Convergence Through Crisis: Disruption opens replacement windows — what decides which technology fills them

Hunter Hughes

Abstract

Emerging economies are often urged to leapfrog: to skip the fossil-intensive stage of development the way many skipped fixed-line telephony. The advice is usually framed as a choice, and it is not. Incumbent capital persists because it is sunk, because switching is optional, and because the option to defer has value. What dislodges it is a crisis that makes the incumbent unavailable or unaffordable — at which point an optional upgrade becomes a forced replacement, and the replacement is drawn from whatever is available, financeable and permitted in that particular month. The decision window is weeks; the asset lasts decades.

The evidence for this pattern is now unusually clear. Pakistan imported about 17 gigawatts of solar panels in 2024, double the previous year, becoming one of the world's largest markets in the space of a year — a response to electricity prices that had roughly doubled since 2021 rather than to any national programme. South Africa's rooftop solar capacity rose from 983 megawatts in March 2022 to about 7.3 gigawatts by 2025, through the worst years of load-shedding. Lebanon's installed solar rose thirteenfold, from roughly 100 megawatts in 2020 to about 1,300 megawatts in 2023, after the collapse of the state supply. Vietnam installed 9.7 gigawatts of rooftop solar in 2020, of which about 6 gigawatts arrived in the final month before a tariff deadline. Nigeria, after removing its petrol subsidy, went from roughly 11,000 converted compressed-gas vehicles to more than 100,000 in about a year.

The same windows also close the wrong way, and the report gives those cases equal weight. During its load-shedding peak Eskom spent 21.5 billion rand on diesel for open-cycle turbines in the 2023 financial year, rising to 23.4 billion the year after — a category that accounted for roughly 19 per cent of generation cost and 2 per cent of output. Ecuador, facing drought-driven blackouts, leased 300 megawatts of Turkish power barges for about US$250 million. Lebanon's private diesel generators survived the solar boom because most households could not also afford storage. In each case the crisis forced a replacement decision and the incumbent technology won it, because it was the option that could be delivered that month.

From these cases the report derives a three-gate test applied at the moment of forced replacement — is the clean option available, is it financeable, and is it permitted — and argues that diesel and rented thermal capacity win by default on all three. The implications are about preparation rather than persuasion: standing import and duty treatment, pre-cleared grid codes and net-metering rules, credit available at the point of purchase rather than through a programme cycle, pre-qualified installers, and procurement templates for storage and hybrid capacity that can be executed in weeks. Countries that pre-position those five things convert their next crisis into a re-equipment at a lower cost and emissions frontier; countries that do not will rent fossil capacity at emergency prices and add the bill to their debt.

Executive Summary

Nobody replaces working equipment because a study recommends it. They replace it when it stops working, stops arriving, or stops being affordable — and what they buy that month is decided by what is on the shelf.

FINDING 01

Crises force the replacement decision

Sunk capital persists while switching is optional. A price shock, a blackout or a collapse converts an optional upgrade into a forced purchase, and the whole national stock turns over faster in two years than in the preceding twenty.

FINDING 02

The window is weeks; the asset is decades

Whatever can be delivered, paid for and legally connected in the weeks after the shock becomes the capital stock for the next twenty years. Persuasion arrives too late to matter; only pre-positioning does.

FINDING 03

Diesel wins the default

A generator is in stock, cheap to buy, and needs no permission. A rented power barge can be moored in months. Unless the clean option clears the same three gates in the same weeks, the crisis entrenches the incumbent.

The leapfrog literature treats technological skipping as a choice a country makes. The record suggests something narrower and more useful: leapfrogs happen when the incumbent becomes unavailable, and they take whatever shape the market can supply at that moment. Pakistan did not decide to become one of the world's largest solar markets; its electricity prices roughly doubled and households left the grid. South Africa did not plan seven gigawatts of rooftop solar; it had 335 days of load-shedding in 2023.

This report examines five windows that produced rapid clean replacement and three that produced the opposite, and derives from them a test that can be applied before the next crisis rather than after it.

