Energy
PrivateThe cheapest listed way to own Alberta's AI-power build-out at the bottom of a violent merchant cycle — but Brookfield's in-the-money option on the crown-jewel hydro fleet quietly taxes the exact upside the data-centre thesis depends on. Constructive, not yet de-risked.
Research
The verdict
The cheapest listed way to own Alberta's AI-power build-out at the bottom of a violent merchant cycle — but Brookfield's in-the-money option on the crown-jewel hydro fleet quietly taxes the exact upside the data-centre thesis depends on. Constructive, not yet de-risked.
TransAlta Corporation is one of Canada's largest independent power producers — a ~100-year-old Calgary-based generator that has spent the last decade pivoting off coal into gas, hydro, wind, solar and storage. It owns and operates a diversified fleet across Canada, the United States and Australia. Registry: slug transalta, ticker TA.TO, status public, primary topic energy.
How it actually makes money — three engines:
Customers / counterparties: hyperscalers (Meta on Horizon Hill; CPP Investments + Brookfield on the Keephills data-centre MOU), industrial off-takers, load-serving utilities, and the Alberta pool itself. customers.csv is empty, so all counterparty detail here is ``.
Contract structure: a barbell — a merchant Alberta core exposed to spot volatility, ballasted by a growing book of long-dated take-or-pay/tolling/PPA contracts (Centralia to 2044, Horizon Hill to Meta, the prospective 230 MW Keephills PPA). The strategic direction is to shift the mix toward contracted while keeping merchant upside for the coming Alberta recovery.
Map: fuel/inputs → generation → grid → end customer.
n/a for specific OEM contracts).Chokepoints / single-source dependencies:
Real, durable moats:
Weak spots (anti-moat): the merchant Alberta core has no pricing power — it is a price-taker in a commodity market with a supply glut. The moat is in the asset siting and dispatch, not the revenue line. Bargaining power over customers is low in merchant sales, but high in a bilateral data-centre deal where TransAlta controls the scarce interconnect.
Five reporting segments. FY2025 Adjusted EBITDA, all ``:
| Segment | FY2025 Adj. EBITDA | Q4-2025 Adj. EBITDA | Read |
|---|---|---|---|
| Hydro | C$285M | C$39M | Crown jewel; stable, high-margin, dispatchable |
| Wind & Solar | C$338M | C$102M | Largest segment; contracted + merchant renewables |
| Gas | C$438M | C$96M | Biggest EBITDA line; the transition workhorse |
| Energy Transition (ex-coal) | C$100M | C$16M | Shrinking (Centralia/Sheerness wind-down) |
| Energy Marketing | C$85M | C$21M | Trading desk; volatility-monetising |
| Corporate | C$(142)M | C$(27)M | Overhead |
| Total Adjusted EBITDA | C$1,104M | C$247M |
(Note: an earlier automated web summary garbled the FY Wind & Solar line as "C$43M" — impossible given Q4 alone was C$102M. The C$338M figure above is from the primary press release and is the one to trust. Flagging the discrepancy per provenance discipline.)
Trend & cause: Total revenue C$2,405M and adj. EBITDA C$1,104M in FY2025. The direction is decelerating on the merchant lines: Q1 2026 adj. EBITDA fell to C$204M from C$270M YoY, driven by (a) no generation at Centralia Unit 2 during its conversion and (b) lower Alberta power prices. Gas is the largest EBITDA contributor; Wind & Solar second (boosted by consolidating TransAlta Renewables in 2023, Lens 9); Hydro the highest-quality. Energy Transition is deliberately shrinking as coal exits. Geographic split is Alberta-dominant (58% of capacity) with growing US contracted renewables and a small Australian book — segments.csv is empty so a precise geographic EBITDA split is n/a.
The tape, all ``:
Balance-sheet flags: Total consolidated net debt C$3.785B and Adjusted Net Debt / Adjusted EBITDA of 4.3x as of Q1 2026 — elevated, and rising as EBITDA compresses. Cash ~C$151M. This is the pressure point: leverage climbing into a merchant-price trough.
FY2025 context (the full-year print, reported Feb 2026):
Market reaction / what was priced in: the stock trades near multi-year lows (TAC as low as ~US$12.70, below its US$13.58 200-day MA; TA.TO ~C$17.27) — i.e. the market has priced the down-cycle and is skeptical of the recovery narrative. A reaffirmed guide + dividend hike was not enough to re-rate it.
