My first brief for COP31 Türkiye Brief argued that how energy-system models represent solar — generic plant or a differentiated menu of rooftop, utility-scale, tracking, floating and prosumer PV — is not a neutral modelling choice. It shapes which technologies look competitive, and by extension, who gets served first when a country decides where to invest.
That argument focused on where solar sits in the system. There is a second, closely related fork that deserves the same scrutiny: once renewable electricity is generated, how does a model assume it reaches an end use? Directly, as electricity powering a heat pump, an electric vehicle or a factory motor? Or indirectly, converted first into hydrogen or a hydrogen-based synthetic fuel, and only then delivered to a boiler, a ship or a steel plant? The answer a model assumes — often built into its structure well before any scenario is run — has its own distributional consequences, and they compound the ones raised by how solar itself is represented.
The direct/indirect fork, in numbers
A 2024 study using the REMIND integrated assessment model to explore EU pathways to climate neutrality found that direct electrification is consistently the dominant strategy: across a range of scenarios, electricity’s share of final energy in 2050 settles between 42% and 60%, while hydrogen and hydrogen-based synthetic fuels contribute a smaller 9% to 26% (Schreyer et al., 2024). Crucially, the two strategies are not evenly spread across the economy. Direct electrification wins almost everywhere it can — passenger cars, low-temperature heating in buildings — while hydrogen-based routes concentrate in the sectors that are genuinely hard to electrify: aviation, shipping, chemicals feedstock, and primary steelmaking. The study estimates that only around 15% of final energy is a genuine toss-up between the two strategies; for the rest, the sectoral role is fairly clear once realistic technology and cost assumptions are used.
A separate, more granular study of Danish 100% renewable energy scenarios reaches a compatible conclusion from industry data. Testing different levels of electrification against three independently developed Danish energy-system scenarios, it found that direct electrification of industrial process heat should generally be favoured over switching to hydrogen-based process systems, even in cases where hydrogen would provide more operational flexibility to the system (Sorknæs et al., 2022). The efficiency losses in producing, storing and reconverting hydrogen are large enough that, outside of applications where no electric alternative exists, direct use of electricity remains the lower-cost, lower-emissions choice.
Why does this matter for a brief about modelling and equity? Because how confidently a model captures the value of variable renewables — not just their cost — turns out to swing outcomes by roughly the same order of magnitude as the direct/indirect split itself. Work incorporating a profitability-adjusted measure of levelised cost into a global integrated assessment model found that representing how a technology’s value changes as its market share grows, rather than treating solar and wind at a flat cost, shifted their combined 2050 electricity share across a range as wide as 53% to 81% depending on assumptions alone (Gurgel et al., 2023). A model that under-represents this dynamic will tend to under-value distributed and variable resources precisely where they are most useful — which loops back to the argument in my first brief about rooftop and prosumer PV being systematically under-modelled.
The same country, two models, two different answers
If the cross-study variation above still leaves room to argue that different countries simply have different resource endowments, a more recent sector-coupled study of the German industrial sector closes that gap. Using an enhanced version of the REMod energy-system model with detailed representation of production routes for steel, chemicals, ammonia, methanol and cement, the study runs four scenarios spanning a deliberately engineered range from favouring direct electrification to favouring indirect electrification via hydrogen and synthetic fuels (Kaiser et al., 2025).
Direct electrification turns out to be robust across the board: even in the scenario most favourable to hydrogen, electricity still supplies 47% of German industrial final energy by 2045, against 52% in the electrification-favouring case — a narrow four-point spread. But that headline number hides sharply uneven certainty underneath it. For crude steel, chemicals and light-duty vehicles, the electrification share barely moves across scenarios (under 5 percentage points), meaning the technology mix is effectively already decided. For trucks, cement and high-temperature industrial process heat, the spread exceeds 20 points — these are the genuinely contested sectors where scenario assumptions, not physics, decide the outcome.
