Burden of Adjustment
Monday 3 March 2025
The evolution of global power generation adheres to the Additive model of energy transition. In the Additive model, wind and solar maintain their fast and beautiful expansion, but wind up as a layer on top of legacy sources. This is in contrast to the Transformative model of energy transition, which looks at a chart like the one below and assumes that wind and solar will, after reliably taking 100% of annual marginal growth, get to work on the underlayer of incumbent energy sources in power, forcing them into decline. But there is no historical, nor economic justification for such a view. The Transformative model represents nothing more than the current consensus reality, shared by many climate observers.
Notice that starting in 2018, annual additions of wind and solar began to cover 100% of total system growth. To be sure, starting in 2021, these other sources did grow again, rising up beyond that 25000 TWh level. That said, wind and solar have unquestionably helped suppress the growth rate of power sources, ex wind+solar since that time. The Additive model embraces and explains this outcome. In the Additive model, the new energy sources not only begin to dominate marginal growth, but by doing so they effectively suppress the growth of legacy sources. However, because the model conforms to the Jevons effect, the default trajectory of global power will eventually see those other sources either growing again, or maintaining their hold, as the global economy continues its expansion.
Editor’s Note: Cold Eye Earth is now aligned with the Additive model of energy transition in which renewables and other clean energies are, on the current trajectory, destined to wind up merely as a new layer of energy, sitting on top all legacy energy sources. The advantage to this alignment is manifold.
First, the Additive model is not in conflict with the continued growth of renewables, whose value proposition gets ever more robust. Solar in particular continues to impress for its cost and speed, and there is little question that eventually solar, along with wind and storage, will absorb 100% of marginal growth in global power.
Second, the Additive model aligns with energy history. While each of the previous transitions saw a new energy source supersede the incumbent source for a while—setting off a slowdown in its growth rate—all transitions eventually incorporated the incumbents into the overall mix, with each source continuing to expand, at different rates of course. Nothing illustrates this general arc more clearly than humanity’s 275 year relationship with coal, and you can read a compact version of that history in Cold Eye Earth’s February 3 issue, Momentum Lost.
Third, the Additive model reminds us that energy history shares the stage with economic history, and that it’s not just energy technology which transforms the world, but the “pull” from economic growth, and population growth, which are highly determinative of outcomes. As Cold Eye Earth has shown, the number that should interest us most is not annual renewables growth, but rather, broader economic growth and energy demand. Jevons was correct.
Fourth, the Additive model conforms to the whole earth, whole system view of the climate problem, which knows no borders. It’s completely unsurprising, for example, that slow-growing regions like Europe are able to achieve emissions declines, when the bulk of global manufacturing sits in China. Indeed, decarbonization success in one domain is still largely offset by the continued growth of carbon in other domains—not a new or novel idea.
Finally, the Additive model unmasks the weakness of the Transformative model, how it’s propped up with unsupported assumptions (consensus reality), and has no historical example. The Transformative model, which assumes the economic attractiveness of renewables will successfully impose their discipline on less efficient incumbent sources and their delivery infrastructures—driving them out of the market—actually has no solution to the turnover problem, in which legacy energy sources and their applications don’t lose enough of their value to be thrown away. Instead, legacy investments consistently retain a “good enough” set of qualities that allows them, whether cars or power plants, to trudge on for many years.
Indeed, the Transformative model may have borrowed too many assumptions from the diffusion curves of new technologies, where consumers typically can afford to write-down legacy products to zero. A new mobile phone can easily drive out a landline. But even the latest and best utility-scale solar plant doesn’t render the bulk of incumbent power uneconomic. In short, the Additive model accepts that the new energy source, in this case wind and solar, is excellent at covering marginal growth, but not very good at all at “replacing” existing incumbent sources—except perhaps on very long timelines, measured in decades.
In just twenty-five years, China’s manufacturing output has gone from being roughly half of US output, to double US output. And there’s no indication this trend is going to halt, either. In share terms, China’s value-added manufacturing has gone from a single digit share of global output at the start of the new century, to over 30% of the world’s output today. This manufacturing output is in service to the world economy—it is not the footprint of China’s demand, but our demand—and it’s confirmation that the world remains very much within the confines of the industrial age. To be more explicit, the existence of post-industrial regions from France to Canada, and from the US to Germany is a local, not a global phenomenon. There is no post-industrialism (yet) in China, however, because the world itself is not post-industrial.
Just to remind, global emissions grew again last year, because fossil fuel consumption grew again last year, because we share one planet, one global economy, one global system of manufacturing output and trade, all translated into collective human energy consumption.
