Floating Solar vs. Ground-Mounted Solar: How Do Construction and Operating Costs Compare

Floating solar usually costs more to build than ground-mounted PV. A 10 MW U.S. benchmark found a $0.26/W_dc, or 25%, floating-PV premium—about $2.6 million at that scale. Ground-mounted solar normally has the cost advantage when flat, inexpensive land is available. Floating solar becomes more competitive when it avoids expensive land, grading, or grid extensions. For either option, compare lifetime cost and net electricity output, not EPC price alone.^[1]

The modules themselves may be similar. Most of the cost difference comes from everything around them. Ground PV needs land preparation, foundations or piles, roads, and vegetation management. Floating PV replaces some of that work with floats, anchors, mooring, shore works, and water-side maintenance.

The answer can change completely from one site to another. Cheap, level land beside a substation strongly favors ground PV. A reservoir beside existing electrical infrastructure, especially where nearby land is expensive, can narrow the gap.

How Large Is the Construction Cost Difference?

One useful like-for-like reference is NREL's Q1 2021 floating-PV cost benchmark. In 2020 U.S. dollars, NREL modeled a 10 MW_dc fixed-tilt floating system at about $1.29/W_dc, compared with $1.03/W_dc for fixed-tilt ground PV over bare ground.^[1]

The difference was:

$1.29/W − $1.03/W = $0.26/W

At 10 MW:

10,000,000 W × $0.26/W = $2.6 million

That equals an installed-cost premium of roughly 25% under the study assumptions. Floats and anchoring were major contributors to the difference.^[1]

Use the $0.26/W figure as context, not as today's quotation. The real premium can move with plant size, module format, float sourcing, water depth, wind loading, water-level change, shoreline access, and anchoring method.

These are sensitivity cases, not average market prices. They simply show how a difference of a few cents per watt turns into millions of dollars once project capacity reaches tens of megawatts.

Why Does Floating Solar Cost More to Build?

The biggest physical difference is the structural system.

A normal ground array transfers loads through racking into driven piles, ground screws, or another foundation. FPV needs a buoyant platform, module supports, float connections, anchor points, and mooring lines that let the array move without drifting away.

NREL's 10 MW reference case assumed average conditions including about 50 m water depth, 10 m water-level variation, a 1 m swell, wind loading around 40 m/s, and a snow load of 20 psf.^[1] Those are benchmark assumptions, not design specifications for every reservoir.

A shallow, protected industrial pond may need relatively simple mooring. A deep reservoir with a large seasonal level change may require longer lines, different anchor locations, and considerably more engineering.

For a 50 MW example, suppose one mooring design adds $0.08/W and a more difficult site adds $0.16/W:

50,000,000 W × ($0.16 − $0.08) = $4 million

The plant capacity has not changed at all. That $4 million comes from the site and structural conditions.

Module choice also changes BOS quantities. Tongwei's high-efficiency module portfolio includes utility-scale formats, while products such as the TWMNF-66HD provide a reference for high-power large-format modules. The cheapest module does not automatically produce the cheapest floating system. Size, weight, electrical layout, and float compatibility matter too.

Which Ground-Site Costs Can Floating Solar Avoid?

Ground PV avoids floats and mooring, but land itself can be expensive to prepare.

Depending on the site, the EPC scope may include clearing, grading, drainage, erosion control, access roads, fencing, piles, and soil improvement.

The World Bank identifies avoided land acquisition and lower site-preparation needs as potential advantages of floating solar, especially where suitable land is scarce or has competing uses.^[2]

Suppose difficult ground adds an illustrative $0.04/W in grading, drainage, and civil work to a 40 MW land project:

40,000,000 W × $0.04/W = $1.6 million

If the floating alternative adds $3.5 million for floats and mooring but avoids that $1.6 million civil bill, the remaining gap falls to $1.9 million before land and grid costs are considered.

Land can move the numbers again. A utility-scale plant may need hundreds of acres, depending on module efficiency, spacing, tilt, tracking, roads, setbacks, and terrain. If the land is already owned and has little competing value, ground PV keeps a strong advantage. If the land must be purchased or is valuable for another use, the water surface starts to look more interesting.

The reservoir is not “free,” though. Floating projects can still need water-use agreements, leases or concessions, environmental review, shoreline access, dam-safety coordination, and anchoring approvals. Those costs belong in the model.

Existing infrastructure can matter as much as land. Reservoirs beside hydropower stations or industrial sites may already be close to roads, substations, and transmission. That does not guarantee a cheap interconnection, but it can avoid costs that a remote ground site would have to carry.

How Do Operating Costs Compare?

Ground-mounted PV usually gives technicians easier physical access. They can drive into the plant, inspect strings from land, and replace most components without boats or floating platforms.

Its recurring work may include vegetation control, road maintenance, drainage, fencing, and erosion repairs.

Floating PV removes most vegetation work under the array, but it adds components that do not exist on a conventional ground plant: floats, float connections, mooring lines, anchors, and moving water-side cable routes.

IEA PVPS identifies FPV-specific issues including humidity, wave-induced mechanical loading, soiling, biofouling, and maintenance-access challenges. Their importance varies from one site and system design to another.^[3]

Small yearly cost differences can become large over a 25- or 30-year project life.

