1 Introduction

Rapid renewable capacity additions are widely interpreted as evidence of accelerating power-sector transition (IPCC 2022). Yet additions do not necessarily imply the physical displacement of fossil-fuel infrastructure. When renewable build-out occurs without concurrent contraction of fossil operating stock, the system may appear to be decarbonizing while remaining structurally dependent on carbon-intensive assets.

This distinction matters because long-lived capital stock and the institutions built around it can slow structural change, reinforcing path dependence and carbon lock-in (Seto et al. 2016; Bertram et al. 2015). The climate significance of such stock persistence is underscored by the literature on committed emissions: existing and planned energy infrastructure can embed large future emissions trajectories if operated along historical patterns (Steven et al. 2010; Tong et al. 2019; Pfeiffer et al. 2018). In this sense, visible renewable growth does not by itself establish that a power system is undergoing climate-relevant fossil replacement.

Recent work further suggests that power-sector transition cannot be reduced to a single dimension of progress. Generation shares, renewable capacity additions, and project-pipeline composition each capture a different part of the transition process, but they are analytically non-equivalent. Generation shares describe delivered electricity; additions describe realized build-out and investment momentum; and project-pipeline composition refers here to the observable development pipeline of announced, pre-construction, and under-construction power assets. This provides a forward-looking signal of development orientation. Each is valuable, but none directly identifies whether fossil assets are being physically replaced at the unit level. Moreover, the project pipeline is not equivalent to realized commissioning, so forward-looking signals do not guarantee physical build-out (Alova et al. 2021). Conventional indicators therefore do not necessarily align realized operating-stock change with forward-looking development intent (Kennedy et al. 2023; Kern and Rogge 2016; Defeuilley 2019; Harsh et al. 2019; Johnstone et al. 2020).

A further challenge is that this conceptual fragmentation is compounded by a systemic informational asymmetry in global asset tracking. Additions and project pipelines are often recorded with relatively high granularity, whereas retirement timing and decommissioning processes remain unevenly documented across regions and technologies. As a result, asset entry is often visible while asset exit remains only partially observed, complicating how fossil exit is monitored and interpreted (Mills et al. 2017; Malik et al. 2020; He et al. 2020; Maamoun et al. 2020; Edianto et al. 2023). From an energy–environment perspective, this is not merely a data limitation: it is a phase-out monitoring challenge, because incomplete visibility of asset exit creates uncertainty in how fossil persistence and retirement progress are assessed.

Taken together, these conditions create a critical blind spot in transition monitoring: existing readings frequently conflate renewable additions, pipeline optimism, and realized fossil displacement. This conflation makes it difficult to distinguish true asset reallocation from additive transition, in which systems grow greener but also heavier in total installed capacity. Related scholarship on utility adaptation, supply-side governance, stranded assets, and phase-out policy similarly suggests that renewable deployment and fossil exit are linked yet distinct processes, mediated by institutions, ownership structures, infrastructure lock-in, and explicit policy design rather than by deployment momentum alone (Alova 2020; Piggot et al. 2020; Gürsan and de Gooyert 2021; Rinscheid et al. 2021; van Asselt 2021b; Cherp et al. 2021; Trencher et al. 2023; Devine-Wright and Peacock 2024).

To address this gap, we use harmonized global plant records (Global Energy Monitor 2024b, a; Byers et al. 2018) to construct an auditable asset-level stock–pipeline diagnostic of the global power transition from 2010 to 2026. This study focuses on electricity-generation assets in the power sector and does not cover heat systems, liquid fuels, solid fuels, or end-use energy consumption. Our contribution is threefold. First, we build a unit-level global operating-stock panel to track physical asset entry and exit. Second, we distinguish otherwise conflated transition regimes by separating realized stock reallocation from forward-looking pipeline orientation. Third, we provide a monitoring-oriented diagnostic of fossil persistence under visible renewable growth, thereby highlighting the importance of retirement transparency, stranded-asset governance, and deliberate policy-managed phase-out (Jenkins 2014; Rozenberg et al. 2020; Angelika von Dulong et al. 2023; Robert Fofrich Navarro et al. 2026; Erickson et al. 2018; Lazarus and van Asselt 2018; van Asselt 2021a; Mohlin et al. 2019; Oshiro and Fujimori 2021; Peng et al. 2023).

