sector multiple up more than half a turn to 10.1x across 2023 to Q1 2026 from 9.6x across 2020 to 2022, and Capstone’s Precision Manufacturing Index trading at 17.1x EV to trailing EBITDA as at 31 March 2026, above the Dow Jones Industrial Average at 16.0x (Capstone Partners, Precision Manufacturing Market Update, 6 May 2026). A premium to the index is the market pricing the segment as structurally advantaged rather than cyclical.

The counter-examples matter as much. Packaging recorded 59 transactions year to date in 2026, down 21.3%, against the broader industrials industry’s 12.1% rise, with consumer-exposed demand named as a structural headwind (Capstone Partners, Packaging Market Update, 27 July 2026). And listed energy is the cheapest thing in the small-cap market, at 7.0x EV/EBITDA for S&P SmallCap 600 Energy as at 30 June 2026 against 13.5x for the S&P 600 overall (VRC, Q2 2026 Equity Markets Report, reporting S&P Capital IQ). Public energy strategics are weak-currency buyers into 2027, which leaves the field to sponsors and to newer structured capital.

Upstream oil and gas is the reminder that energy is not one market. US upstream deal value fell to $9.1 billion in Q2 2026, the third-lowest quarterly total since 2020, down 76% on the quarter, with more than 40% of it from a single federal lease sale. Enverus reads the weakness as a bid-ask problem from crude volatility rather than a demand problem and expects a materially busier second half (Enverus Intelligence Research, 5 August 2026).

What does a seller now have to prove?

That the specific business sits on the side of the market capital is rotating toward, with evidence a diligence team can test. Theme adjacency is no longer accepted as a substitute.

The clearest example is reshoring. US manufacturing construction spending peaked at $239 billion in June 2024 and has since fallen 21%, driven by a 44% collapse in computer, electronic and electrical construction from its July 2024 peak. Excluding electronics, spending has risen only 5.6% since tariffs began, which the analysis calls lower than would be expected for a boom, and the conclusion is that it is too early to call a reshoring boom in the US (IoT Analytics, Industrial Macro Pulse, 12 May 2026, on US Census survey data to Q1 2026). If a buyer’s thesis on a business is a reshoring tailwind, the macro data does not yet support it. Underwrite the specific customer program instead.

The second is the AI-exposure question, which is now a pricing input rather than a narrative. Among the hundred largest corporate transactions, AI was cited in the strategic rationale for roughly a third of deals in 2025 and 17% in the first half of 2026 (PwC, through 31 May 2026): buyers became more disciplined about where it creates durable demand and where it compresses value. PwC names AI exposure in professional services as an explicit risk, and Capstone’s stated screen for accounting platforms in 2027 is readiness and organic growth quality ahead of market size or acquisition runway.

The practical translation is that a labour-arbitrage services business is now underwritten over a three to five year horizon on whether the technology accelerates or erodes it. That answer has to exist, in writing, with evidence, before a process launches.