No transcripts on disk; this lens is `` from press releases and coverage, so treat as directional, not verbatim.
What management is focused on (consistent across the last several quarters): (1) the Alberta recovery thesis — repeated framing that the market will "recover meaningfully through the end of the decade as demand growth increases, including with data-centre load"; (2) contracting the fleet (Centralia toll to 2044, Horizon Hill/Meta, Keephills PPA); (3) capital discipline + dividend growth (7th consecutive annual increase); (4) fleet transition (coal exit via Centralia conversion and Sheerness mothball).
Tone shift: the language has migrated from defensive transition (managing coal exit, 2023-24) to offensive optionality (data-centre / AI-load, 2025-26). The data-centre MOU is the new centre of gravity of the story. What they have stopped emphasising: near-term Alberta spot-price upside (they've conceded the trough and pushed the recovery out to "end of decade"). That honesty is a modest positive — they are not pretending the glut isn't real. The risk is that "end of decade" is doing a lot of work in a story sold to investors today.
Peer set: Alberta/North American IPPs and contracted-renewable generators. Market caps are well-sourced; EV/EBITDA multiples from aggregators are inconsistent and flagged.
| Company | Ticker | Mkt cap | EV/EBITDA | Note |
|---|---|---|---|---|
| TransAlta | TA.TO / TAC | C$5.12B | ~9.4x (FY25) / ~10.3x (FY26E) | EV ~C$10.34B ÷ EBITDA C$1.104B / C$1.0B guide-mid |
| Capital Power | CPX.TO | C$11.08B | "16.28x" — suspect | Alberta merchant peer; historically ~8–10x, so 16x looks distorted → treat as n/a — not cleanly sourced |
| Boralex | BLX.TO | C$3.81B | "16.37x" | Contracted renewables (genuinely higher); being acquired by Brookfield Renewable |
| AES Corp | AES | US$10.26B | ~14.9x | Going private at US$15.00/sh, approved 2026-06-26 |
| Brookfield Renewable | BEP | ~US$7B (LP units) | reported 4.9x — unreliable | Multiple clearly mis-tagged; n/a — not cleanly sourced |
Read: on the numbers I can source cleanly (TransAlta's own EV/EBITDA of ~9–10x and an FCF yield of ~8–10%), TransAlta is the cheapest name in the group. The discount is earned — merchant Alberta beta, a GAAP loss, coal-transition drag, and the Brookfield hydro overhang — but the direction matters: two peers (AES, Boralex) are being taken out by private capital at mid-teens multiples in the same window TransAlta trades at ~10x. Private capital is paying up for exactly this kind of dispatchable/contracted fleet. That is the single most important comp signal, and it is a valuation floor argument, not a fabricated multiple.
TransAlta dividend yield ~1.6% (C$0.28 / ~C$17.27) — low yield, but that is the point: unlike a yieldco, TransAlta retains FCF (payout only ~18–24% of guided FCF) to fund the C$3.5B growth plan and delever. It is a total-return/optionality name, not an income name.
What has moved TransAlta >5%, and what it reveals ``:
Pattern: the market reacts to (1) the Alberta spot power price above all else, and (2) discrete structural events (Brookfield deals, RNW take-in, data-centre news). It is not an EPS-reaction stock — GAAP earnings are noise (impairments). The next real catalyst is a definitive Keephills PPA or an AESO/REM clarity event, plus any inflection in the Alberta forward curve.
Leadership is mid-transition (April/May 2026):
Track record: the prior team executed a credible, disciplined transition (coal → gas/renewables) and consistently grew the dividend (7 straight years). Capital allocation has been reasonable: the C$1.38B RNW privatisation at an 18.3% premium simplified the structure and pulled renewable cash in-house; growth capex is disciplined (C$3.5B / 1.75 GW to 2028, funded from retained FCF). But two allocation questions dog them: (1) the RNW take-in used cash + stock near a cyclical high — arguably rich; (2) the 2019 Brookfield hydro financing (7% cost + a deep-in-the-money conversion option, Lens 10) looks, with hindsight, like an expensive way to raise C$750M that mortgaged the crown jewel.
Skin in the game — a red flag. Insiders own <1% (~C$16M); institutions own 67%. Low insider ownership means management is not meaningfully co-invested with shareholders — a governance weakness, especially with a Brookfield option and a new, internally-promoted CEO. Archetype: professional managers, not owner-operators. insider-transactions.csv absent → precise recent-buying data n/a.
