In my previous piece, “Balance of Power – Assault of the Cost Curves,” I focused on the economics — how falling cost curves for solar, wind and storage are reshaping capital allocation. This post examines the system-level consequence of that shift. When capital flows move, infrastructure follows. When infrastructure scales, the energy system flips. I refer to this structural phase shift — where electrification, storage and cost curves converge to reorganise the global system — as Bettrification.
Energy transitions do not flip in a single dramatic moment. They shift in layers. Capacity tips first, as new technologies outbuild the old. Firming tips next, as storage stabilises high-penetration grids. Generation tips last, when most electricity comes from the new regime.
Using historical build rates and current deployment trends, I’ve mapped what the next decade looks like if today’s acceleration continues. The chart and accompanying data table below form the backbone of this analysis — a structured projection of installed capacity (GW), storage discharge (GW), and electricity generation (TWh) through 2035. I’ve been refining and stress-testing these figures for the past two months, iterating assumptions against real-world deployment data and utilisation trends. No heroic assumptions. No speculative policy leaps. Just compounding growth already visible in global installation data, extended forward with consistent methodology.
Over the past two months I have stress-tested multiple variables: solar and wind CAGR ranges against historical build-out data, storage duration assumptions as average fleet hours expand (with average utility-scale duration moving toward ~4.5 hours by 2030 and approaching ~9 hours by 2035 as longer-duration chemistries scale), fossil decline sensitivity based on utilisation compression rather than abrupt shutdowns, and total demand growth trajectories incorporating AI, electrification and industrial load. The projections shown are not a single straight-line extrapolation, but the result of iterative scenario testing within realistic deployment bands.
The crossover years are structural:
• 2027 – Solar installed capacity overtakes fossil capacity
• 2030 – Solar + Wind generate 56% of global electricity
• 2030 – Storage discharge reaches parity with fossil capacity
• 2035 – Solar + Wind generate 83% of global electricity
Chart: Global Power System Crossovers (2016–2035)
Figure 1. Installed capacity (GW), storage discharge capacity (GW), and electricity generation (TWh) projections through 2035. The chart visualises the layered crossover: capacity → firming → generation.
Data Table: Capacity, Storage and Generation Projections
Table 1. Underlying numerical projections used in the model, including solar, wind, fossil capacity (GW), storage discharge (GW), generation (TWh), and duration assumptions.
Quantitative Snapshot
• Solar capacity scales ~4× from ~2.5 TW (2025) to ~9.2 TW by 2030
• Storage discharge scales from sub‑1 TW to ~3.7 TW by 2030
• Fossil capacity peaks before 2030 and trends downward thereafter
• Total electricity generation continues rising throughout the transition
This is not ideology. It is arithmetic. Cost curves drive build rates. Build rates shift capacity. Capacity shifts utilisation. Utilisation shifts profitability. Profitability reshapes the system.
Curtailment Is the Catalyst
Curtailment is often framed as a weakness of renewables. In reality, it signals overbuild. As solar penetration rises, midday prices collapse. Merchant margins compress. Unpaired projects are clipped first. Within one to two build cycles, the market adapts.
No developer willingly spills electrons at zero or negative prices. Storage captures the margin. Solar-plus-storage becomes the rational configuration in high-penetration grids. Volatility is the market forcing pairing and accelerating firming.
Curtailment is not the problem. It is the trigger.
Firmed VRE and the Demand Surge
Electricity demand is rising again, driven by AI workloads, hyperscale data centres, electric vehicles, heat pumps and industrial electrification. The instinctive response from legacy thinking is that coal must be prolonged or nuclear massively expanded to meet that demand.
But the data shows something different. Annual solar and wind additions already exceed new fossil additions. Storage is compounding. Total generation in the model continues to rise even as fossil capacity declines.
In the projections shown, “firmed VRE” is not a slogan — it is implicitly represented by the interaction between solar + wind capacity and storage discharge capacity. The binding constraint for reliability is not raw nameplate renewables, but dispatchable discharge (GW) available during non-solar hours. By 2030, storage discharge reaches parity with fossil capacity in the model, assuming average fleet duration approaches roughly 4.5 hours by 2030 and extends toward ~9 hours by 2035 as longer-duration storage becomes mainstream. The duration assumption matters: discharge capacity (GW) reflects how much power can be delivered at once, while duration (hours) reflects how long that power can be sustained. Energy (GWh) is simply power × time. A 1 GW battery running for 4.5 hours delivers 4.5 GWh. At ~4.5 hours, storage primarily covers evening ramps and intra-day variability; at ~9 hours, it increasingly covers overnight gaps and multi-hour wind lulls. This means the system possesses equivalent firming power (in GW terms) by 2030 to manage non-solar hours, and materially deeper resilience by 2035. Storage discharge parity in 2030 does not imply fossil exit, but rather equivalent firming power in instantaneous terms; duration and utilisation determine the depth of displacement.
