Every decision about a solar asset eventually becomes a question about one number. Does this project clear our hurdle rate? Is this portfolio worth more held or sold? Should we refinance now or wait? All of them resolve to the internal rate of return, and for most owners the solar IRR they underwrote at financial close quietly drifts over the asset's life, usually in the wrong direction.
That drift is the part the model does not show you. IRR is treated as a closing-day calculation, fixed once the CAPEX, the PPA, and the debt are set. In practice, a large share of the return is decided afterward, in how the plant is actually run: how fast faults are found and fixed, how much yield soiling and downtime quietly remove, how closely real availability tracks the number in the financing model.
This guide is written for asset managers and owners who want to understand what actually moves solar IRR, not just how to compute it. It covers the cash flow model behind the metric, the inputs that carry the most weight, and the operational levers that keep the delivered return closer to the underwritten one.
What IRR actually measures for a solar asset
The internal rate of return is the discount rate at which a project's net present value equals zero. Put plainly, it is the annualized return implied by the full stream of cash flows across the asset's life, from the initial outlay through 25 or more years of revenue and cost. If a solar project's IRR is 12%, the investment behaves as though it compounds at 12% a year.
Its value to an asset owner is comparability. Because IRR reduces a complex, multi-decade cash flow to a single annualized rate, it lets you rank a solar asset against alternative uses of capital and against a required hurdle rate, regardless of project size. That is why it appears in investor presentations, lender term sheets, and secondary-sale valuations alike.
Two versions matter in practice. Project IRR (or unlevered IRR) measures the return on the total capital invested, before debt. Equity IRR (or levered IRR) measures the return to equity holders after debt service, and it is typically higher because leverage amplifies the equity return. When someone quotes a solar IRR without saying which one, that is the first question to ask.
The cash flow model behind solar IRR
Calculating solar IRR means building the annual cash flow stream the metric discounts. For a utility-scale asset the model has six moving parts.
Initial investment (CAPEX). The all-in build cost: modules, inverters, trackers, balance of system, land, grid connection, and development. This is the year-zero outflow.
Revenue. For a contracted plant, annual generation multiplied by the PPA price; for a merchant plant, generation multiplied by captured market prices. Generation is where operational reality enters the model, because it is a function of irradiance, the performance ratio, and availability, not just nameplate capacity. The solar performance ratio is the single metric that connects design expectations to delivered output, and the PPA structure determines how that output converts to revenue.
Operating costs (OPEX). O&M, asset management, land lease, insurance, and a decommissioning provision. O&M is a thin but relentless line item. NREL's Annual Technology Baseline puts fixed O&M for utility-scale PV at roughly $24 per kW-AC per year, and unlike CAPEX it recurs, with inflation, every year of the model.
Degradation. Output falls slightly each year. NREL's fleet research finds a median module degradation rate near 0.5% per year, with a fleet-wide performance loss of about 0.75% per year, and NREL notes this rate feeds into nearly every solar financing agreement. Small as it sounds, it compounds across 25 years and steadily reduces late-life revenue.
Incentives and tax. Investment tax credits, accelerated depreciation, and any local mechanisms. These front-load cash flows into the early years, which mathematically lifts IRR because early cash is discounted less.
Terminal value. A residual or salvage value at end of life, often modeled conservatively, sometimes as a repowering or refinancing event.
With those flows laid out year by year, IRR is the rate that sets their discounted sum to zero. Any spreadsheet or financial model solves it iteratively. The arithmetic is not the hard part. Choosing inputs that reflect how the asset will actually perform is.
The inputs that move solar IRR most
Not every input carries equal weight. Three dominate the sensitivity of the result.
The first is revenue timing and certainty, driven by the PPA price and contract terms. A fixed-price offtake stabilizes the entire model; merchant exposure widens the range of outcomes in both directions.
The second is the cost of capital, which sets the hurdle the IRR is measured against. A project returning 9% is attractive under a 7% hurdle and unfinanceable under an 11% one. IRR is never judged in isolation; it is judged against what the capital costs.
The third, and the one owners most often underestimate, is realized generation over the asset's life. This is where the model meets the meter. Two plants with identical CAPEX, PPA, and financing can deliver materially different IRRs purely because one is run better than the other. That gap is not a financing variable. It is an operational one.
Why operational performance is a financial lever
Here is the point most IRR discussions miss. Once an asset is built and financed, CAPEX and debt are fixed. The one input still moving, every single day, is delivered generation, and that is governed by how the plant is operated.
Two operational metrics translate directly into the revenue line of the IRR model. Availability measures the share of time the plant is able to produce. NREL's PV fleet analysis found median system availability of 0.99, but a lower-decile (P90) value of 0.95, meaning the weaker plants lose around 5% of potential output to downtime that never appears on a nameplate. Performance ratio captures how much of the available irradiance actually becomes sellable energy after soiling, temperature, mismatch, and inverter losses.
