BankableReal Bet? · The Blue Hydrogen Teardown · one-page verdict

Blue hydrogen: overhyped as a new vector, a real bet as a retrofit

The cheapest clean hydrogen on paper — and four dead projects in eight months. Both things are true, and they are the same truth: the molecule was never the problem, the carbon is. There are exactly three things you can do with it. Companion to the video episode; every number below traces to the cited research file.

Making hydrogen was never the hard part. The carbon is — and it doesn't disappear, it only changes form and address.

Swipe the table to see every column →

Answer	The physics	The economics	Verdict
Bury it — blue SMR/ATR + CCS	fleet captures ~60% by design; standards want 90–95%; methane leakage sits upstream of any capture box	$2.00–3.50/kg vs grey $1.50–2.50; 45Q at $85/t pays back ~$0.77/kg	REAL BET as a retrofit only OVERHYPED as a new energy vector
Sell it — turquoise methane pyrolysis	no process CO₂ at all; 7–12 kWh/kg, ~1/5 of electrolysis; but 1 kg of H₂ always comes with 3 kg of solid carbon	carbon black market ~18 Mt/yr caps the route at ~6% of hydrogen demand	WORTH WATCHING solves carbon honestly, hits a market-size wall
Skip it — burn the gas power for data centers	no hydrogen, no capture, no storage liability	the buyer signs today — the Teesside site is literally becoming an AI data center	THE COMPETITOR what actually outbid blue wasn't green — it was the AI data center

Every figure keyed to the frozen claim ledger [S1–S54]. Full [S#] mapping in the workbook's Sources sheet.

The graveyard: eight months, four projects

ExxonMobil Baytown — one of the largest clean-hydrogen plants ever attempted (~1 Bcf/day), paused November 2025. The CEO's reason: no signed offtake customers.
bp H2Teesside — 1.2 GW, application withdrawn December 1, 2025. The site is now planned as an AI data center. Read that twice.
Air Products — Louisiana: spending stopped, the CCS and ammonia pieces listed for sale. Alberta: announced 2021 at $1.3B for 2024, now $3.3B and end-2027, flagged as an underperforming asset — the CEO called part of the miss self-inflicted.
Suncor + ATCO — 300 kt/yr, announced 2021, FID planned 2024, still no FID; Alberta's major-projects list now says earliest 2029.
Four different companies, four different balance sheets, one shared cause of death: nobody would sign the contract.

It's not a technology problem

Steam methane reforming is a century-old process making ~95% of the world's hydrogen. Shell's Quest has been storing CO₂ in Alberta since 2015 — about 9 million tonnes so far. The equipment works.
Blue is the cheapest clean hydrogen on paper: $2.00–3.50/kg against green PEM at ~$5. Its theoretical floor in cheap-gas regions sits just above $1/kg — but note the label: theoretical, and no operating project reaches it at 95% capture.
What fails is not the plant. It is the business model: a premium product with no premium buyer.

The capture-rate gap — and the methane wall behind it

The running fleet captures about 60% of its CO₂ — by design, not malfunction. The cheap carbon (the concentrated process stream) gets captured; the flue gas is left alone. Standards and subsidies want 90–95%. ATR targets 95% and is projected to take ~37% of new capacity by 2030 — from roughly zero today.
Then the wall no capture box can touch: upstream methane. Over 20 years methane warms more than 80× CO₂. At the US-average 1.5% leakage, even 100% capture still leaves a carbon intensity above 2.5 kg CO₂e per kg H₂. At 0.4% leakage with 90% capture you're near 2 kg. The gas source decides how clean 'blue' can ever be.

