Solar Sizing for Kuna Data Centers

Capacity Needs: MW-to-Load Ratios and Acreage Estimates

Powering a continuous 1,050 MW data center load with solar alone would require on the order of four to five times more solar nameplate capacity, because Idaho’s solar plants only produce ~20–26% of their rated output on average (capacity factor, CF)nwcouncil.org 1eta-publications.lbl.gov 2. In other words, each 1 MW of steady load demands ~5 MW of PV capacity given Idaho’s ~20% CF resource. We quantify three scenarios (low/mid/high) reflecting fixed-tilt vs. tracking systems and best/worst insolation assumptions:

• Low Output Scenario (Fixed-Tilt, CF ~18%) – ~5,800 MW of solar required to net 1,050 MW on average. At ~7.5–8 acres per MW for utility PVdocs.nrel.gov 3, this equates to roughly 44–46,000 acres of solar farms. Such a buildout (≈180 km²) is enormous – about 70 square miles of panels. The upfront capital expenditure (CAPEX) at today’s high-end prices (~$2/W_ac) would be on the order of $11–12 billion for full 1,050 MW offset (all-in, before tax credits).
• Mid Output Scenario (Single-Axis Tracking, CF ~22%) – ~4,800 MW of PV needed to offset the load, occupying roughly 38–40,000 acres (60–63 sq miles). Tracking boosts energy yield ~5 percentage pointseta-publications.lbl.gov 4 but also slightly increases land per MW. At median utility-scale costs (~$1.3–1.5/W), CAPEX is around $7 billionutilitydive.com 5.
• High Output Scenario (Optimized Tracking, CF ~26%) – ~4,000 MW of PV could suffice under best-case conditions (high DC/AC ratio, bifacial modules, peak insolation)eta-publications.lbl.gov 6pv-magazine-usa.com 7. This still requires on the order of 32–33,000 acres (~50 sq miles) of solar arraysdocs.nrel.gov 8. Assuming aggressive cost reductions (~$1/W_ac), the CAPEX might be $4–5 billionutilitydive.com 9. (This optimistic scenario relies on top-tier technology and low pricing that may not fully materialize.)

Table: Solar Capacity, Land, and CAPEX to Offset 1,050 MW Load (100% annual energy matching)

All scenarios underscore the huge footprint of solar required. For context, Idaho’s largest solar farm today is 120 MW on ~1,000 acresidahopower.com 10 – the data centers would need dozens of such plants. Per-MW Load breakdown: To offset Meta’s 250 MW facility alone at mid CF (~25%), roughly 1,000–1,250 MW of solar (≈8,000–10,000 acres) is needed, costing on the order of $1.5–3 billion. The larger Gemstone 800 MW park would require about 3,200–4,000 MW of PV (≈25–32k acres) at a ~$5–9 billion investment. Combined, the 1050 MW load corresponds to roughly 4.5 GW PV spread over ~35–40k acres, costing $8–12 billion (mid-range scenario) for the solar infrastructure alone. These estimates include full EPC scope – panels, inverters, racking, new substations and interties, land acquisition, construction, plus typical developer fees and contingencies.

Full-System Costs and Component Breakdown

Utility-scale solar costs in 2025 are on the order of $1–1.5 million per MW_ac for a complete projecteta-publications.lbl.gov 11. However, we include all balance-of-system (BOS) and integration expenses in our analysis. Key cost components, roughly ordered by share, are:

