Understanding, measuring, and projecting technology growth in energy transitions
Doktorsavhandling, 2026
Meeting climate targets depends on how fast low-carbon technologies replace fossil fuels. Wind and solar power are now among the cheapest sources of new electricity, which many take to mean their growth will be rapid, seamless and uniform. Techno-economic models, in which deployment is largely cost-driven, project much the same. But how does growth actually unfold, and how far ahead can it be anticipated?
This thesis shows that growth instead proceeds in four phases – formative, accelerating, steady, and slowdown – each shaped by drivers and barriers that shift as technology and policy co-evolve. It develops tools that analyse growth as it unfolds, without imposing a fixed shape or destination, and locate shifts between phases directly in deployment data. The findings contradict three common assumptions. Growth does not follow a simple S-curve: for wind and solar, initial acceleration typically ends at around 3% of a country's electricity generation, far earlier than an S-curve's midpoint implies, giving way to prolonged steady growth punctuated by stalls and re-accelerations. Learning rates are not constant: across 145 cost series, the cost decline per doubling of deployment steepens as technologies mature, yet annual cost reductions slow, because each doubling takes longer. And technologies do not spread evenly: across more than 90 technologies in over 200 countries, later adopters pass through the phases in less time but deploy more slowly and saturate lower, consistent with a weaker capacity to absorb new technologies.
A hindcast-validated probabilistic model, PROLONG, projects a baseline for what is likely given the growth observed across countries, rather than what is cost-optimal or required by a target. Its central projections for wind and solar grow much like the IPCC's 2 °C-compatible pathways, and faster than current policy scenarios. The COP28 pledge to triple renewables by 2030 sits near its 95th percentile, and would require major economies to accelerate wind and solar growth by 1.4 to 5 times. Together these findings replace a smooth, cost-driven picture with a phase-based one, in which barriers change from phase to phase rather than fading as technologies become cheap, and the support that sustains growth evolves with them.
This thesis shows that growth instead proceeds in four phases – formative, accelerating, steady, and slowdown – each shaped by drivers and barriers that shift as technology and policy co-evolve. It develops tools that analyse growth as it unfolds, without imposing a fixed shape or destination, and locate shifts between phases directly in deployment data. The findings contradict three common assumptions. Growth does not follow a simple S-curve: for wind and solar, initial acceleration typically ends at around 3% of a country's electricity generation, far earlier than an S-curve's midpoint implies, giving way to prolonged steady growth punctuated by stalls and re-accelerations. Learning rates are not constant: across 145 cost series, the cost decline per doubling of deployment steepens as technologies mature, yet annual cost reductions slow, because each doubling takes longer. And technologies do not spread evenly: across more than 90 technologies in over 200 countries, later adopters pass through the phases in less time but deploy more slowly and saturate lower, consistent with a weaker capacity to absorb new technologies.
A hindcast-validated probabilistic model, PROLONG, projects a baseline for what is likely given the growth observed across countries, rather than what is cost-optimal or required by a target. Its central projections for wind and solar grow much like the IPCC's 2 °C-compatible pathways, and faster than current policy scenarios. The COP28 pledge to triple renewables by 2030 sits near its 95th percentile, and would require major economies to accelerate wind and solar growth by 1.4 to 5 times. Together these findings replace a smooth, cost-driven picture with a phase-based one, in which barriers change from phase to phase rather than fading as technologies become cheap, and the support that sustains growth evolves with them.
technology growth
technology diffusion
probabilistic projections
climate change mitigation
energy transitions
Författare
Avi Jakhmola
Chalmers, Rymd-, geo- och miljövetenskap, Fysisk resursteori
Policy-driven growth of technologies to accelerate climate action
Nature Reviews Earth and Environment,;Vol. 7(2026)p. 235-252
Reviewartikel
Rethinking S-curves for policy-driven energy technologies
Joule,;(2026)
Artikel i vetenskaplig tidskrift
Probabilistic projections of global wind and solar power growth based on historical national experience
Nature Energy,;Vol. 11(2026)
Artikel i vetenskaplig tidskrift
H. Wang, A. Jakhmola, D. Tong, Q. Zhang, K. He and J. Jewell (2026). The paradox of faster learning and slower cost reduction as technologies mature.
