IDACORP Has Seven Gigawatts of Maybe and a Multiple That Already Priced It
Meta's servers are live and the load forecast jumped to 8.9% — but Micron's second fab still has no contract, and the AI-infrastructure tape has gone quiet
IDA · Earnings Call · 2026-10-29
The forecast finally showed up in the P&L
IDACORP has spent four straight calls promising an industrial inflection. This is the quarter it arrived: industrial revenues rose 27% year over year, the new large-contract revenue line added roughly $12 million of operating income in Q3 (~$20 million year to date), and management is now guiding to “our preliminary load growth CAGR for the 2027 IRP is 8.9% for retail sales for the next 5 years” — Unidentified Speaker · 2026-10-29. Translating that into physical terms: about 19,000 gigawatt-hours of retail load in 2027 growing to roughly 29,000 by 2032. The single most important sentence on the call was not a forecast, though — it was an operating fact. “Meta officially began operations at its data center south of Boise in September and is now ramping up” — Unidentified Speaker · 2026-10-29. For two years, data center load at Idaho Power was a slide, an NDA and a set of inquiry letters; this quarter it became a meter spinning. That is the difference between a story and a business — and it is why Meta 1 jumped into the company's keyword hierarchy almost immediately, a phrase that also surfaced among global themes this quarter, a rare case of a single hyperscaler facility being a market-level token.Seven gigawatts, of which roughly two are real
The pipeline discussion was the most substantive in a year. Management had been anchoring on a "4 gigawatt" figure for several quarters, frozen since at least 2025's ~3.8 GW. This quarter it broke open:The segmentation is unusually honest for a utility: roughly 1 GW with financial commitment, 1 GW paid-for study work, and ~5 GW of tire-kicking. The committed slice is what is inside the 8.9%. Everything else is optionality — which is precisely how the 2032 RFP was constructed. Over 100 bids, an 11-project shortlist deliberately larger than the committed need, and, for the first time, a company-owned project ranked first in both years. The 2031 winner is a 430 MW advanced simple cycle plant; the 2032 candidate is a 432 MW combined cycle unit — a genuinely new keyword for this company, and a technical admission that single-cycle peakers are no longer the efficient answer. That shift matters because for a year management's resource narrative was batteries-plus-gas-peakers, with SMRs politely deferred — “Our summary is that we like SMR technology. At this time, we do not love the pricing.” — Adam Richins, Vice President of Energy Resources · 2026-07-30 Now the honest answer on 2032 is a large combined cycle, delivered with a notable lack of conviction: “It feels like it's going in that direction, but we still need to kind of look at the load forecast” — Unidentified Speaker · 2026-10-29.Of that 4 gigawatts, I'd say 1 gigawatt is what we often refer to as ... promising. ... And then around 2 gigawatts of the 4 feel preliminary ... But we have another 3 gigawatts that have come in the door since we talked about the 4-gigawatt number. So really, 5 gigawatts are preliminary, and again, another 1 gigawatt is promising, and another 1 gigawatt is being studied.
The contract that keeps not getting signed
Here is the tension worth holding onto. In July, Micron Fab 2 was described as an active negotiation with ground preparation underway: “We are in ESA discussions, but as Lisa mentioned, those are confidential, so we cannot really speak to those.” — Adam Richins, Vice President of Energy Resources · 2026-07-30 In April, the line was “Fab 2 is not in the 8.3%” — Adam J. Richins, Unknown · 2026-04-30. Three calls later, an analyst asked directly whether the "before year-end" target still held, and the answer softened to “we feel like it's going well, and certainly we're hopeful that we'll meet those” — Unidentified Speaker · 2026-10-29. Fab 2 remains outside the 8.9% load forecast, outside the CapEx plan, and outside the generation stack. The largest single incremental load in the territory is still un-contracted going into the guidance-setting quarter. The rate case followed a similar pattern of deferral. Idaho Power is skipping a 2026 general rate case entirely on the strength of large-load revenue — a real validation of the growth-pays-for-growth construct, and a theme that has quietly dropped out of the company's keyword set after topping it earlier this year. But 2027 is when B2H, hydro refurbishments and system hardening convert from construction work in progress to plant in service, which stops AFUDC and starts depreciation. Buckham: that “increases the likelihood of us filing a general rate case next year” — Unidentified Speaker · 2026-10-29. The tell is the tax credit. Additional amortization fell to $6.3 million for the nine months versus $39 million last year, guidance was cut to under $10 million from under $15 million, and the EPS range rose to $6.40–$6.50. Less reliance on ADITCs means earned ROE is converging on allowed ROE — higher-quality earnings, but a thinner shock absorber exactly when the plant-in-service wave hits.The tape has stopped paying for the story
The market backdrop is what makes this quarter interesting rather than merely good. The global AI data center complex is up huge over twelve months — dozens of advancers across semis, power and cooling — yet it is down more than 5% over both the 90- and 30-day windows, with a wide majority of constituents negative. The trade that made data-center-adjacent utilities expensive has stalled. IDACORP is a clean read on that. The stock is 7.1% below its July 2 peak of $154.55, down 2.5% over 90 days and essentially flat, while the fundamental narrative accelerated. At 24.7x net income, up 22% year over year, there is not much room for a slip in a contract date. The build is real and expensive: capex of $437 million in the June quarter, up 31% year over year, against $3.7 billion of long-term debt, pushing interest coverage to 2.6x versus 6.0x in 2017. Roughly 93 cents of every revenue dollar went back out the door as capital spending in that quarter. Management has prefunded aggressively — $873 million of unsettled forwards, equity needs covered through 2027 and likely into part of 2028 — and is willing to let the payout ratio drift:We don't have a concern if for some period of time we fall below that 50 to 60% dividend target range. ... What that gives us though is very significant dividend horsepower for the future.