• PV Modules: Solar panels typically make up ~20–30% of project upfront cost. Module prices have fallen dramatically (utility module MSP ~$0.30/W)utilitydive.com 12, aided by Chinese manufacturing scaleiea.org 13. For our 1,050 MW load case, modules alone represent an outlay in the low billions (e.g. ~$1.2 billion if 5 GW DC at $0.24/W_dc). We assume a module reserve fund as well – by year ~15–20 some panel replacements are budgeted (module output degrades ~0.5–0.7%/yr)atb.nrel.gov 14.
• Inverters and Electrical Equipment: Inverters (DC-to-AC converters) and power transformers account for ~5–10% of CAPEX. Recent trends show inverter costs rising due to supply tightnessutilitydive.com 15. For ~5 GW of PV, central inverters and medium-voltage transformers could total on the order of $300–500 million. These are typically replaced or overhauled around year 10–15, so our cost includes an inverter refresh reserve.
• Structural BOS (Racking, Trackers, Site Work): The mounting structures, single-axis tracker systems, steel posts, wiring, and other BOS hardware plus installation labor are a major cost chunk (often ~40% of utility PV cost)atb.nrel.gov 16. The expansive scale here means literally millions of modules mounted on thousands of tracker rows. We assume robust site prep and grading costs as well (especially for 40k acres). The BOS + labor for our full build runs into several billion dollars (very roughly $0.5–0.7/W). Notably, large projects can see economies of scale (pre-assembly, automated pile driving, etc.), but also face labor cost escalation – NREL found 2022–23 BOS labor and inverter costs jumped, pushing utility PV from $1.07/W to $1.16/Wutilitydive.com 17.
• Transmission Tie-in and Substations: To integrate ~4–5 GW of new solar, multiple high-voltage interconnection points are needed. We include new substations and gen-tie lines linking the solar farms to Idaho Power’s 138 kV/230 kV network. A rough allowance of ~$150k–200k per MW for interconnection facilities is included (several hundred million in total). Indeed, network upgrade costs have risen as interconnection queues grow. For example, a single 320 MW solar project for Meta requires a new 230 kV intertie (the Pleasant Valley and Blacks Creek farms tie into the system near Boise)pv-magazine-usa.com 18puc.idaho.gov 19. Our cost stack assumes the solar developers or Meta fund these grid upgrades (consistent with special contract terms).
• Soft Costs, Land, and O&M: Permitting, engineering, project management, developer fee, and land acquisition add ~10–15%. In Ada and Elmore County, agricultural land values can range from $3k–10k/acre; using ~$5k/acre, securing ~40k acres might cost ~$200 million (though many projects lease land at ~$200–300/acre/year)renewableenergyworld.com 20boisedev.com 21. We assume either outright purchase or 25-year lease payments equivalent to that. We also factor 25 years of O&M (~$15 kW/yr falling to $10 kW/yreta-publications.lbl.gov 22) – covering panel cleaning, inverter maintenance, property taxes, insurance, and site security. Over 25 years, O&M cumulates to ~$250k per MW (present value), which for ~5 GW is ~$1.25 billion. While O&M is not an upfront CAPEX, a funded reserve or PPA pricing would account for it.

Bottom line: a full 100% solar supply for 1,050 MW continuous load is a double-digit billion-dollar endeavor, and on the order of 40,000 acres of land development. (Even a 50% solar offset would still cost many billions and cover ~20k acres.) These figures assume ideal conditions; any delays, tariff changes, or higher interest rates could push costs higher. Notably, our estimates do not include energy storage – incorporating batteries to shift solar into nighttime would add further billions (see Chapter 9). In reality, meeting 24/7 demand would likely require a portfolio of resources; relying solely on solar oversupply would be highly inefficient (massive midday surplus and curtailment).

Solar Resource and Seasonal Performance in Kuna

Southwestern Idaho has a good but not superb solar resource. Boise’s average solar insolation is about 5.3 kWh/m²/day, roughly 36% higher than cloudy locales like western Washington, but ~19% lower than the desert Southwest (e.g. Nevada)solarenergylocal.com 23. This puts Kuna’s annual PV capacity factor in the ~20–22% range for fixed-tilt systemsnwcouncil.org 24, and up to ~26% with single-axis trackingeta-publications.lbl.gov 25. However, seasonal variation is pronounced: in mid-summer, Boise sees ~7.0–7.4 kWh/m²/day of sun, whereas in December it averages only ~2.5 kWh/m²/day. Solar output in winter can be a third (or less) of summer levels – for example, Idaho’s solar capacity factor drops to ~10% or lower in December but exceeds 25% in July. Cloud cover and snow accumulation can further reduce winter production.

The implication is that solar overbuilding would be needed to cover the same load in winter. Our sizing above was based on annual energy; in practice, a 100% solar strategy would either massively overgenerate in summer (with huge curtailment) or fall short in winter without storage. For instance, meeting a 250 MW load in December (2.5 sun hours) would require on the order of 2,100 MW of PV just for midday, whereas the same 250 MW in July (7 sun hours) needs ~750 MWsolarenergylocal.com 26. The mismatch between constant data center load and peaky solar output is exactly why Idaho Power is pairing batteries with the new solar farms[25][26]. On a yearly basis, each 1 MW_ac of PV in Ada County produces on the order of 1,750–2,300 MWh per year (fixed vs. tracking)nwcouncil.org 27eta-publications.lbl.gov 28. Thus, the 1,050 MW of data centers (using ~9.2 million MWh/year if operated at full load) would require ~9–10 million MWh of PV generation annually, consistent with the ~4–5 GW of capacity calculated.