Does it pencil? One subsidy carries the math

Grey at $1.50–2.50/kg, blue at $2.00–3.50: a premium of roughly a dollar for a molecule the customer cannot tell apart.
45Q pays $85/tonne for geological storage — and after the July 2025 budget bill, enhanced oil recovery and utilization pay the same $85. Capture ~9 kg per kg of hydrogen and the credit is worth ~$0.77/kg — about three-quarters of the blue-grey gap.
Canada adds a refundable 50% investment tax credit on capture equipment. But Alberta's TIER shows the policy-risk hole: a nominal C$95/tonne industrial carbon price against a compliance-credit market trading near C$20. The spreadsheet says 95; the market pays 20.
So blue pencils where the subsidy is bankable and the storage is already proven — and almost nowhere else. The sector runs on policy, and policy just showed it can be repriced.

Where blue actually works: four conditions

Read them straight off the one big FID that did happen — Shell's Polaris (~650 kt CO₂/yr, Scotford): existing molecule (refinery already uses the hydrogen) · existing storage (Quest, running since 2015) · existing customer (your own plant) · existing permits.
Blue as a retrofit of an asset you already own, decarbonizing a molecule you already sell to yourself. Miss any one of the four and you are on the graveyard list.

Turquoise: the honest answer with a small door

Methane pyrolysis makes no process CO₂ at all — nothing to capture, compress, inject or monitor. It eats 7–12 kWh/kg, roughly a fifth of electrolysis, and no water.
The catch is arithmetic: CH₄ = C + 2H₂, so every kilogram of hydrogen brings 3 kg of solid carbon. Sell it as carbon black and the ~18 Mt/yr global market absorbs at most ~6 Mt of hydrogen — ~6% of demand. A peer-reviewed 2026 analysis flags exactly this: carbon supply outrunning real demand is a fundamental oversupply risk.
It is real: Monolith has run commercially in Nebraska since 2020; BASF and ExxonMobil signed a demo at Baytown (2,000 t H₂ + 6,000 t carbon — exactly 3:1) weeks after pausing blue. Worth watching, not yet a thesis. And it inherits the methane-leakage wall in full.

The third answer: don't make hydrogen

The Teesside site becomes a data center. Monolith is reportedly planning one on its own Nebraska plant. The gas, the grid connection, the water, the permits — everything a blue project assembles is exactly what an AI data center wants to buy.
When the same inputs have two bidders and one signs today at scale while the other asks you to wait for policy, the market has already voted. What outbid blue hydrogen wasn't green or turquoise. It was the AI data center.

China: same physics, different problem

China's hydrogen is dominated by coal: ~20.7 Mt of coal-to-hydrogen at a carbon intensity of ~19–20 kg CO₂/kg — about 2.5× the gas route. Costs: coal ~RMB 12, gas ~RMB 15, electrolysis RMB 20–30 per kg; adding CCS to coal-to-hydrogen costs roughly +RMB 7/kg.
National CCUS capture capacity is ~9.4 Mt/yr against coal-to-hydrogen emissions of ~390 Mt/yr — 2.4%. The tool and the problem differ by about two orders of magnitude. For China the blue question is not price; it is capacity.

Scored

Maturity 7/10 — century-old reforming, a decade of running storage; highest in the series.
Economics 6/10 — the only clean hydrogen that comes within a dollar of the incumbent, and one subsidy covers most of it.
Scale-up 3/10 — storage permits, leakage scrutiny, and four dead projects in eight months.
Moat 4/10 — geology and existing assets are real moats; the molecule itself is a commodity.
Timing 4/10 — 45Q is at its peak and EOR now pays parity; the buyer market just walked next door to data centers.

Bankable — which hard-tech bets actually pencil out. The editable cost workbook came with this download: set the capture rate, gas price and carbon price yourself and watch which answer survives. The video is at bankable.show.

Prepared from public data by David Zheng, P.Eng. — over a decade commercializing first-of-a-kind energy technology. The ~$0.77/kg 45Q conversion and the 6% turquoise ceiling are derived figures (shown in the workbook). Illustrative, not project-grade engineering. We judge technologies, never stocks; nothing here is investment advice. Advisory: bankable.show/advisory · hello@bankable.show