17 GW

Solar panels imported by Pakistan in 2024 — double the previous year

Ember, Global Electricity Review 2025

7.3 GW

South African rooftop solar, up from 983 MW in March 2022

Eskom and National Treasury estimates

13×

Increase in Lebanon's installed solar between 2020 and 2023, after the state supply collapsed

Contemporary reporting

R21.5bn

Eskom's FY2023 diesel bill: about 19% of generation cost for 2% of output

Eskom disclosures

Section 1 explains why capital stock is sticky until it is forced. Section 2 sets out five windows that produced clean replacement, and Section 3 three that did not. Section 4 gives the three-gate test that distinguishes them, Section 5 the pre-positioning it implies, and Section 6 the consequence for convergence. Sections 7 and 8 give recommendations and conclusions.

A country does not decide to leapfrog. It decides, years earlier, what will be in stock, financeable and legal on the morning the lights go out. The preparation argument

1. Why Nothing Changes Until It Has To

The persistence of old capital is rational, which is why exhortation does so little and disruption does so much.

A working diesel generator, a connected grid supply, a petrol engine: each represents money already spent. The relevant comparison for its owner is not the lifetime cost of a new clean alternative against the lifetime cost of the incumbent, but the cost of the new asset against the marginal running cost of the one already owned. On that comparison the incumbent usually wins, and the owner rationally defers — retaining the option to switch later, when prices may be lower and the technology better understood.

Three things break that logic, and all of them are forms of disruption:

The incumbent becomes unaffordable. A tariff increase, a subsidy removal, a currency devaluation that reprices imported fuel. The marginal running cost jumps, and the comparison inverts.
The incumbent becomes unavailable. Load-shedding, fuel queues, a collapsed utility. What was a cost comparison becomes a question of whether the service exists at all.
The incumbent is destroyed. A flood or storm removes the asset, and the replacement decision is taken at a moment when nothing is sunk any more.

In each case the option to defer disappears and the purchase happens now. That is the mechanism this report is about, and its most important property is speed: the deliberation that normally takes a decade compresses into weeks, and the choice set narrows to what can actually be obtained.

2. Five Windows That Produced a Leapfrog

Different shocks, different countries, the same shape: a forced replacement met by a technology that happened to be cheap, available and installable without permission.

2.1 Pakistan: price

Pakistani electricity prices roughly doubled between 2021 and 2024, driven in part by capacity payments on thermal contracts. Households and businesses responded by leaving the grid where they could. Pakistan imported about 17 gigawatts of solar panels in 2024 — twice the previous year's volume — making it one of the largest markets in the world for new solar within a single year, almost entirely through unsubsidised private purchases (Ember). No national programme organised this. Cheap Chinese modules met a population being priced off the grid, and the two found each other.

Figure 1 — Pakistan's solar imports double in a year

Gigawatts of solar panels imported. The 2024 figure is Ember's; the 2023 bar is shown as approximately half of it, which is what "double the amount imported the year before" implies, and is labelled as inferred rather than reported.

2.2 South Africa: availability

South Africa endured 205 days of load-shedding in 2022 and 335 days in 2023. Rooftop solar capacity, which stood at 983 megawatts in March 2022, reached about 4,412 megawatts by June 2023, roughly 5,800 megawatts by mid-2024 and about 7.3 gigawatts on the most recent estimates — installed by households and firms, outside the utility's procurement entirely.

Figure 2 — South African rooftop solar through the load-shedding years

Estimated installed rooftop photovoltaic capacity, megawatts. From Eskom and National Treasury estimates as reported between 2023 and 2025. These are modelled estimates rather than a metered registry, and different vintages differ by several hundred megawatts.

2.3 Lebanon: collapse

After 2019 the Lebanese state's electricity supply effectively failed, with Électricité du Liban providing only a few hundred megawatts against demand several times larger. Solar installation went from about 14 megawatts added in 2020 to 663 megawatts added in 2022 alone, with cumulative capacity rising roughly thirteenfold from about 100 megawatts in 2020 to some 1,300 megawatts by 2023. This was not an energy transition in any planned sense; it was households buying the only electricity they could get.

Figure 3 — Lebanon's solar capacity after the collapse

Cumulative installed solar capacity, megawatts, as reported in contemporary assessments. Estimates vary between sources; the magnitude and the timing are consistent across them.

2.4 Vietnam: a deadline as an artificial crisis

Vietnam's second rooftop feed-in tariff paid a fixed price for systems commissioned by 31 December 2020. Installations rose from 378 megawatts-peak in 2019 to 9,731 in 2020, an increase of about 2,474 per cent — and roughly 6 gigawatts of that arrived in the final month, with more than 3,000 megawatts commissioned in the last three days.