Cloud variability and outages: Idaho’s high-desert climate offers many sunny days, but also spring storms and occasional wildfire smoke that can dim output. Hourly generation profiles would show sharp ramp-downs in the evening (even with tracking, sunset in June around 9 pm still leaves an all-night gap). Net MWh output per MW will also degrade ~0.5% per year as panels ageatb.nrel.gov 29, so by year 25 a given solar farm might produce ~12% less than when new. Our sizing does not even account for that long-term degradation or downtime for maintenance – in practice additional capacity or repowering is needed to maintain 100% renewable supply over decades. All these factors underscore that solar can substantially offset the AI centers’ energy needs, but cannot supply round-the-clock power on its own. There will be hours, especially winter evenings, when even 5× overbuilt PV has zero output – necessitating either massive storage, backup generation, or drawing from the grid mix (likely gas/hydro). The next chapter will explore how storage and dispatchable resources must fill this gap.

(On the plus side, Idaho’s solar output is well-correlated with peak summer loads (air conditioning and irrigation)idahopower.com 30. So added solar does help the grid on hot afternoons. But the data centers’ flat 24/7 demand provides less “peak-shaving” benefit – they create a new baseline load that persists into night.)

Developers and Delivery: Idaho’s Solar Build-Out and Players

Idaho Power’s latest plans call for an unprecedented surge of utility-scale solar to meet both normal growth and large customer procurements. In the 2021–2025 period, the utility went from virtually no PV to contracting well over 700 MW of solar (most paired with batteries)renewableenergyworld.com 31idahopower.com 32. The 2025 Integrated Resource Plan projects ~1.4 GW of new solar by 2026 and more beyond. Who is building these farms? Largely out-of-state solar developers, often partnering with Idaho Power via power purchase agreements (PPAs). We compiled a “developer tracker” of key IRP-listed solar projects:

Key observations: Nearly all utility-scale solar in Idaho to date has been led by national developers – Duke (a North Carolina utility) built the first two farms; rPlus (Utah, backed by AES) is building Meta’s projects; Savion (Shell-owned, Missouri) is pursuing the huge Blacks Creek farm. Idaho-based solar firms are rare – the notable one is Clēnera, a Boise-born developer (responsible for 2016’s Grand View Solar and now Crimson Orchard). Clēnera’s local presence is an exception; most contracts went to experienced external firms. This raises the question of in-state economic benefit – construction jobs are local, but corporate profits and specialized expertise often flow out. On the flip side, out-of-state developers bring needed capital and know-how to get projects financed.

We also see repeat vendors and issues: Duke was tapped twice (perhaps due to successful execution of Jackpot, though it experienced some curtailments early on). Savion’s double attempt – Blacks Creek moving forward but its earlier 400 MW Ada County proposal (Powers Butte) was killed by NIMBY resistance. This suggests permitting risk in more populated areas; even though land was secured, local governments can block projects over wildlife or aesthetics (Powers Butte’s 2,385 acre footprint drew fire as a “blight”). Idaho Power’s IRP specifically notes relying on those big projects, so delays or cancellations put pressure on the utility’s capacity plans. Technical hiccups have occurred too: the Black Mesa solar+storage project (developed by a newcomer LLC pivoting from oil & gas) reportedly had equipment problems in its early months – industry chatter points to a faulty inverter capacitor bank that had to be replaced (affecting the 40 MW battery operation). While not publicly detailed in the IRP, this illustrates integration challenges when deploying first-of-a-kind resources (Black Mesa was Idaho’s first large battery site)docs.idahopower.com 41. Idaho Power had to work with the developer to get it sorted out, as Micron was counting on that output.

Going forward, the 2025 IRP’s solar additions will likely be awarded to those same players or similar large developers through RFPs. No Idaho-headquartered EPC contractor has suddenly appeared; instead, firms like Sundt (national EPC for rPlus) handle constructionpv-magazine-usa.com 42. The upside is that multiple developers are competing, which can lower PPA prices. The downside: Idaho is depending on timely delivery by external actors – any financial troubles or supply chain issues they face could ripple into project delays. For example, if module tariffs or import bans hit a developer’s panel supply, a project could slip past its needed in-service date (jeopardizing data center renewable goals and grid resource margins). The IRP highlights 2025–2027 as critical for resource additions[48][49]; thus far, some projects (Pleasant Valley 1) have stayed on track, but continued vigilance is needed to ensure the remaining planned farms come online as promised.

