How Deriva Raised $820 Million of Construction Debt in Nine Days

Deriva Energy signed two construction loans nine days apart in October 2026, worth $820 million between them. For the two years before that, every financing Enerdatics records for the company refinanced something already built. The two also priced very differently per megawatt, and the reason has little to do with technology.

Key takeaways

• Deriva Energy secured $540 million of debt on 21 October 2026 for the construction of the Sullivan solar project in Sullivan County, Indiana, with Royal Bank of Canada acting as both administrative agent and collateral agent.
• The lending group comprises Royal Bank of Canada, DNB Bank, Mizuho Bank, NatWest and Wells Fargo, with Skadden advising Deriva Energy and Milbank advising the lenders.
• Enerdatics records the Sullivan project at 350 MW, connecting to the Gibson to Merom 345 kV transmission line in MISO through two separate queue positions, J1691 and J1786, both with interconnection agreements already executed.
• Nine days earlier Deriva raised $280 million from Canadian Imperial Bank of Commerce and KeyBanc Capital Markets for the 302 MW Frontier battery in Oklahoma, which Oklahoma Gas and Electric will own on completion under a build-transfer agreement.
• Enerdatics records 80 US solar debt financings in 2026 totalling $31.41 billion, against 97 in 2024 totalling $21.45 billion, so deal count has fallen while average size has risen from about $221 million to about $393 million.

Why did Deriva's borrowing change character in October?

Because it stopped paying for the past and started paying for the future. Enerdatics holds five Deriva Energy financings. The three before this month were all raised against assets already generating: $141 million from MUFG Bank in September 2025 for the long-term operation of three solar projects, a $127 million green bond in May 2025 refinancing the Ledyard wind and Pisgah Ridge solar projects, and $49 million of senior secured notes from Prudential Financial in February 2025 for the North Allegheny wind project.

Those are balance sheet exercises. Refinancing an operating asset replaces expensive or maturing debt against cash flows that already exist, and it adds nothing to the portfolio. The two October facilities do the opposite. Both fund construction of assets that produce nothing yet, and together they commit more than two and a half times the capital of the previous three combined.

The timing is tight enough to be deliberate. A sponsor that brings two bank groups to close within nine days has run both processes in parallel for months. Deriva holds nearly 7 GW of operating assets across the major US power markets, so this is not a company short of scale. It spent two years optimising what it owned and has now returned to building.

What makes one loan so much heavier per megawatt than the other?

Who owns the asset when the builders leave. Against the 350 MW Enerdatics records for Sullivan, the $540 million facility works out at about $1.54 million per MW. The Frontier facility works out at about $0.93 million per MW across 302 MW. The record also states a 400 MW alternating current rating, and that denominator gives about $1.35 million per MW, so the figure depends on which rating is applied. On either reading Sullivan carries materially more debt per megawatt.

Technology accounts for some of that, since a solar farm and a standalone battery have different capital costs and revenue shapes. But the structural difference is larger than the technical one. Frontier is being built for Oklahoma Gas and Electric to acquire at completion, so its loan only has to survive until handover and is repaid from the sale proceeds. Enerdatics set out that arithmetic when the facility closed, noting that a loan sized for a handover rather than a holding period produces a per-MW figure unlike the rest of the market. Sullivan has no counterparty waiting. Deriva is building it to keep, so the debt has to be sized against revenue the project will earn over many years rather than against a fixed sale price.

Set against assets their sponsors intend to hold, Sullivan looks ordinary rather than heavy. BrightNight and Cordelio Power raised $414 million for the 300 MW Box Canyon solar project in Arizona in May 2024, which worked out at about $1.38 million per MW. Sullivan sits modestly above that, two and a half years later.

Why does a project already in construction have a 2030 operating date?

The queue positions are the clue. Sullivan carries two interconnection identifiers in MISO, both with agreements executed, and connects to a single 345 kV line running between the Gibson and Merom generating stations. Two queue positions on one site usually means capacity was secured in two applications and the project assembled from them, a common route to a large connection in a congested market but one that can leave the halves energising on different schedules.

The dates are long either way. The facility closed in October 2026 against commercial operation expected in March 2030, about three and a half years for a project of this size. SOLV Energy and Burns and McDonnell are both engaged as engineering, procurement and construction contractors, an unusually heavy arrangement pointing the same way, toward a build delivered in stages rather than one campaign. Lenders pricing across that span carry construction risk far longer than a single-phase solar farm needs.

Enerdatics records 24 US solar debt financings of $400 million or more announced in 2026, and Sullivan's $540 million sits in the middle of that group rather than near the top. The largest are an order of magnitude bigger, including roughly $4.95 billion for IPX Power's 2.75 GW Darden project and about $2.62 billion for Enlight's CO Bar complex. What distinguishes Sullivan is not size but the club. Five banks for $540 million is tight, where Darden drew twenty lenders.

What does the deal signal for US solar project finance?

That the market is concentrating into fewer, larger cheques. US solar debt financings have fallen from 97 in 2024 to 80 in 2026 while the dollars have risen by almost half, so the average facility has grown by roughly three quarters in two years. Fewer sponsors reach the market, and those that do raise more each time. A company with 7 GW already operating is on the right side of that split.

The second signal is about agents. Royal Bank of Canada holds both the administrative and collateral agent roles here, and appears as a lender in several of the year's largest US solar facilities, including Swift Current Energy's $750 million package in August and IPX Power's $800 million facility in the same month. In a market of larger, tighter clubs, the banks that can both lead and hold become the gatekeepers, and sponsors without those relationships will find the club harder to assemble than the project.

Frequently asked questions

Who is lending to the Sullivan solar project?Five banks: Royal Bank of Canada, which is also administrative agent and collateral agent, together with DNB Bank, Mizuho Bank, NatWest and Wells Fargo. Skadden acted as legal counsel to Deriva Energy and Milbank to the lenders.

How large is the Sullivan solar project and when will it operate?Enerdatics records it at 350 MW in Sullivan County, Indiana, developed through Sullivan County Power, LLC, connecting into MISO via the Gibson to Merom 345 kV line. It is in construction now, with commercial operation expected in March 2030.

How does the $540 million compare with other US solar loans this year?It sits mid-pack. Enerdatics records 24 US solar debt financings of $400 million or more in 2026, out of 80 worth $31.41 billion, with the largest about $4.95 billion.

Enerdatics tracks US solar project finance by sponsor, lender and agent role, including all 80 debt financings recorded in 2026. Explore Enerdatics for the full comparables set behind this analysis.