Vietnam is the control case for this report's argument. There was no shock, no shortage and no price spike: only a deadline that made deferral impossible. The response was the same as a crisis response, which suggests the operative variable is the removal of the option to wait rather than hardship as such. A policy can manufacture that; it does not have to be suffered.

Figure 4 — Vietnam's rooftop solar, 2019 and 2020

Megawatts-peak of rooftop solar installed in each year, with the portion of 2020's total commissioned in December shown separately. From industry reporting on Vietnam's FIT2 deadline.

2.5 Nigeria: subsidy removal, and a partial substitution

Nigeria removed its petrol subsidy in 2023, and pump prices rose sharply. The government's compressed natural gas initiative reports that converted vehicles rose from roughly 11,000 in 2023 to more than 100,000 within about a year, conversion centres from seven to over 170, and private investment mobilised of about US$1.02 billion.

This case is included precisely because it is not a clean leapfrog. Compressed natural gas is a fossil fuel; the substitution cuts cost and local pollution but not carbon to anything like the degree electrification would. It illustrates the report's central claim in its least flattering form: the window was real, the replacement was rapid, and what filled it was whatever the country could actually supply at scale in a year.

3. When the Window Closes the Wrong Way

The same forcing mechanism that produced Pakistan's solar boom produced South Africa's diesel bill and Ecuador's rented barges. Nothing about a crisis favours clean technology.

3.1 Eskom's diesel

While South African households were installing rooftop solar, the utility was buying diesel. Spending on open-cycle gas turbines rose from about 10.1 billion rand in the 2022 financial year to 21.5 billion in 2023 and 23.4 billion in 2023/24. In the 2023 financial year those turbines accounted for roughly 19 per cent of Eskom's generation cost and 2 per cent of its output.

Figure 5 — Eskom's diesel spending on emergency turbines

Billions of rand spent on diesel for open-cycle gas turbines by financial year, as reported at the time. In FY2023 this category accounted for about 19 per cent of generation cost and 2 per cent of output.

The comparison is the point. Faced with the same crisis, the household chose a twenty-year asset and the utility chose a fuel bill, because the household could install in a week and the utility could not build in a week. Speed of delivery, not merit, decided both.

3.2 Ecuador's barges

When drought cut Ecuador's hydropower in 2024, the government leased thermal generation barges: between September 2024 and March 2025, about 300 megawatts from three vessels at a reported cost of some US$250 million. Rented thermal capacity is the purest expression of a badly-prepared window — expensive, fossil, temporary, and leaving no asset behind at the end of the lease.

3.3 Lebanon's surviving generators

Even in the clearest leapfrog case, the incumbent survived. Lebanese private diesel generators continued operating alongside the solar boom, because panels without storage do not cover the evening, and most households could not afford both at once. The lesson is that a partial substitution leaves the incumbent in place with a smaller but durable role — and the gap that keeps it there is usually storage.

The asymmetry that decides it

A diesel generator is in stock at a dealership, costs little up front, and needs no permission to run. A rented barge can be moored in months on a single contract. Solar with storage is cheaper over its life and slower to obtain: it needs modules in the country, credit at the point of purchase, an installer, and — where it touches the grid — a connection rule that already exists. The incumbent wins the window by default, and only deliberate preparation changes that.

4. The Window Test

Three questions, asked about the moment of forced replacement rather than about the technology in general.

Whether a crisis produces a leapfrog or a lock-in turns on whether the clean option clears three gates in the weeks the decision is actually taken:

Available. Is the hardware physically in the country, or obtainable within the window, with someone able to install it? Lead time is the binding constraint, not price.
Financeable. Can the buyer pay at the moment of purchase? A technology with low lifetime cost and high up-front cost fails this gate precisely when incomes are under stress — which is to say, exactly during the crisis.
Permitted. Is it legal to install and connect without a process longer than the window? Net-metering rules, import treatment, grid codes and certification all act as timing constraints, whatever their merits.