Both erode quietly. A string outage, a drifting tracker, a soiled array, or an inverter fault does not announce itself; it shows up as a slightly lower monthly generation number that, in a slow operating model, is noticed weeks after the loss began. Every week of elevated mean-time-to-resolution is revenue that is gone the moment the sun sets. And because that lost revenue lands in the early and middle years of the model, where discounting bites least, it hits IRR harder than the same loss would late in life.
This reframes O&M. It is not a cost to minimize against the IRR model; it is a lever inside it. The shift from a reactive posture to a predictive one is, in IRR terms, a decision to protect the revenue line rather than let it leak.
A worked view: how a performance gap moves the return
Consider an illustrative 100 MW plant generating about 180,000 MWh a year under a $40/MWh PPA, roughly $7.2 million in annual revenue. On those assumptions, every 1% of lost yield is about 1,800 MWh, or around $72,000 a year.
A plant operating at the lower-decile 0.95 availability is losing on the order of 5% of output, close to $360,000 every year, before degradation or soiling are counted. Recover even half of that through faster detection and resolution and you add roughly $180,000 to annual revenue, straight to the cash flow line the IRR model discounts. Sustained across the asset's life, a persistent two-to-three-point gap between run-well and run-poorly is not an operational footnote. It is a visible move in the equity IRR your investors were promised.
The numbers above are illustrative, chosen to show the mechanism rather than to benchmark any specific plant. The logic holds at any scale: the return is decided as much in the field as in the financing.
How Areg AI connects operations to IRR
Most asset owners run the financial model and the operational reality in two separate systems. The model lives in a spreadsheet updated quarterly; the plant's actual performance lives in a monitoring tool that flags problems but does not price them. The gap between those two is where IRR quietly slips.
Areg AI is built to close it. The Financial Dashboard tracks performance ratio, technical and contractual availability, and PPA KPIs, including guaranteed minimums and liquidated-damages exposure, and turns that operational data into financial insight rather than leaving it as raw telemetry. The Forecast & Statistics engine models expected yield against actual output, so a developing shortfall is surfaced while it is still cheap to correct, and a live digital twin keeps the whole site visible in real time.
The point is not another dashboard. It is that when detection is wired to resolution, a fault flagged in the morning becomes a work order and a dispatch, not a line in a report someone reads next week. That is the operational difference that keeps delivered generation, and therefore delivered IRR, close to the number in the model.
The bottom line
Solar IRR is not a figure you set once at financial close and then watch. The closing-day calculation is a projection. The realized return is earned, or lost, over 25 years of operating decisions, in how quickly problems are found and how completely they are fixed.
The owners who protect their returns are the ones who stop treating the financial model and the operating plant as separate things. Underwrite the IRR carefully. Then run the asset as though the return depends on it, because it does.
To see how operational data maps to financial performance in real time, book a demo.
FAQ
What is a good IRR for a solar project?
There is no single figure, because a good IRR is one that clears the project's cost of capital for its risk level. A merchant plant carries more price risk than a contracted one and is expected to return more; a leveraged equity IRR runs higher than an unlevered project IRR on the same asset. The right benchmark is always your own hurdle rate, not a headline number.
What is the difference between project IRR and equity IRR?
Project IRR (unlevered) measures the return on total capital invested before any debt. Equity IRR (levered) measures the return to equity holders after debt service. Because debt is typically cheaper than the project return, leverage lifts the equity IRR above the project IRR. Always confirm which one a quoted figure refers to.
How is solar IRR different from ROI or payback period?
ROI is a simple ratio of total gain to cost and ignores timing. Payback measures only how long until you recover the outlay. IRR accounts for the timing and size of every cash flow across the full asset life, which is why it is the standard for comparing solar assets against other investments and against a cost of capital.
Which inputs affect solar IRR the most?
Revenue certainty (PPA price and terms), the cost of capital it is measured against, and realized lifetime generation. The first two are set at financing. The third keeps moving, through availability, performance ratio, and degradation, for the life of the asset.
Can improving O&M actually change IRR after a plant is built?
Yes. Once CAPEX and debt are fixed, delivered generation is the main remaining variable, and it is governed by operations. Faster fault detection and resolution recovers yield that would otherwise be lost, and because that revenue lands in early and mid-life years where discounting is lightest, it has an outsized effect on IRR.
What degradation rate should I use in a solar IRR model?
NREL's fleet research supports a median module degradation rate near 0.5% per year, with a fleet-wide performance loss closer to 0.75% per year. Hotter climates degrade faster. Use a rate consistent with your technology and site rather than a generic default, since it compounds across the full model.