A. Jakhmola and J. Jewell (2026). Declining international support jeopardises deep decarbonisation.
For more than a century, wind and solar power were too expensive to be considered viable. That is no longer true. They are now among the cheapest ways to make electricity. The question today is not whether the world can move away from fossil fuels but how fast it will. It is tempting to assume that cheap technology will usher in a transition that unfolds rapidly, smoothly and uniformly. This thesis asks whether that is true.
It draws on two centuries of evidence, covering more than 90 technologies in over 200 countries, from first use until growth stopped. That evidence contradicts three common assumptions. First, growth is not smooth. Technologies grow in phases. For wind and solar, the fastest phase ends much sooner than expected, and what follows is a long period of growth in fits and starts, as markets, technological systems, government policies and public support pull in different directions. Second, costs do not fall at a steady pace. They fall fastest while a technology’s growth is accelerating. As growth slows, so does the decline in cost. Third, technology does not spread evenly. Countries that start late move through these phases in less time, yet usually build less, plausibly because they have less of the skills, institutions and finance needed to put it to use.
Together these findings replace the picture of a cheap technology spreading by itself with a more complicated one, in which growth depends on politics, capacity and sustained effort. The thesis develops ways to identify which phase a technology has reached while it is still growing. It also builds a model that translates the evidence on growth already under way across countries into a projection of what future deployment is likely, rather than what would be cheapest or what climate targets demand. That projection is consistent with scenarios where the world warms by about two degrees but falls short of more ambitious international goals. Meeting them would demand accelerating global deployment to speeds few countries have sustained so far. The work does not end when clean technology becomes cheap. What stands in the way changes as a technology matures, from cost and viability to grids, land and politics, and the support that keeps it expanding must change too. How fast technology grows is not predetermined but depends on choices that societies make.
It draws on two centuries of evidence, covering more than 90 technologies in over 200 countries, from first use until growth stopped. That evidence contradicts three common assumptions. First, growth is not smooth. Technologies grow in phases. For wind and solar, the fastest phase ends much sooner than expected, and what follows is a long period of growth in fits and starts, as markets, technological systems, government policies and public support pull in different directions. Second, costs do not fall at a steady pace. They fall fastest while a technology’s growth is accelerating. As growth slows, so does the decline in cost. Third, technology does not spread evenly. Countries that start late move through these phases in less time, yet usually build less, plausibly because they have less of the skills, institutions and finance needed to put it to use.
Together these findings replace the picture of a cheap technology spreading by itself with a more complicated one, in which growth depends on politics, capacity and sustained effort. The thesis develops ways to identify which phase a technology has reached while it is still growing. It also builds a model that translates the evidence on growth already under way across countries into a projection of what future deployment is likely, rather than what would be cheapest or what climate targets demand. That projection is consistent with scenarios where the world warms by about two degrees but falls short of more ambitious international goals. Meeting them would demand accelerating global deployment to speeds few countries have sustained so far. The work does not end when clean technology becomes cheap. What stands in the way changes as a technology matures, from cost and viability to grids, land and politics, and the support that keeps it expanding must change too. How fast technology grows is not predetermined but depends on choices that societies make.
Mechanisms and actors of Feasible Energy Transitions (MANIFEST)
Europeiska forskningsrådet (ERC) (950408), 2021-03-01 -- 2026-02-28.
Drivkrafter
Hållbar utveckling
Styrkeområden
Energi
Ämneskategorier (SSIF 2025)
Energisystem
DOI
10.63959/chalmers.dt/5943
ISBN
978-91-8103-486-8
Doktorsavhandlingar vid Chalmers tekniska högskola. Ny serie: 5943
Utgivare
Chalmers
EA, Hörsalsvägen 11, Chalmers.
Opponent: Prof. Dr. Evelina Trutnevyte, Renewable Energy Systems group, University of Geneva, Switzerland