The more striking result, for the purposes of this brief, is what happens when the authors compare their own model’s output to Germany’s official Long-Term Scenarios project, which uses a different, more granular sector-specific model (FORECAST, resolving over 60 industrial sub-processes against this study’s 16). For the same country, the same target year and broadly similar narratives, industrial electricity share comes out at 65–71% in the sector-coupled model used here, versus 45–64% in the sector-specific government reference scenarios. The authors attribute the gap not to a disagreement about German industry’s physical potential, but to how much technological detail and cross-sector coupling each model resolves. This is close to a controlled experiment for the claim running through both of my briefs: the model’s architecture, not just the country or the resource base, can move the answer by twenty percentage points.
There is an equity dimension buried in the same results. Higher direct electrification systematically reduces German industry’s dependence on imported hydrogen and synthetic fuels — 134 TWh of imports in the electrification-favouring scenario versus 253 TWh in the hydrogen-favouring one. Whichever pathway a national model favours therefore also decides, by construction, whether that country remains a durable importer of green hydrogen or reduces that dependency over time — and, on the other side of the trade, which countries are being modelled into the role of long-term hydrogen exporter. A model that defaults toward indirect electrification without interrogating why is not just a technical choice about heat pumps versus boilers; it is quietly assigning countries and regions to one side or the other of a future energy trade relationship.
Where decentralized electrification fits
This is where the direct/indirect distinction meets the decentralization argument from my last piece. Direct electrification — a rooftop system paired with a heat pump, a battery and an EV, or a mini-grid serving a village clinic — is inherently a technology set that can be deployed incrementally, close to where people live, without waiting for large shared infrastructure to be built first. It is, in other words, the natural electrification pathway for the roughly 675 million people worldwide still without electricity access, and for the decentralized systems that reviews of progress on SDG 7 estimate could serve around 41% of that unelectrified population.
Indirect electrification runs on a different logic. Producing hydrogen or synthetic fuels at meaningful scale requires centralized investment: electrolysers sized for industrial offtake, dedicated transport and storage infrastructure, and in many proposed strategies, an export market. None of this is a criticism of hydrogen as a technology — for shipping, aviation and primary steel, there may be no realistic substitute, and the studies above are clear that some hydrogen buildout is indispensable. But it does mean that the beneficiaries of an indirect-electrification-heavy pathway are, by construction, different from the beneficiaries of a direct-electrification-heavy one: heavy industry, export-oriented producers and long-haul transport operators on one side; households, small businesses and underserved communities relying on direct, often distributed, electricity on the other.
If a national energy model treats “electrification” as a single aggregated demand-side variable — without distinguishing which end uses are met directly and which are routed through a hydrogen or synfuel intermediary — it cannot show a government which of these two beneficiary groups its investment plan is actually prioritising. The 2022 National Energy Plan for Türkiye, built on the Türkiye Energy Model referenced in my first brief, projects rising solar capacity and a role for hydrogen, but the public documentation does not make clear how the plan’s optimization allocates end uses between direct and indirect electrification, or what assumptions about hydrogen cost and value drive that split. That is not a claim that the underlying choices are wrong — only that, as with the solar representation question, the assumption is largely invisible from the outside.
What this adds to the list
My first brief closed with a list of distinctions a model designed for an equitable transition might need to preserve: where generation is located, how it interacts with networks, whether it sits on land or rooftops, how consumers themselves participate. To that list I would now add: which end uses are assumed to be met by direct electricity and which by hydrogen-based fuels, and on what cost and value assumptions that split rests. Getting the solar side right is necessary but not sufficient. A model can represent rooftop, prosumer and utility-scale PV in exhaustive detail and still steer investment away from the households and communities it should be reaching, if the demand side quietly assumes that decarbonizing heat, transport or industry runs mostly through a centralized hydrogen economy rather than through the electricity those same distributed resources can supply directly.