I have also stress-tested higher demand scenarios, including accelerated AI deployment and faster electrification uptake. Under higher demand trajectories, crossover years shift modestly — typically by one to two years — but the structural direction remains unchanged. The system absorbs incremental load through continued compounding of VRE and storage, rather than through prolonged fossil expansion.
Firmed variable renewable energy is therefore scaling quickly enough to absorb incremental demand growth without structurally extending coal or requiring large-scale new nuclear buildouts in most regions. The system does not require a rescue from legacy baseload. It requires continued compounding of what is already winning on cost.
Nuclear as Hedge, Not Growth Engine
Nuclear remains valuable as a hedge technology. It provides resilience in specific geographies with limited renewable resources, dense industrial clusters requiring high reliability, or constrained land availability. It may also play a strategic role in defence infrastructure, remote grids, and future space exploration applications. In those contexts, nuclear offers firm, weather-independent output that complements — rather than competes with — VRE.
However, nuclear is not the growth engine of the 2020s. Deployment speed, capital intensity and cost trajectories matter. Solar, wind and storage are scaling at orders of magnitude faster rates. Capital is flowing accordingly. In a system governed by cost curves, the technologies that scale fastest and cheapest determine the centre of gravity. Nuclear can remain an important hedge within the system, but it is unlikely to define its dominant capacity trajectory this decade.
Utilisation Collapse and Economic Redundancy
Installed capacity flipping is only the first phase. The deeper shift comes when fossil utilisation rates begin to fall. As VRE penetration rises, coal and gas plants run fewer hours. Coal, typically higher-emitting and less flexible, faces sharper utilisation pressure in high-penetration markets, while gas increasingly shifts toward peaking and balancing roles. Load factors decline. Margins compress.
Infrastructure does not disappear overnight. It becomes economically redundant before it is dismantled. That redundancy is the quiet mechanism of transition.
The Acceleration Toward Energy Abundance
As cheap VRE scales and storage absorbs volatility, the marginal cost of electricity trends downward in surplus periods. This dynamic begins to unlock a new phase: energy overcapacity rather than scarcity. Lower-cost electricity enables desalination, electrofuels, green hydrogen, precision fermentation and massive computational workloads.
This is not simply decarbonisation. It is structural abundance emerging from compounding cost curves and firmed renewables at scale.
Emissions and the S-Curve Effect
Many climate projections assume gradual policy-driven adoption. Technology disruption rarely follows gradual linear paths. S-curves compress timelines. When capacity growth accelerates, utilisation shifts faster than expected. When utilisation shifts, emissions fall faster than projected under linear assumptions.
In the model, fossil capacity begins declining before 2030, but the more important dynamic is utilisation compression. Even without immediate plant retirements, falling load factors materially reduce CO₂ and methane intensity of the power sector. Under continued VRE and storage compounding, emissions decline accelerates relative to static-capacity assumptions commonly embedded in long-range outlooks.
This is not an argument for complacency. It is an argument that technology-driven displacement can move faster than many legacy projections anticipate.
It is also a global argument with regional variation. Crossovers will not occur simultaneously everywhere. China, Europe and parts of the United States may reach capacity and generation inflection points earlier due to scale and deployment velocity. Emerging markets with lower electrification levels may experience a different sequencing. The structural direction, however, is consistent: as cost curves fall and capital shifts, each region moves along the same layered pathway — capacity, firming, generation — at its own pace.
Net Zero as Systems Outcome
Net Zero is often framed as a policy target. In reality, it is a systems consequence of power-sector transformation.
Electricity is the keystone sector. Once power generation decarbonises at scale, electrification of transport, heat and industry compounds the effect. Every additional electric vehicle, heat pump or electric furnace plugged into a rapidly decarbonising grid accelerates economy-wide emissions decline.
The sequencing matters. First, clean capacity overtakes fossil capacity. Then firming ensures reliability without emissions. Then generation dominance pushes fossil utilisation toward structural irrelevance. At that point, Net Zero in the power sector is no longer aspirational — it becomes a default trajectory driven by economics.
Economy-wide Net Zero follows from three reinforcing loops:
- Clean power lowers marginal electricity cost.
- Lower electricity cost accelerates electrification.
- Electrification reduces direct combustion emissions and increases demand for clean power, reinforcing scale.
Under sustained VRE and storage compounding, the power sector can approach near-zero operational emissions well before mid-century in leading regions. As grids decarbonise, hard-to-abate sectors shrink in relative importance and become the focus of targeted solutions rather than systemic drag.
Net Zero, in this framing, is not a single political decision. It is the emergent result of capacity tipping, firming scaling and generation dominance converging within a cost-driven system.
System Constraints and Boundary Conditions
No model operates in a vacuum. The projections assume continued resolution of three non-economic constraints: transmission, materials, and market design.
Transmission. Generation and storage only create value if the grid can move electrons. Interconnection queues are expanding — in the United States alone, well over 1.5 TW of generation and storage projects sit in interconnection queues — and high-voltage transmission build times in developed markets often exceed a decade. Storage can buffer local congestion, but it cannot substitute for wires. In the near term, congestion — not curtailment — may become the binding constraint in certain regions. The crossover timing therefore assumes steady, if imperfect, grid expansion alongside generation growth.