Table 1 — The episodes against the three gates

Episode	Shock	Available	Financeable	Permitted	Outcome
Pakistan, 2024	Tariffs roughly doubled	Yes — cheap modules, open imports	Partly — cash purchase, mostly better-off buyers	Yes — off-grid needs no approval	17 GW imported in a year
South Africa, 2022–24	335 days of load-shedding	Yes — mature installer base	Partly — bank finance and a tax rebate for some	Yes — registration eased during the crisis	~1 GW to 7.3 GW of rooftop
Lebanon, 2020–23	State supply collapsed	Yes — imports and informal installers	No — cash only, no banking system	Yes — nothing to connect to	13× growth, but generators survive
Vietnam, 2020	Tariff deadline	Yes — regional supply chain	Yes — guaranteed 20-year tariff	Yes — the policy was the permission	9.7 GW in a year, 6 GW in a month
Eskom, 2022–24	Fleet failure	No — new plant takes years	Partly — balance sheet distressed	Partly — procurement rules slow	R21.5bn of diesel in one year
Ecuador, 2024	Drought cut hydropower	No — no storage or solar pipeline ready	No — no capital programme in place	Partly — emergency powers used for leases	300 MW of rented barges, US$250m

The gate assessments are the author's, applied to the episodes documented in Sections 2 and 3. They describe the position at the moment of the shock, not the country's general policy stance.

Figure 6 — What happens when the window opens

A conceptual schematic of Table 1, not a quantitative model. The gates are sequential in practice: hardware that cannot be financed is not available to the buyer who needs it, and hardware that cannot be connected is not worth financing.

5. Pre-positioning: What to Do Before the Crisis

Every item on this list has to exist before the shock. None of them can be created inside the window.

Standing import and duty treatment. Tariff exemptions and customs classifications for modules, inverters, batteries and efficient appliances should be permanent and pre-published, not announced as emergency relief three months into a crisis. Pakistan's boom ran on open imports meeting collapsed module prices; a duty regime settled in advance is what makes that possible elsewhere.

Pre-cleared connection rules. Net metering, small-scale embedded generation registration, safety certification and grid codes should be written, published and tested in calm conditions. A rule that takes four months to obtain is a prohibition in a window that lasts eight weeks.

Credit at the point of purchase. The gate most often failed is finance, because up-front cost collides with a shock that has just reduced incomes. Standing instruments — on-bill repayment, asset finance through utilities or telecoms, pre-agreed credit lines to installers, partial guarantees — must be live before the crisis, since a new programme takes longer to design than the window lasts.

An installer base and a certification registry. Hardware without installers is not available in any practical sense. South Africa's speed rested on an existing commercial installer industry; Lebanon's rested on an informal one. Training and certification are slow to build and cannot be improvised.

Procurement templates for firm capacity. Utilities lose these windows because they cannot build in weeks. The counter to a rented barge is a pre-negotiated framework for battery storage and hybrid capacity — standardised contracts, pre-qualified suppliers, pre-cleared sites — so that an emergency can be answered with an asset instead of a lease.

Manufacturing the window

Vietnam's 2020 surge came from a deadline rather than a disaster, which means the forcing function can be created deliberately. A tariff that steps down on a published date, an efficiency standard that takes effect, a subsidy that phases out on a schedule: each removes the option to wait without requiring anyone to suffer a blackout first. The difference between a manufactured window and a crisis window is that the state chooses the timing — and can therefore ensure the three gates are open when it arrives.

6. Why This Is a Convergence Question

Crises are not rare events in emerging economies. They are the mechanism through which the capital stock actually turns over.

A country that catches its windows re-equips at each one, arriving at the frontier in a series of jumps rather than a smooth climb — the pattern that made mobile telephony look like a choice when it was mostly an accident of timing. A country that misses them pays three times: at emergency prices during the shock, in fuel bills for the rented capacity afterwards, and in the debt service on both.

The second pattern is corrosive in a specific way that connects to earlier reports in this series. Emergency fossil procurement is expensive precisely when fiscal space is smallest, and it adds to the debt stock that already carries a vulnerability premium. Ecuador's US$250 million of barge leases bought no asset; Eskom's diesel bought 2 per cent of output for 19 per cent of cost. Both are transfers from the investment budget to the fuel bill, made under time pressure, and both make the next window harder to catch.

The optimistic reading is that the windows keep coming. Every price shock, every drought-driven blackout, every subsidy reform is another forced replacement — another chance to install the better asset if it happens to be available, financeable and permitted that month. Which of those is true on the day is decided years earlier, by whether anyone prepared.

