TransAlta Grew the Prize to 1.6 GW — and Pushed the Signature Date Into 2027
Battle River goes into mothballs, Boreal gets bigger on paper, and the definitive agreement that was supposed to land 'in year' has now drifted a third time.
TAC · Earnings Call · 2026-11-03
The quarter was fine. The calendar was not.
TransAlta's third quarter was operationally unremarkable in the way good utilities usually are: adjusted EBITDA of $264 million, free cash flow of $93 million or $0.29 a share, 92.6% fleet availability, and a reaffirmed 2026 guidance range. The problem — and the reason this call matters — is that the Alberta merchant backdrop continues to soften underneath it. “Spot prices averaged approximately $36 in the third quarter, down from $51 in the third quarter of 2025.” — Mike Politeski, Unknown - likely senior executive involved in financial operations (e.g., CFO or similar) · 2026-11-03 Management papered over that with the hedge book (about 2.1 TWh at $67 for the balance of 2026, 6.7 TWh at $65 for 2027) and with dispatch optimization that let the gas fleet realize $55/MWh — a 53% premium to spot — and hydro $43/MWh, a 19% premium.
So the financials are held together by trading and contracts, not by the market. Which puts all of the narrative weight on the two things management can control: portfolio surgery and the data centre file. Both moved this quarter — in opposite directions.
The timeline that keeps sliding
Start with the thing that did not happen. In February, the message was that “we do expect definitive agreements to be completed in year.” — John Kousinioris, President and Chief Executive Officer · 2026-02-27 By the July call, it had become “we continue to remain very confident in our ability to bring forward our data center option here later in the year.” — Joel E. Hunter, President and Chief Executive Officer · 2026-07-31 On Tuesday, Joel Hunter finally said the quiet part out loud:
“I would not say it is probably 2026. it is probably into sometime early into 2027 at this point in time once we understand exactly the amount that we can offer our customers.” — Joel E. Hunter, President and Chief Executive Officer · 2026-11-03
That is three consecutive calls in which a definitive agreement moved right. And the reason given has not changed either: everything is gated on the Alberta Electric System Operator deciding how much of TransAlta's parked thermal fleet counts as usable underutilized generation. That was the same ask in November 2025, in February, and in July. The regulator's answer is the binding constraint on this stock's growth story, and it is the one variable management cannot schedule.
The offset is real, though: the prize got bigger while the paperwork got later.
when we announced the MOU back earlier, you know, I guess, February, we talked up to 1 GW. Well, we now know that could be a higher number and you saw that from the open house here.
The Boreal Technology Hub — the Keephills-adjacent project with Brookfield and CPP Investments — is now framed at up to 1.6 GW of electrical load versus the 1 GW contemplated in the original MOU, with the 230 MW Phase 1 allocation as the anchor and room for up to six facilities on site. The open house drew just over 180 people and produced no protests, and the AUC approval for the Keephills project landed in October. Management is now explicitly calling the 1.6 GW "phase 1" of something larger. That is a company that believes it is negotiating from strength while its counterparties wait.
Parking steel to sell optionality
The genuinely new item this quarter was the mothballing of Battle River units 4 and 5, plus an extension of the Sundance 6 mothball outage. Hunter's framing was blunt: “This is purely an economic decision and really to preserve the long term optionality for both of these assets.” — Joel E. Hunter, President and Chief Executive Officer · 2026-11-03 Forty-five employees stay on site to keep the units warm.
That is an economic decision that reads as a statement about the forward market. These are gas-fired steam units designed to run above 90% capacity factor; they have been running near 20%. When capacity that efficient is cheaper to park than to dispatch, the short-term Alberta market is telling you something the hedge book is hiding.
There is also a quiet irony worth flagging. Back in February, management pointed to the already-mothballed Sundance 6 and Sheerness 1 as the pathway to the full 1 GW: “with both Sundance 6 and Sheerness I being mothballed, just those two units alone provide a pretty clear path where we could certainly be able to ramp up and meet the up to 1 gigawatt.” — John Kousinioris, President and Chief Executive Officer · 2026-02-27 Nine months later, the fleet of parked units has grown, not shrunk, and the load that was supposed to justify restarting them has not been contracted. "Optionality" is the right word — but optionality that keeps accumulating is a cost, not an asset, until someone signs.
The tape has already voted on data-centre power
Here is the contrast that matters. Over the last thirty days, the market's own data-centre-power basket has been sold hard: AI data center names are down roughly 21% with 55 decliners, data centre capacity names down about 30%, data centre operators down more than 32%, and the power generation business cluster down over 23%. Simultaneously, nearly every regulated utility that reported in the past week — Talen, PGE, Xcel, AES, IDACORP, Pinnacle West, Portland General, WEC, Entergy, Fortis — led with Large Load interconnection, data centre tariffs, and rate base growth. The theme is now sector boilerplate; the trade is not.
That is precisely where TransAlta's differentiated angle lives. Hunter's pitch is jurisdictional: “We are the only jurisdiction that really, I. I would say, is long power. Within North America.” — Joel E. Hunter, President and Chief Executive Officer · 2026-11-03 He name-checked moratorium regimes in Texas, Pennsylvania and New York as the foil. Speed-to-power via existing steel is a genuinely company-unique asset — and it is the exact asset whose monetization is stuck at the system operator.
What the market is actually paying for
The quarter's other actions were unambiguously good housekeeping. The Colorado peakers closed — 318 MW, US$1 billion, roughly $110 million of contracted annual EBITDA across 26–27 year terms — and 49% of the 502 MW Oklahoma wind portfolio went to Ares for US$305 million. Both are the asset recycling programme in action, and Mike Politeski was explicit that this is early: “We have over 90 assets in the portfolio. Many of them are out of our outside of our core areas.” — Mike Politeski, Unknown - likely senior executive involved in financial operations (e.g., CFO or similar) · 2026-11-03 A small asset went out the door for $27 million "under the radar" during the quarter.
The Colorado deal also, quietly, buys a second US platform: Colorado still generates 20–25% of its power from coal, and TransAlta now has a contracted beachhead in WECC plus a relationship with the seller's sponsor. That is a more durable story than a 1.6 GW press release.
The tell
The word that has climbed to the top of this company's keyword board is not "data center" anymore — it is data center load, capacity factor, and utilization mechanics. The conversation has shifted from ambition to plumbing. That is what happens when the commercial case is settled internally but the regulatory case is not.
TransAlta is a $4.2 billion company whose equity story now hinges on a single regulatory determination, wrapped in a bigger number than last quarter and a later date than last quarter. Both can be true. But investors should note which one management chose to lead with.