Michael Barnard in conversation with Earth Set

Amy and Michael chat in the Octopus armchairs

September 2026

Featuring: Michael Barnard, Chief Strategist at The Future is Electric, TFIE

At September’s Earth Set, Michael’s central challenge was to the assumptions that sit underneath many of those forecasts in the hardest parts of the energy transition, for example shipping, aviation, hydrogen, steel, cement and energy infrastructure. Too often, he argued, we take today’s demand for a fuel, material or service, extrapolate it forwards, and then ask how to decarbonise it.

But what if the demand itself changes? Barnard’s shorthand for the transition is:

“Molecules shrink to where molecules are necessary. Electrons grow to eat everything that can be eaten.”

Shipping: decarbonisation removes the cargo

Shipping provides perhaps the clearest example.

Around 40% of global maritime freight tonnage is coal, oil and gas. Add raw iron ore and Barnard puts the figure at around 55% of freight tonnage, accounting for roughly 60% of shipping fuel demand because these commodities travel particularly long distances.

So a decarbonised economy does not need to find a clean fuel to move everything ships carry today. A very large part of the cargo disappears.

Barnard contrasts projections of up to 450 million tonnes of future maritime liquid-fuel demand with his own estimate of around 70 million tonnes by 2100. The difference comes both from falling fossil-fuel freight and from electrifying much of what remains: inland shipping first, then short-sea shipping, with transoceanic vessels increasingly hybridised.

So rather than asking how to replace today's bunker fuels molecule-for-molecule, the question becomes how much liquid fuel is actually required after the transition has removed both cargo and energy demand.

Buildings: don't assume China builds twice

The same problem appears in forecasts for cement and construction.

China has dominated global cement demand during an extraordinary period of urbanisation and infrastructure construction. Barnard argues that this is now slowing as China moves from first-build towards a more mature economy, and that simply projecting its historical construction rates forwards therefore badly overstates future materials demand. The cement industry itself is beginning to recognise this. Barnard cited a World Cement Association projection of around a 30% reduction in cement demand.

But his second point was about comparators. The alternative to high-carbon cement is not necessarily an identical quantity of low-carbon cement. Buildings can use less material through structural optimisation and finite-element analysis; different materials can substitute for concrete; and mass timber can replace reinforced concrete in suitable applications. Barnard estimates that one tonne of mass timber can displace around five tonnes of reinforced concrete, while retaining carbon absorbed during the tree's growth.

The relevant question is therefore not simply how do we decarbonise cement? It is how much cement will we actually need?

Hydrogen: “I can make it green, but I can't make it cheap”

Hydrogen was the most obvious example of Barnard's other recurring concern: technologies around which policy and investment can build faster than physical deployment. His objection is not that green hydrogen cannot be produced. It is the assumption that it will become a cheap, general-purpose energy carrier.

“I can make it green, but I can't make it cheap.”

The problem, he argues, becomes clearer when the whole system is included: renewable electricity, electrolysis, compression or liquefaction, storage, transport and then conversion into useful energy. And increasingly there is a physical comparator. In China at the end of 2025, Barnard's analysis counts roughly 18,000 fuel-cell heavy trucks against around 366,000 battery-electric heavy trucks: around 20 battery trucks for every hydrogen one. Across medium- and heavy-duty vehicles, his estimate puts the ratio above 40:1.

His wider point is that announcements, pilots and funding commitments can make technologies appear much closer competitors than their actual deployment suggests. His preferred redirection of European energy spending was characteristically uncomplicated:

“Stop spending money on carbon capture and sequestration and hydrogen. Put it into HVDC and electrification.”

Aviation: does expensive fuel destroy demand?

Aviation produced one of the evening's strongest disagreements. Barnard argues that long-haul aviation is one of the places where molecules genuinely remain necessary. But he expects sustainable aviation fuel to cost perhaps five to seven times more than today's fossil jet fuel, and therefore expects demand to respond.

He describes 91% of aviation as “discretionary”: not that 91% of flights will disappear, as he later had to clarify, but that most flying is undertaken by choice rather than because it is economically unavoidable.

His model therefore does not accept continually rising aviation demand as a given. Shorter journeys increasingly electrify or move to rail; some business travel remains virtual; and higher long-haul ticket prices reduce demand. His resulting projection is around 110 million tonnes of aviation fuel annually by 2100. That assumption was challenged directly from the audience. People do plenty of things that aren't economically productive because they value them. So what government will actually impose a policy that makes flying to Spain substantially more expensive?

Barnard corrected the suggestion that he expected 90% of flights to disappear, but maintained that the price of genuinely sustainable aviation eventually has to feed through into behaviour. It exposed one of the central uncertainties in long-range transition modelling: the economics can suggest one trajectory; politics and human behaviour may produce another.

From invention to deployment

Perhaps the most consequential argument of the evening was also the simplest:

“We don't need many new innovative technologies. Deployment is the problem. Governance is the problem. Program management is the problem. Policy is the problem.”

Barnard estimates that around 90% of the job is now deploying technologies we already have. Wind, solar, batteries, heat pumps, EVs and transmission already work. The harder problems increasingly concern grids, planning, regulation, utilities, finance and public acceptance.

That makes social licence part of the infrastructure problem, rather than an inconvenient communications exercise bolted on afterwards. National support for renewable energy matters little if a transmission line, wind farm, substation or data centre cannot be built where it is actually needed. Barnard argued for earlier community involvement, tangible local benefits and a much clearer sharing of the value created by new infrastructure.

The point came up again during the discussion of data centres: communities should not simply inherit additional electricity demand and higher costs while most of the economic value accrues elsewhere. New infrastructure needs to demonstrate what it contributes locally, whether through heat, grid resilience, investment or other benefits.

Electrification, but make electricity cheap

That creates a particular problem for Britain. The UK wants households and businesses to electrify transport, heating and industry while maintaining a large gap between the cost of electricity and gas.

Barnard's prescription: “Make electricity cheap. Electricity can be green. Gas can't be green.”

He pointed to gas setting marginal electricity prices, grid constraints and curtailment, and the way taxes and policy costs are distributed between electricity and gas.

Whatever the precise package of market reform, the contradiction is obvious: persuading households to buy heat pumps and businesses to electrify industrial processes is considerably harder while electricity remains the expensive energy carrier. This also returned the discussion to Barnard's broader point. The principal barriers are increasingly not missing technologies. They are the systems surrounding them.

Getting the assumptions right

The conversation ranged considerably further, through carbon capture, data centres and AI, aviation offsets, seasonal thermal storage, Canada and the changing politics of energy security.

Not all of Barnard's conclusions went unchallenged. Nor should they. His projections depend on strong assumptions about future demand, prices and behaviour, particularly in aviation and industrial materials.

But the more useful challenge was methodological. Before attempting to decarbonise today's demand, ask how much of it survives the transition. Before backing a technology, compare it with all the plausible alternatives, not simply the incumbent. Look at deployment rather than announcements. And test the economics of the whole system, not just the interesting component.

As Barnard put it: “I don't claim to be right. I just claim to be less wrong than most.”

For a transition involving trillions of dollars of capital allocation, being less wrong would be a useful start.