7. Recommendations

Grouped by who controls each gate.

7.1 Energy ministries and regulators

Publish permanent connection and certification rules for small-scale generation and storage, and test the process in advance so that its actual duration is known.
Settle import and duty treatment for clean hardware permanently, rather than granting exemptions during emergencies when the shipping decision has already been made.
Hold pre-negotiated frameworks for storage and hybrid capacity with pre-qualified suppliers and cleared sites, so an emergency can be answered with an asset rather than a lease.
Consider manufacturing a window — a stepped tariff, a standard with a date — where the capital stock needs turning over and no crisis is obliging.

7.2 Finance ministries and development finance

Keep consumer and SME asset finance live before the crisis. On-bill repayment, guarantee facilities and installer credit lines are the difference between a leapfrog available to everyone and one available to those with cash.
Budget for the default. If no pre-positioned alternative exists, the emergency line will be spent on fuel or leases; putting a number on that in advance is the most persuasive case for the standing alternative.
Treat crisis response as capital formation. Where emergency capacity is unavoidable, structure it so the country owns an asset at the end.

7.3 Utilities

Assume the customers will leave. When reliability fails, the best-resourced customers self-supply first, taking their load and their revenue; planning should treat that as the base case, not a risk.
Sell what the customer is about to buy anyway — rooftop packages, storage, financed hybrid supply — rather than losing the relationship entirely.

7.4 Development partners and analysts

Measure lead times, not just costs. The binding constraint in a window is how fast something can be obtained, installed and connected, and it is almost never reported.
Document the failed windows as carefully as the successful ones. Rented barges and emergency diesel are the counterfactual the success stories are measured against.

8. Conclusions

The question is not whether emerging economies can leapfrog. It is what will be on the shelf the next time they are forced to.

Capital stock is sticky because deferral is rational, and it turns over when deferral becomes impossible. That is why the most dramatic clean-energy expansions of the last five years happened in countries that were not running clean-energy programmes: Pakistan importing 17 gigawatts of panels as tariffs doubled, South Africa's rooftops going from under one gigawatt to more than seven through 335 days of load-shedding, Lebanon's capacity rising thirteenfold after the state supply failed. Vietnam shows the same behaviour produced by a deadline rather than a disaster, which is the most encouraging fact in this report: the forcing function can be designed.

But the mechanism is indifferent to what it installs. The same load-shedding that put solar on South African roofs put 21.5 billion rand of diesel through Eskom's turbines in a single year, for 2 per cent of output. The same drought that might have justified storage in Ecuador produced 300 megawatts of rented barges at a quarter of a billion dollars. Lebanon's diesel generators are still running because panels without batteries do not cover the evening. Crises force a decision; they do not improve it.

What decides the outcome is whether the better option clears three gates during the weeks the decision is live: available, financeable, permitted. Those gates are set by import regimes, connection rules, certification registries, installer capacity and standing credit — all of which take years to build and none of which can be created inside the window. The work of catching the next crisis is therefore ordinary and unglamorous, and it has to be done now, in the calm, for a shock whose date nobody knows.

References

Every quantitative claim above is attributed inline. The principal sources are collected here.

EmberGlobal Electricity Review 2025 — Pakistan's 17 GW of solar imports in 2024, double the previous year. Context on tariffs and the bottom-up nature of the boom in CNN and Business Recorder.
Eskom and National Treasury estimatesSouth African rooftop solar at 983 MW in March 2022 and 4,412 MW by June 2023 (Energy Monitor), about 5,791 MW by mid-2024 and roughly 7,345 MW on later estimates (Moneyweb).
Reporting on Lebanonpv magazine on 663 MW added in 2022 against 14 MW in 2020; MERIP and The Century Foundation on the thirteenfold rise to about 1,300 MW and on why diesel generators survived.
PV TechVietnam rooftop solar records major boom as more than 9GW installed in 2020 — 378 MWp in 2019 to 9,731 MWp in 2020, with about 6 GW commissioned in December alone.
Presidential Initiative on Compressed Natural Gas, NigeriaPi-CNG, with conversion and investment figures as reported by Nigerian Tribune — roughly 11,000 converted vehicles in 2023 to more than 100,000 within a year, and about US$1.02 billion of private investment.
Eskom diesel spendingMoneyweb and Bloomberg on R21.4–21.5bn in FY2023 against R10.1bn the year before, and the share of cost against output.
Reporting on Ecuador's power bargesDialogue Earth and regional coverage — three leased vessels totalling about 300 MW at a reported cost of some US$250 million between September 2024 and March 2025.
H HeuristicsRelated reports in this series: financing the synergy, on the vulnerability premium that emergency borrowing compounds; scaling low-cost technologies, on why some goods diffuse and others do not; and food, energy, water, on the drought that produced Ecuador's crisis.

