Rock, Not Gas

Alberta can turn wellhead gas into near-zero-CO₂ power for data centres by splitting methane into hydrogen and solid carbon, and selling the carbon to aluminum smelters as a substitute for anode coke. The power pays the bills. The carbon is the additional revenue stream, and, it turns out, the real ceiling on how big this gets.

The way I like to frame it is similar to the petroleum revolution, where oil was discovered to have many byproducts that were useful for a huge variety of things, from metals to pharmaceuticals. That diversity of uses is what created the long-term economic durability and surpluses of the oil industry. There are similar technologies and opportunities for other energy sources emerging now.

1The pain points

Alberta has more than 20.6 GW of data-centre load asking to connect to a grid that peaks at about 13 GW, and as of June 2026 it makes every large project bring its own power. There is a clearly identifiable market opportunity to generate power for these behemoth projects, which are power-constrained, well funded, and willing to pay to get their required power on an expedited timeframe.

Power constraints, especially within Canada, are only deepening. On-site turbines are sold out for years: GE Vernova's backlog hit 100 GW in April 2026, with only about 10 GW of slots left across 2029–30. More cannot be produced without foundries, which themselves sit on a multi-year timeline with massive capex, and high-end blade casting is effectively a duopoly between Howmet Aerospace and Precision Castparts. (I wrote about that bottleneck in more depth here.)

Finally, the long-term political and ultimately economic viability of data centres depends on avoiding massive amounts of emissions. Gas without capture creates a carbon liability as Alberta's carbon price climbs toward $130/t, and capture needs pipelines and sequestration wells most sites don't have nearby.

2Why now

The ideal timing to take advantage of this gap in the buildout was several months ago, but methane pyrolysis is being commercialized, enabling a new niche to be targeted.

Pyrolysis is the process of heating chemicals until they are broken apart into simpler components. Methane can be heated to between 800–1,200°C to separate hydrogen and carbon: every 4 kg of methane yields 1 kg of hydrogen and 3 kg of solid carbon. The hydrogen can be used to power fuel cells, while the carbon comes out as a solid, like soot or sand. Even better, no CO₂ is made in the process, as the carbon in the methane is never put in contact with oxygen.

Two companies that I am aware of currently work with this process. Monolith is commercial in Nebraska, selling its carbon to Goodyear for tires, and Alberta's Aurora Hydrogen, backed by Shell, Chevron and Williams, targets commercial units in 2027.

Wellhead gasMontney, Alberta

→

Pyrolysis reactorAurora Hydrogen · no CO₂ made

1 kg hydrogen → power

Fuel cellsBloom, PowerCell

→

Data centre20–50 MW, on site

3 kg solid carbon → anodes

Carbon shippedRail to Quebec

→

Aluminum smeltersRio Tinto, Alcoa, Alouette

The final piece that makes the timing unique is that additional revenue stream. Aluminum smelters need a special carbon, called anode coke, to make aluminum. That coke is a leftover from heavy-crude refining, and as refineries shift to light crude like US shale oil, supply is shrinking while smelters still need the same amount. One industry report put prices up 80–100% in 2026, to US$800–900/t. The fortunate byproduct of pyrolysis is that the solid carbon produced could be of a similar quality, meaning it can be used in the aluminum process.

Pyrolysis leaves roughly half of methane's energy locked in the carbon, but Alberta gas prices can offset that. A site burns about C$28 of gas for every MWh of power sold, and earns about C$51 back from the carbon made in the process.

The main headwind is that buyers pay for speed, not cleanliness. NET Power, the best-funded oxy-fuel company, shelved its technology in August 2026 to sell plain gas power instead, and Alberta's 7.5 GW Wonder Valley gas project has stalled at the proposal stage amid consultation disputes.

3Sizing it

The assumptions:

• One site is a 100 MW data centre running 90% of the year: about 790,000 MWh a year and, at ~0.15 t of carbon per MWh, about 120,000 t of carbon.
• Carbon sells at US$250/t, half the US$490/t anode-coke index, to allow for an unqualified product.
• Power sells at C$80/MWh: Alberta's 2025 pool price was C$43.68/MWh, plus a premium for firm, near-zero-CO₂ power with no queue.
• By 2040, 2 GW of Alberta's 20.6 GW queue is built this way. Carbon beyond the ~1.3 Mt Canadian smelters can absorb (3.3 Mt of aluminum) sells into steel or construction at US$100/t.

If it works, every day a 100 MW site ships about three railcars of solid carbon to Quebec instead of putting roughly 1,000 t of CO₂ into the air. That's about 355,000 t a year, comparable to taking 75,000 cars off the road. The ~108 t of hydrogen made alongside that carbon each day keeps the full 100 MW running, about 790 GWh a year, without a single turbine. That takes 100 MW of orders out of a queue where GE Vernova can build only about 20 GW a year.

4Competition

The major competition comes from competing technologies. Because hydrogen is supply-constrained, the main threat is from dirty on-site power (firms like Bloom, and VoltaGrid, which raised $1B and signed a 2.3 GW deal with Oracle) and clean firm power (closed-loop geothermal like Rodatherm, and small modular reactors in the 2030s).

The final competitive risk is if an oil incumbent or smelter vertically integrates this, and does not license or purchase the technology from one of the firms doing the initial commercialization.

5What a venture would need

A venture pursuing this would need to establish three things:

• The reactors. Buy or otherwise source the Aurora units.
• A smelter. Quebec has the concentration, with eight smelters producing 2.9 Mt of Canada's 3.3 Mt of primary aluminum.
• A hydrogen buyer. The market is large but almost entirely dirty: the world used nearly 100 Mt of hydrogen in 2024, and less than 1% of it was made with low emissions.

There are three ways the venture could go to market:

• Supply existing industrial users. Sell pyrolysis hydrogen to Alberta refineries and fertilizer plants, which already use grey hydrogen made from gas without capture. There is no data-centre dependency and the buyers already exist, so this is the lower-risk path.
• License to a smelter. Put pyrolysis units at or near a Quebec smelter, which produces its own anode carbon and sells the hydrogen locally. This turns the vertical-integration risk into the business model.
• Build with Aurora. Partner with Aurora and build the data centre as the anchor customer. This is the most ambitious option and the closest to the thesis.

6What kills it

Two things would be fatal. The first is that the carbon fundamentally fails to meet the anode spec. The second is that the carbon price collapses, either in the market (about 11 sites would saturate Canadian demand) or because it is simply not economically viable to produce with this method.