Income statement / earnings quality: the FY2025 C$(190)M net loss is almost entirely non-cash — impairments (C$37M wind, C$27M energy-transition, C$37M required divestitures, net of C$17M reversals) plus a C$63M decommissioning-provision revision from lower discount rates. Against that, FCF was +C$514M and operating cash flow +C$646M. Cash flow exceeds earnings by a wide margin — the benign direction of divergence (the opposite of the classic fraud pattern where earnings outrun cash). Adjusted pre-tax earnings were +C$181M. Verdict: the loss is accounting, not cash — a positive tell, though recurring impairments on the transition fleet are a real economic signal that coal-era assets keep losing value.
Balance sheet: the genuine risk is leverage (4.3x Adj. Net Debt/EBITDA, C$3.785B net debt) rising as EBITDA compresses. Still investment-grade (BBB-low, DBRS, stable), but a further Alberta leg down would pressure the rating and the dividend. Decommissioning/reclamation liabilities on retired coal are a long-tail obligation.
Related-party / structural — the Brookfield hydro option (the headline red flag). In 2019 Brookfield injected C$750M (C$350M exchangeable debentures 2019 + C$400M redeemable prefs 2020, 7.0% coupon) convertible after Dec 31, 2024 into an equity interest in TransAlta's Alberta Hydro Assets. The top-up option lets Brookfield take an additional 10% of the hydro entity if the 20-day VWAP ≥ C$14, and up to the full 49% if VWAP ≥ C$17. With TA.TO at ~C$17.27, that option is in-the-money. So the very asset that makes the AI-power/REM thesis work (dispatchable hydro) can be diluted up to 49% to a related party at a fixed 7% cost — a structural value leak precisely when hydro becomes most valuable. This is the most underappreciated item in the whole file.
Regulatory findings:
n/a beyond the above.EPS is the wrong metric for a merchant generator with non-cash impairment noise — TransAlta itself guides on Adjusted EBITDA and FCF/share, so the projection is built there and EPS is indicative only. Base built bottom-up from the reaffirmed FY2026 guide. No forecast.ts logged (watchlist run). All ``, inputs labeled.
Anchor (FY2026 guidance, reaffirmed): Adj. EBITDA C$950M–C$1,050M; FCF C$350M–C$450M (C$1.18–C$1.51/sh). ~296.7M shares.
| Scenario | FY2026 | FY2027 | FY2028 | Logic |
|---|---|---|---|---|
| Bear | EBITDA ~C$950M; FCF ~C$1.15/sh | ~C$900M; ~C$1.00/sh | ~C$900M; ~C$1.05/sh | Alberta glut persists (~C$30-40/MWh); REM (2027) imposes a location discount on key assets; Sheerness/Centralia downtime drags; leverage stays ~4.3x, dividend growth pauses |
| Base | EBITDA ~C$1,000M; FCF ~C$1.35/sh | ~C$1,050M; ~C$1.45/sh | ~C$1,150M; ~C$1.60/sh | Pool price stabilises then grinds up as data-centre load begins arriving; Centralia toll (to 2044) + Far North (310 MW) + growth capex add contracted EBITDA; one Keephills PPA (230 MW) signed |
| Bull | EBITDA ~C$1,050M; FCF ~C$1.50/sh | ~C$1,200M; ~C$1.75/sh | ~C$1,400M+; ~C$2.00/sh | AI load re-tightens Alberta toward C$60-80/MWh; multiple Keephills phases (toward 1 GW) contracted; REM rewards dispatchable/hydro; re-rate toward peer multiples |
Indicative adjusted EPS: roughly breakeven-to-C$0.50 (base, 2026) rising toward C$0.75-1.25 by 2028 in the base/bull as EBITDA leverages up — but GAAP EPS may stay depressed if transition-asset impairments recur. Treat EPS as noise; FCF/share and the Alberta forward curve are the metrics that matter.
The single swing variable: the Alberta pool price trajectory 2026-2028. At C$32/MWh the merchant book barely earns; each ~C$10/MWh of sustained recovery is worth a large step-up in merchant EBITDA. The bet is whether AI data-centre load arrives fast enough (2027-2030) to absorb the supply glut before the balance sheet and the Brookfield option do damage.
