Ownership, Strategic Focus, and Asset ProductivityEvidence from U.S. Electricity Generation

Electricity is the bottleneck of the AI build-out — and, in the near term, that bottleneck runs through the existing fossil fleet

Data center demand is arriving faster than new generation can be built, so the fleet already on the grid is what meets it. Since the repeal of the Public Utility Holding Company Act (PUHCA) in 2005, private equity has been acquiring fossil generation assets at scale. Who owns them — and what ownership does to how they run — has become a first-order question for strategy, energy markets, and climate at the same time.

Operational engineering is the value-creation channel everyone asserts — and much less is known about how it works

The canonical account gives private equity three levers: financial engineering, governance engineering, and operational engineering. The first two are well mapped. The third — “they improve operations” — usually enters as a residual. This paper opens that box with a specific mechanism: operating experience converts across technologically similar assets, so the portfolio’s composition determines whether know-how accumulated on one plant can be redeployed on the next.

The question originates in practice. Advising industrial companies at McKinsey, I regularly observed comparable assets performing differently under different owners; this paper examines that variation systematically.

1. What are the productivity consequences of ownership change for physical assets? Preview answer
2. How does the composition of an owner’s portfolio connect firm-level strategy to the performance of the individual assets it holds? Preview answer
3. What are the technological and geographic boundaries within which operating knowledge transfers across a portfolio? Preview answer

Six data layers, linked at the individual generating unit

Every claim in the paper rests on the same spine: the physical unit. High frequency operations, transaction-level ownership, and facility-level digital demand join at the unit level, so productivity, ownership, and demand can be read off the same asset at the same moment — for a quarter century.

Practitioners describe the mechanism directly

From semi-structured interviews with practitioners across PE-backed platforms, utilities, OEM service organizations, and advisory firms (Harvard University IRB Protocol IRB25-1112). Interviews were confidential and non-attributable; identifying details have been generalized. The first two quotes describe knowledge moving across distance within a technology class; the last describes it stopping at a technology boundary — together, the two sides of the dispersion result.

The interview record maps onto Nonaka’s (1994) knowledge-conversion framework. Focused ownership accelerates the three modes tied to unit-level efficiency; diversified ownership retains the advantage in cross-technology recombination — value that appears at the portfolio level, not in unit-level heat rates.

Adapted from Nonaka (1994), as in the paper’s interview appendix. Focused ownership accelerates socialization, externalization, and internalization — the three processes most directly linked to unit-level operational efficiency.

Operating knowledge converts across similar machines — and the sharpest evidence is who fails to produce the gains

The mechanism makes a sharp prediction about who should fail to produce the gains. If the improvements came from leverage, monitoring, or incentive design, financially oriented owners would generate them too. They do not: efficiency effects for financial owners hover at zero throughout, while operating specialists’ effects deepen steadily after acquisition. The capability, not the capital structure, does the work.

On the ground, the gains come from a repertoire of well-known, attention-intensive adjustments — not proprietary technology or heavy capital spending. One performance consultant estimated that a typical heat-rate audit surfaces 2–4% of recoverable losses, nearly all from fixes like these:

• Turbine airflow calibration — fine-tuning inlet guide vane angles to specific load profiles
• Combustion setpoints — tightening temperature margins toward optimal combustion
• Start-up sequencing — re-ordering warm-up and ramp procedures by outage type and weather
• Maintenance calibration — timing hot-gas-path inspections to each unit’s cycling intensity
• Control-system drift — catching sensor and setpoint degradation that raises no alarms
• Auxiliary-load optimization — cutting the 3–5% of gross output that plants consume internally

What separates owners is not knowing these levers exist — everyone does. It is whether the organization allocates the sustained, technology-specific attention needed to find and hold marginal gains across a fleet. A portfolio of similar units makes performance differences immediately legible: a plant running the same turbine model as its sister units but posting a higher heat rate invites a question — what is different about this one? — that a heterogeneous fleet cannot ask as precisely.

Alternative explanations, and the evidence that bears on them

Long-lived physical assets, discretionary operating practices, and active financial engineering make several rival readings plausible from the outset. The paper takes each seriously, and the verdicts are deliberately calibrated: “ruled out” only where the measurement makes the alternative impossible, more guarded where the evidence narrows the space without closing it.

No single alternative — and no plausible combination — reconciles flat pre-trends, within-unit gains in a balanced panel, persistence through six years, survival within fuel type, and a measurement technology immune to reporting discretion. The remaining channels read as complements to the knowledge-convertibility mechanism, not substitutes for it.

The productivity of a physical asset may not be intrinsic to the asset

A supermodular model of capability, experience, and focus

To explain the heterogeneity, the paper develops a supermodular model in which the owner’s operating capability governs the complementarity between technology-specific experience and portfolio focus. Experience raises productivity most where the portfolio concentrates on technologies the owner knows; without operating capability, the complementarity is inert. The model turns the mechanism into predictions — gains that grow with the experience–focus interaction, concentrate among operating specialists, and stop at technology boundaries — each borne out in the evidence above.

1. What are the productivity consequences of ownership change for physical assets?

   Acquired units improve thermal efficiency by 2–5% with clean pre-trends, and their hazard of retirement falls by roughly three-quarters. Ownership change alters both how an asset runs and how long it lasts — twin outcomes that cut in opposite climate directions.

2. How does the composition of an owner’s portfolio connect firm-level strategy to the performance of the individual assets it holds?

   Through convertibility. The gains concentrate almost entirely among operationally focused specialists whose portfolios are technologically concentrated: experience pays where the portfolio makes it transferable. Financially oriented owners holding similar assets show no comparable improvement.

3. What are the technological and geographic boundaries within which operating knowledge transfers across a portfolio?

   Operating knowledge transfers within, but not across, technology classes — and it travels across geographically dispersed same-technology portfolios. The binding boundary is technological, not spatial.