Metadata

Keywords: leapfroggingtechnology adoptionenergy crisisrooftop solarPakistan solarload sheddingdistributed generationcapital stock turnoverdiesel lock-inpower bargesenergy policyeconomic convergencecrisis responsepreparedness
Topics: Energy Transition Emerging Markets Technology Infrastructure
JEL classification: O33, Q42, O13, L94, Q48 — technological change and diffusion; alternative energy sources; agriculture and natural resources in development; electric utilities; energy policy
Data and method: This report synthesises published data and contemporary reporting on six national episodes rather than producing new modelling. Pakistan's 2024 solar imports are from Ember's Global Electricity Review 2025 as reported at the time; South Africa's rooftop capacity from Eskom and National Treasury estimates reported between 2023 and 2025; Lebanon's installed solar from pv magazine and subsequent assessments; Vietnam's 2020 rooftop additions from PV Tech and industry reporting; Nigeria's vehicle conversions from the Presidential Initiative on Compressed Natural Gas and Nigerian press reporting; Eskom's diesel spending from Bloomberg, Moneyweb and Eskom disclosures; and Ecuador's power barge leases from regional reporting. All figures were verified against their sources in September 2026. Rooftop and distributed capacity estimates in several of these countries are inferred from utility modelling and import data rather than metered registries, and are therefore approximations; the report uses them to establish magnitude and direction, not precision. The three-gate window test in Section 4, Table 1 and the pre-positioning list in Section 5 are the author's analysis, assembled with the cited episodes already known, and are offered as a preparedness checklist rather than a predictive model. Figure 6 is a conceptual schematic. The report is analytical rather than predictive.
Report: H Heuristics Digital Report № 2026-16 · Published 17 September 2026
Licence: CC BY-NC-ND 4.0
Cite as: Hunter Hughes (2026). Convergence Through Crisis: Disruption opens replacement windows — what decides which technology fills them. H Heuristics Digital Report 2026-16. https://digitalreports.hheuristics.com/reports/convergence-through-crisis/

↑ Back to top

[1] EmberGlobal Electricity Review 2025 — Pakistan's 17 GW of solar imports in 2024, double the previous year. Context on tariffs and the bottom-up nature of the boom in CNN and Business Recorder.

[2] Eskom and National Treasury estimatesSouth African rooftop solar at 983 MW in March 2022 and 4,412 MW by June 2023 (Energy Monitor), about 5,791 MW by mid-2024 and roughly 7,345 MW on later estimates (Moneyweb).

[3] Reporting on Lebanonpv magazine on 663 MW added in 2022 against 14 MW in 2020; MERIP and The Century Foundation on the thirteenfold rise to about 1,300 MW and on why diesel generators survived.

[4] PV TechVietnam rooftop solar records major boom as more than 9GW installed in 2020 — 378 MWp in 2019 to 9,731 MWp in 2020, with about 6 GW commissioned in December alone.

[5] Presidential Initiative on Compressed Natural Gas, NigeriaPi-CNG, with conversion and investment figures as reported by Nigerian Tribune — roughly 11,000 converted vehicles in 2023 to more than 100,000 within a year, and about US$1.02 billion of private investment.

[6] Eskom diesel spendingMoneyweb and Bloomberg on R21.4–21.5bn in FY2023 against R10.1bn the year before, and the share of cost against output.

[7] Reporting on Ecuador's power bargesDialogue Earth and regional coverage — three leased vessels totalling about 300 MW at a reported cost of some US$250 million between September 2024 and March 2025.

[8] H HeuristicsRelated reports in this series: financing the synergy, on the vulnerability premium that emergency borrowing compounds; scaling low-cost technologies, on why some goods diffuse and others do not; and food, energy, water, on the drought that produced Ecuador's crisis.