Bull case. TransAlta is a deep-value cyclical at the trough with a free AI call option. You are buying Alberta's largest hydro fleet + a dispatchable gas/renewables portfolio at ~9-10x trough EBITDA and an ~8-10% FCF yield, while (a) private capital takes out peers (AES, Boralex) at mid-teens multiples, setting a valuation floor 30-60% higher; (b) the exact demand shock that fixes Alberta — hyperscale AI load — is arriving, with TransAlta uniquely positioned via shovel-ready zoned land + interconnect at Keephills and a CPP+Brookfield JV already at the table for up to 1 GW; (c) Horizon Hill proves it can already sign a hyperscaler (Meta); (d) the dividend is well-covered (18-24% of FCF) and growing (7 straight years) while it delevers; (e) REM's 2027 reliability/dispatchability incentives should reward its hydro + gas. Contrarian view the market refuses to see: this is not a dying coal utility, it is the cheapest listed way to own Alberta's data-centre power build-out — and the market is still pricing the last cycle's supply glut, not next cycle's demand shock.
Bear case (permanent-impairment risks). (1) The Brookfield hydro option — up to 49% of the crown jewel can be converted away at a fixed 7% cost, in-the-money above C$17, structurally leaking the very upside the thesis rests on. (2) Merchant timing risk — data-centre load is a 2027-2030 story requiring definitive agreements, regulatory approval and grid build-out; the balance sheet (4.3x) and cash flows must survive 1-2+ more years of C$30-40/MWh Alberta first, and dividend growth could stall. (3) REM (2027) is a coin-flip — locational marginal pricing could impose location discounts on TransAlta's assets; genuine design uncertainty. Pre-mortem (18 months out, thesis broke): the Keephills MOU never converts to a binding PPA (or CPP/Brookfield build their own generation), Alberta stays oversupplied as even more renewables come online, gas stays cheap, leverage forces a dividend cut, Brookfield exercises the hydro top-up — and the "AI-power optionality" turns out to have been a slide in an investor deck, not a contract. Is the multiple too high? No — ~10x is defensible; the risk is to the earnings, not the multiple.
Dismantling the bull case: What structurally breaks how this makes money? The whole equity is a levered bet on one commodity price in one province — the Alberta pool — and that market just added 11% supply in a year and printed its lowest February on record. Renewables keep getting built (by everyone, including the data-centre developers themselves), which structurally suppresses the pool price the merchant book depends on. The AI-load "fix" is a double-edged sword: hyperscalers increasingly want to build or contract their own behind-the-meter generation — CPP and Brookfield are capital partners in the Keephills JV, entirely capable of financing generation that competes with, rather than buys from, TransAlta's merchant fleet. Where is revenue concentrated? 58% of capacity in one merchant market — remove the Alberta recovery assumption and the base case collapses. Why is the moat weaker than bulls think? The moat is grid siting, not generation — and a well-capitalised competitor (or the hyperscaler itself) can replicate a Keephills-style interconnect elsewhere in Alberta. Most dangerous competitor bulls underrate: the hyperscalers' own vertical integration into power, plus Capital Power (a better-run, more contracted Alberta peer). Worst capital-allocation move: the 2019 Brookfield hydro financing — 7% money that handed a rival an option to seize half the crown jewel; short-hand, management sold the future to fund the present. What must hold for today's price? That Alberta recovers to ~C$50-60+/MWh by 2027-28 and Brookfield doesn't convert and REM doesn't discount their assets and the balance sheet holds. If growth disappoints 20-30% (pool price stuck at C$30-35), FY2027 EBITDA lands nearer the bear ~C$900M, leverage pushes past 4.5x, and the dividend-growth story ends — the stock is a value trap, not deep value. Single scenario that permanently impairs: Brookfield exercises the 49% hydro top-up into a recovering Alberta market while the merchant fleet is still under water — the crown jewel is diluted at the bottom, permanently.
Research Trail
Covered in the Knowledge Base
Energy
The most physically elegant fusion bet per dollar — a no-magnet, no-laser Z-pinch ~9x more capital-efficient than the tokamak leaders — but the April-2026 fission pivot is the tell: Zap is now a near-term fission-revenue company wearing a fusion halo. Underwrite the falsifiable milestones (independently-verified Q≥1 at ~650 kA; a signed 4S/fission milestone contract), not the 2030s grid dream. WATCHING · MEDIUM.
The cleanest listed way to own physical uranium at a discount — at ~527p the market prices YCA's U3O8 near ~$70/lb vs ~$85 spot and ~$91.50 term; the discount, not a uranium view, is the edge. It closes if the buyback + a firm spot bid hold; the thesis breaks if uranium rolls over or the ~15% structural discount becomes permanent. This is levered beta to one variable with a fee lid, not an operating business.