On a 50 MW plant, every $1/kW-year equals:

50,000 kW × $1 = $50,000/year

If one design needs an extra $4/kW-year for inspections and access:

50,000 kW × $4 = $200,000/year

Over 25 years, before discounting:

$200,000 × 25 = $5 million

The $4/kW-year figure is only an example. It is not an FPV industry average. The point is that O&M should not be dismissed just because the annual number looks small next to EPC cost.

For context, DOE's Q1 2025 benchmark for a representative 100 MW_dc utility-scale ground-PV system uses O&M of $20/kW_dc-year. Its modeled market price is $1.12/W_dc, with dollar values adjusted to 2024 U.S. dollars.^[4]

At 100 MW:

100,000 kW × $20/kW-year = $2 million/year

That DOE value is a current ground-PV benchmark, not a floating-PV O&M rate. It also should not be directly combined with NREL's older 25% floating premium because the studies use different dates, assumptions, and dollar bases.

Can Higher Energy Yield Recover the Floating Solar Premium?

Water can lower module operating temperature under some conditions. Some reservoirs may also have less dust than dry land sites. Both can help output.

The problem starts when those possible benefits are turned into an automatic FPV production bonus.

IEA PVPS says floating-PV performance advantages depend strongly on system design, environment, and site conditions and cannot be generalized to every installation.^[5]

Use site-specific irradiation, temperature, module layout, tilt, losses, and operating conditions instead of simply adding 5% or 10% to the ground-PV forecast.

Suppose the ground option is modeled at 80 GWh/year and the site-specific floating model predicts 3% more output:

80 GWh × 3% = 2.4 GWh/year

At an illustrative $50/MWh:

2,400 MWh × $50/MWh = $120,000/year

If floating construction costs $4 million more and all other inputs were identical:

$4,000,000 ÷ $120,000 ≈ 33.3 years

This is not a full payback model. It ignores financing, degradation, O&M, taxes, and changing electricity value. It simply shows that a modest energy gain may not be enough by itself to recover a multimillion-dollar construction premium.

If the floating site also avoids several million dollars in land, civil work, or transmission extension, the result can change quickly.

How Can Land and Grid Costs Change the Answer?

A combined example is more useful than looking at one cost item at a time.

Assume two 30 MW alternatives. These are example figures, not supplier quotations.

If you look only at the floating structure, FPV appears $5.4 million more expensive.

Once $1.5 million of ground civil work, $1.2 million of grid extension, and $1.5 million of land cost are added to the ground option, the modeled difference falls to $1.2 million.

Now change the site. If the ground parcel is already owned, flat, and next to the substation, most of those avoided costs disappear. Floating PV again carries something much closer to its original structural premium.

This is why one universal “floating solar premium” is not very useful for investment decisions. Calculate the technology premium and site-specific savings separately.

Tongwei's utility-scale project portfolio also shows how much large PV projects differ in capacity and operating environment. Cost comparison should begin with the actual site rather than a generic plant layout.

Compare Lifetime Cost per MWh, Not Just EPC Cost per Watt

$/W is useful during procurement because it normalizes construction cost by installed capacity. It does not tell you how much usable electricity the project will deliver over its life.

A proper lifetime comparison should include:

• initial EPC cost;
• land lease, purchase or water-use cost;
• grid interconnection;
• annual O&M;
• insurance;
• major replacement and repair allowances;
• annual net generation;
• degradation;
• curtailment;
• financing;
• taxes and incentives where applicable; and
• decommissioning obligations.

IRENA reports that the global weighted-average LCOE of utility-scale solar PV projects commissioned in 2025 was about $44/MWh.^[6] That is useful market context, not a target LCOE for any particular floating or ground-mounted project.

Two reservoirs only 100 km apart can still produce very different lifetime costs if one has difficult mooring and the other has easy grid access. The same is true for ground sites with different soils, grading, and transmission requirements.

Compare both options using the same discount rate, project life, electricity-value assumptions, degradation basis, and interconnection boundary.

Which Site Usually Has the Lower Cost?

Ground-mounted PV has the clearest cost advantage when suitable land is already available, civil work is simple, and the grid is nearby.

Floating PV becomes more competitive when the reservoir solves an expensive problem such as land acquisition, difficult grading, long grid extensions, or competition for scarce land.

A project may still choose FPV for reasons other than cost. Those benefits should be shown separately rather than used to hide a higher lifetime electricity cost.

What Should Be Included in a Like-for-Like Cost Model?

Before comparing two EPC totals, make sure both quotations cover the same project boundary.

Apply the same discipline to generation assumptions. Use the same DC capacity, comparable module technology, project life, degradation convention, financial basis, and electricity-value assumptions.

If one bid includes the substation and another stops at the inverter output, the headline $/W figures are not describing the same scope.

Conclusion

Ground-mounted solar is usually cheaper when suitable land and grid access are already available. Floating solar starts with higher structural cost—the NREL 10 MW benchmark showed a $0.26/W premium—but savings in land, civil work, and interconnection can narrow or even reverse the gap. For a real project, compare the same capacity over 25–30 years using site-specific CAPEX, O&M, and net generation. The lowest EPC $/W is not always the lowest lifetime $/MWh.