Solar Supply Chain Realities and Risks

Building gigawatts of solar in Idaho implicates a global supply chain – one that today is heavily concentrated overseas. We summarize the major components and their origin/risk profile:

In summary, Idaho’s solar strategy is deeply entwined with global supply chains. The heavy dependence on Asian-made solar panels is a strategic vulnerability – if trade policies shift or if there’s an international crisis, projects could be delayed or face cost overruns. On the other hand, this dependence has brought cost savings: Chinese competition drove panel prices down >80% over the last decadeiea.org 53, directly benefiting projects like Meta’s. Inverters and batteries present a similar dichotomy – Chinese technological dominance vs. U.S. desire for secure, domestic alternatives. Federal incentives in the Inflation Reduction Act (IRA) are trying to localize more of this supply (e.g. 45X credit for U.S.-made componentsenergy.gov 54), but in 2025 those nascent factories are not yet sufficient for multi-gigawatt demand. Economic security implications: Relying on imported energy hardware can be seen as swapping one dependency (on out-of-state or foreign fuel) for another (on imported technology). If geopolitical tensions with China escalated (trade war, etc.), Idaho’s renewable buildout could stall – undermining both energy goals and local economic benefits (since projects might miss deadlines, and data center expansions might pause without their promised green power). On the flip side, successful deployment of solar farms will create some local jobs and tax base, and improves long-term energy autonomy (sunshine is local and free). The key is navigating the supply chain such that Idaho isn’t left in the lurch mid-project. Thus far, developers have managed by sourcing panels from tariff-exempt countries and using Tier-1 suppliers; continued diligence (and perhaps contingency plans for alternate suppliers) will be needed as the state scales up to gigawatt-level solar.

Energy Contracts: Tying Solar to Meta’s Data Center

Both Meta and Idaho Power have taken steps to ensure these new solar resources specifically benefit the Kuna data center project. Meta did not simply drop 250 MW of load on the grid and consume generic power – it negotiated a bespoke arrangement to supply that load with new renewables. In mid-2023 the Idaho PUC approved a Special Energy Services Agreement (ESA) for Meta’s facility (operated via Meta’s subsidiary “Brisbie LLC”)boisedev.com 55. Under this “Clean Energy Your Way – Construction” program, Idaho Power agreed to procure 100% renewable energy for Meta’s usage, and Meta/Brisbie in turn agrees to fund the associated resources (through special rates/PPA payments)[58][59]. In essence, Meta is bringing its own power plants to Idaho Power’s system – the utility integrates and delivers the energy, and Meta pays all costs, receiving credits for the green energy produced[60][59]. This arrangement was enabled by Schedule 33, a tariff rider for large customers to achieve renewable goals without harming other ratepayers[61][62].

Specific projects: For Meta’s Kuna data center, two dedicated solar farms (and one battery) have been lined up under this program:

• Pleasant Valley Solar – 200 MW (online 2025): This Ada County project by rPlus Energies is explicitly tied to Meta’s ESA. It achieved commercial operation in March 2025, just in time for Meta’s data center opening. Meta will receive the full output (and Renewable Energy Credits) from Pleasant Valley 1[63][64] via its agreement with Idaho Power. A second phase, Pleasant Valley 2 (125 MW in 2026), will further augment supplypv-magazine-usa.com 56.
• Blacks Creek Solar – 320 MW (due 2027): In late 2024, Idaho Power signed a 20-year PPA with Blacks Creek Energy LLC (Savion/Shell) for a massive 320 MW solar farm by end of 2027puc.idaho.gov 57. Notably, the PPA identifies Brisbie (Meta) as a third-party beneficiary of the energy and green attributes. In other words, this project is being built specifically to fulfill Meta’s future needs. Once online, Meta’s Idaho campus will be linked to a total of ~645 MW of new solar (200 + 125 + 320 MW)[67][68] – more than enough on paper to cover its ~250 MW load annually. (Excess credits in good solar years could apply toward other Meta operations or be retained for growth.)
• Kuna Battery Storage – 150 MW (2025): While not a source of energy, this large 4-hour battery (150 MW/600 MWh) is being built near Kuna to firm Meta’s solar supply. The PUC approved a 20-year storage agreement with “Kuna BESS LLC” in late 2023[69][70]. The battery will charge from resources like Pleasant Valley (when Meta’s load is lower than solar output) and discharge in evening hours to cover the data center’s demand[71]. Idaho Power stated this addresses “the mismatch between the constant load of a data center and the intermittent power from solar”[71]. Meta is effectively sponsoring this battery (paying for its capacity via the contract), ensuring its energy is available around the clock. The 150 MW BESS, slated for operation by summer 2025, will be Idaho’s largest battery and a critical reliability asset for the grid as well[72][73].