Materials and supply chains. Solar, wind and storage scaling relies on lithium, copper, silver and rare earth processing capacity. These supply chains are geographically concentrated — lithium processing is heavily concentrated in China, and solar PV already consumes roughly 10–15% of annual global silver supply — while mining lead times are long. The model assumes continued incremental resolution of these bottlenecks through substitution, recycling, chemistry evolution and capacity expansion. Severe geopolitical disruption would slow build rates, though not alter the long-term cost-curve direction.
Demand response and flexible load. The projections treat demand growth explicitly, but demand itself will become increasingly price-responsive. Electrolysers, EV fleets, data centres and industrial loads can shift consumption toward surplus periods. As flexible demand scales, required storage duration may moderate at the margin, because load shifting performs part of the firming function — potentially reducing the average duration requirement by on the order of 30–60 minutes under high-flex scenarios relative to a 4-hour baseline. Curtailment therefore signals both storage opportunity and flexible electrification opportunity.
By the early 2030s, leading regions are likely to achieve routine 24-hour renewable coverage using combinations of solar, wind and storage. South Australia already demonstrates high VRE penetration with growing battery depth. Texas and California are rapidly scaling storage to manage evening ramps. Parts of China and Europe are deploying grid-scale storage alongside aggressive renewable build-out. In parts of Africa, leapfrogging directly to distributed solar-plus-storage systems may enable localised 24-hour reliability without legacy fossil baseload. These regions act as proving grounds for the longer-duration storage trajectory embedded in the model.
Gas as transitional hedge. Within the shrinking fossil fleet, gas is likely to gain relative share as coal retires faster. Gas plants are cheaper and faster to build than nuclear and increasingly operate as flexibility assets rather than baseload. This shift reduces emissions intensity relative to coal but introduces methane and infrastructure lock-in considerations. The model captures declining aggregate fossil capacity, but composition within that category may evolve.
Scope boundary. This analysis focuses on the power sector. Electrification of transport, heat and industry is embedded in the demand growth trajectories but not explicitly modelled sector by sector. Faster electrification accelerates crossover timing pressure; slower electrification moderates it. The structural sequencing — capacity, firming, generation — remains intact under reasonable demand bands.
The Turbulent Middle
The 2025–2035 window will not feel smooth. Expect stranded assets, political backlash, incumbent resistance and capital reallocation shocks. Legacy systems rarely concede quietly. Capacity mechanisms, regulatory friction and market redesign can slow utilisation decline even when economics turn.
But delay is not reversal. Once capacity tips, firming scales and generation dominance follows, the direction of travel becomes self-reinforcing.
2040–2050: The System After the Flip
If the layered transition described above holds, the 2040–2050 period looks structurally post-fossil in leading economies. By around 2040, fossil generation is likely to have exited the core of the power system in most advanced grids, retained only in residual or emergency roles where economically justified. Nuclear persists on the margins in specific geographies — as resilience hedge, strategic infrastructure, or legacy fleet continuation — but no longer defines system direction.
Electric vehicles dominate passenger transport and increasingly heavy-duty segments as battery cost curves continue downward. Electrification becomes the default pathway for mobility, heat and much of industry. The power sector, having crossed capacity, firming and generation thresholds in the 2020s and 2030s, becomes the foundation layer of a fully electrified economy.
Storage durations extend beyond intra-day balancing toward multi-day resilience, complemented by flexible demand that responds dynamically to price signals. In that environment, electricity ceases to be a constraint on economic activity. It becomes an enabling platform for computation, manufacturing, desalination and synthetic fuels.
This is the early architecture of a Type 1 planetary civilisation — not in a speculative sense, but in the practical sense of harnessing abundant, largely carbon-free energy flows at continental scale. The transition decade is about dislocation. The 2040s are about consolidation and optimisation.
We are well into that alignment phase. The question is no longer whether the system flips, but how quickly the old regime accepts its redundancy in the age of Bettrification.
Investor Implications
For investors and capital allocators, the implications are structural rather than thematic. When capacity growth decisively favours one technology stack, capital efficiency follows. Technologies compounding fastest attract supply-chain investment, grid integration spending and policy support.
Three signals matter most:
- Storage pairing becomes standard in high-curtailment markets.
- Fossil utilisation rates decline before formal retirements accelerate.
- Power price volatility increasingly rewards flexible assets over static baseload.
This is not simply a renewables story. It is a capital reallocation story. Money is already moving toward technologies riding compounding cost curves — and away from assets whose utilisation is structurally declining. The system math is no longer abstract. It’s visible in build rates, margins and cash flows — and markets are pricing it in now.
Sources & Data References
All underlying datasets, source documents and reference material used in this model can be found here.
The projections in this piece synthesise IEA, IRENA, Ember, BNEF, Rho Motion, BMI/Fitch Solutions and related industry data, extended through consistent scenario methodology as described above.