The persistence of coal growth, and natural gas growth, and electricity growth in China is therefore our growth, here in the West. Yes, it’s nice that humans have built leading-edge states like California, dominated by knowledge-work, creative design, and very little manufacturing. Emissions are in decline in California, unsurprisingly. Petrol consumption is in decline, natural gas consumption is in decline, and coal has been (nearly) zeroed out of the system. But that’s because the ideas, inventions, and products conceived of in California are manufactured elsewhere. One of the other pitfalls to the Transformative model therefore is that it treats states like California as a proof-of-concept that energy transition inexorably leads to actual decarbonization. Presumably, it’s not necessary to name the logic flaw in that view. But, if you insist: The Fallacy of Composition.
Human societies are free to choose the path of the Transformative model. But that would require economic pain, and how to share that pain is the unsolvable problem in current politics. Because the Transformative model has no solution to the turnover problem or to the path dependency of legacy energy solutions, it needs to add an entirely new component to its case to achieve success: the burden of adjustment. Who will bear that particular burden? The burden of shutting down economically viable power plants, and vehicles? As an example, for the Transformative model to work, we would need to see China placing a moratorium on new coal plants; the US placing a moratorium on new natural gas plants; and both countries engaging in a compensation scheme for owners to shutter still-viable coal and natural gas plants in both countries. Apply the same scheme to cars. Because, without such schemes the Additive model is in control, and therefore the thesis of declining emissions will not come true, and will remain “right around the corner, surely about to happen,” for decades.
The evolution of renewables in US power generation also adheres to the Additive model of energy transition. As Cold Eye Earth has explained previously, the retirement of aging coal plants and uneconomic natural gas plants over the past decade has predictably resulted in a younger power fleet. That’s what happens when you construct vast new natural gas power capacity. Wind and solar meanwhile look great. They continue to expand right up the borderline of legacy sources, and then halt, having done their job of covering all, or nearly all, marginal growth.
Now, if the US had a policy that mandated all marginal growth in its power system had to be met by wind, solar, and storage then there would be little chance natural gas in US power would keep growing. But we know already that wind and solar, over the past decade, had to share with natural gas the opportunity created by successive coal closures. That pattern looks set to continue as 1. the US closes more coal. 2. builds more natural gas capacity, and 3. covers the bulk of total demand growth with new wind and solar. This is the stair-step, ratchet effect that has produced the Additive outcome in previous transitions, in which each year sees incremental additions to the fossil fuel underlayer—just enough to keep it intact, and rolling forward.
Wind and solar compete well for marginal growth, edging out the bulk of other power solutions to be built from scratch. But legacy power generation also competes well, extremely well, and is highly resistant to any sort of (net) erosion effect as renewables pile up along its border.
Economic growth needs to be placed at the center of all climate discussion, and decarbonization forecasts. The IMF just updated its outlook for this year and next, projecting global growth of 3.3% both this year and next. That’s below the historical average of 3.7 percent during the 2000–2019 period, but is still pretty robust. The IMF further estimates that last year saw 3.2% growth, and most forecasts of global emissions growth peg last year’s advance right around 1.00%.
We also have an overly strong growth outlook from the IEA, which estimates that global power demand grew by 4.3% last year, and will grow 4.00% both this year, and next. When we consider both forecasts, it’s not hard to project the following: global emissions will grow again by around 1.00% this year (which is alot, by the way) and renewables will be tested again, as they attempt to cover marginal growth in power.
Factors that could shift the forecast: a recession in the US, or China; extreme weather, causing heating demand or air-conditioning demand to spike; mild weather, which could greatly relax cooling demand, thus easing demand off the coal and natural gas variables. The world could also experience a stronger than expected year of global growth, which would probably boost oil, too. Right now, oil growth is extremely slow, and has actually been contributing to slower emissions output.
What’s important is to get these growth forecasts up on your dashboard first, before trying to count up renewables to get answers. Counting renewables is a largely static exercise, and therefore reliably misses the dynamics of systemic growth. So again, for those who wish to engage in peak emissions forecasting, show your economic model first, and your demand models too, before you try to get all the answers exclusively from renewable supply.
Recommended Reading: AI Data Centre Power and Glory - An Update, by Michael Liebrich (2025) • The Thucydides Trap: Are the U.S. and China Headed for War?, by Graham Allison (2015) • DeepSeek Debates: Chinese Leadership On Cost, True Training Cost, Closed Model Margin Impacts, by SemiAnalysis (2025) • Memo to the President: Manhattan Economy Improving, Thanks to Congestion Pricing, by Dave Colon (2025) • How to build data centers without raising grid costs — and emissions, by Jeff St John (2025) • The Troubled Energy Transition - How to Find a Pragmatic Path Forward, by Yergin, Orszag, and Arya (2025) •
—Gregor Macdonald