Through these agreements, Meta gains a 100% renewable-powered facility on an annual basis, and Idaho Power insulated other customers from subsidizing it. The utility confirmed the special contract is structured so that existing ratepayers won’t see a rate hike due to Meta – Meta covers all associated costs of its serviceboisedev.com 58[61]. Indeed, the PUC’s concern in approving the ESA was to ensure “no harm to other customers,” which it found to be satisfied[75][60]. This model – essentially a green tariff – has also attracted Micron Technology, which did a similar 40 MW solar + 40 MW battery deal (Black Mesa) for its Boise campusdocs.idahopower.com 59, and likely will use more under CEYW as its new fab comes online. Other large tech companies in Idaho could follow suit.

What about the Gemstone 800 MW data center park? That project (led by Diode Ventures) is more speculative and multi-tenant – there isn’t a single end-user like Meta yet publicly committing to renewables. However, if companies like Google or AWS were to locate there, they would almost certainly negotiate similar renewable deals. Idaho Power’s IRP does include generic future solar for expected large loads, but as of now no specific PPA for Gemstone’s 800 MW has been announced. The site only recently got zoning approval in April 2025datacenterdynamics.com 60. It’s conceivable that Gemstone’s eventual tenants will utilize the Clean Energy Your Way program too – meaning Idaho Power would acquire additional dedicated solar (or other renewables) on their behalf. The utility has a responsibility to meet its 100% clean energy by 2045 goal as wellboisedev.com 61, so even absent customer-driven PPAs, the IRP shows thousands of MW of solar and wind being added for the system. In summary, Meta’s data center already has concrete renewable projects tied to it, enabled by innovative contracts, whereas Gemstone’s future load remains a question mark but will likely demand similar clean energy solutions once tenants materialize.

Conclusion: Solar’s Promise and Limits

Solar energy can supply the massive AI loads – but not without significant caveats. To offset 1,050 MW of always-on demand, we quantified requiring up to ~5 GW of PV, tens of thousands of acres of land, and over $10 billion in capital. Such a solar buildout in Idaho is technically achievable over time, given the state’s decent sun and available flat land, but it confronts practical limits: acquiring and permitting 40,000 acres, navigating global supply dependencies, and coping with seasonal/intermittent output. The analysis above highlights that land and integration requirements scale steeply – even at the high end, covering ~50 sq miles with panels yields just ~1 GW average output (enough for the data centers’ needs, but nothing at night or in mid-winter)solarenergylocal.com 62. Furthermore, the solar supply chain risks and project execution challenges introduce uncertainty: a delay in one 320 MW project or a hiccup in battery delivery could leave the data center partially relying on grid power or carbon offsets.

Crucially, solar alone cannot provide 24/7 reliability for critical AI workloads. To truly power these data centers around the clock with clean energy, energy storage and other firming resources must come into play. Idaho Power recognized this by fast-tracking 270 MW of battery storage (150 MW for Meta, 120 MW utility-owned) as a direct response to the data center loads[72][79]. In the next chapter, we will delve into how battery storage, dispatchable generation, and load management are required to bridge the gap when the sun isn’t shining. We’ll examine how the utility plans to prioritize and protect essential AI workloads during grid stress, and the trade-offs between overbuilding solar versus investing in firm capacity. The sheer scale of solar needed introduces new vulnerabilities – from land use conflicts to evening ramp shortfalls – which mean Idaho cannot simply solar-build its way out of reliability challenges. Thus, Chapter 9 will explore how power prioritization and storage become the next puzzle pieces: ensuring that the “green electrons” from Idaho’s vast new solar farms can be stored, shifted, or backed up such that critical AI computations stay online, and less essential loads can be curtailed if needed. The success of Idaho’s AI infrastructure will hinge on managing these constraints – balancing an abundance of daytime solar with the demands of a 24/7 digital economy.