2 Methods

2.1 Data and operating-stock reconstruction

We use unit-/phase-level records from the Global Integrated Power Tracker (GIPT) (Global Energy Monitor 2024b, a). This study focuses on electricity-generation assets within the power sector and does not cover heat systems, liquid fuels, solid fuels, or end-use energy consumption. Following a deterministic audit pipeline, we restrict the analysis universe to in-scope statuses and valid coordinates, deduplicate records, and clean commissioning and retirement years (SI.1). Our primary unit of inference is the macro-subregion (17 subregions within 5 global regions), chosen to preserve global comparability under heterogeneous unit-level data completeness; an illustrative national-scale replication for China and the United States is reported in SI.8.

For each year t, a unit is classified as operating if \(\textrm{StartYear}\le t\) and \(\textrm{RetiredYear}\) is missing or \(\textrm{RetiredYear}>t\). To prioritize physical consistency, we enforce a retirement-year precedence rule: if \(\textrm{RetiredYear}_{clean}\le t\), the unit is excluded from the operating stock even if its categorical status remains labeled as operating. We address missing commissioning years through a main specification and two sensitivity bounds (SI.2), and treat retirement timing as only partially observed because retirement-year records remain incomplete and heterogeneous across regions and fossil types.

2.2 Core stock–pipeline monitoring framework

Our stock–pipeline framework is designed to monitor fossil persistence under visible renewable growth by separating realized operating-stock change from forward-looking development orientation. The primary accounting boundary tracks renewables versus fossil assets and excludes nuclear from the main numerator and denominator. Nuclear is treated separately because its development cycles, governance structures, and retirement dynamics differ substantially from the renewable–fossil comparison that motivates the main framework. By contrast, large hydropower remains within the renewable side of the electricity-generation boundary used here. Storage capacity is not counted as renewable operating generation capacity in this framework, because the accounting boundary is defined over electricity-generation assets rather than flexibility assets. Its role is instead treated as a system-level factor that can shape the relationship between renewable additions, fossil utilization, and fossil stock persistence, and is discussed in the Discussion and Limitations. A nuclear-included sensitivity analysis is reported in SI.3.

Let \(\textrm{Ren}_t\) and \(\textrm{Fos}_t\) denote renewable and fossil operating capacity (MW) at time t. For an interval \((t_0,t_1)\), we define

$$\begin{aligned}& \Delta \textrm{Ren}=\textrm{Ren}_{t_1}-\textrm{Ren}_{t_0},\\&\quad \Delta \textrm{Fos}=\textrm{Fos}_{t_1}-\textrm{Fos}_{t_0},\\&\quad \textrm{Den}=(\textrm{Ren}_{t_0}+\textrm{Fos}_{t_0})+\varepsilon , \end{aligned}$$
(1)

and

$$\begin{aligned} \textrm{SSI}=\frac{\Delta \textrm{Ren}-\Delta \textrm{Fos}}{\textrm{Den}}. \end{aligned}$$
(2)

We define Stock Substitution Intensity (SSI) to measure net renewable-versus-fossil stock reallocation relative to baseline system size, and we use the term “Top-SSI” to refer to the subregions ranking highest in this metric over a given interval. SSI is a stock-based accounting metric rather than a measure of generation displacement, dispatchable adequacy, or emissions substitution. Positive SSI therefore indicates structural change in the operating asset base, not proof that renewable capacity has functionally replaced fossil generation. We classify an interval as expansion-led when \(\Delta \textrm{Fos}\ge 0\), indicating renewable growth alongside non-declining fossil operating stock. To characterize forward-looking development orientation, we define Pipeline Tilt (PT) as the renewable capacity share of the observable project pipeline, referring here to the announced, pre-construction, and under-construction development queue. PT is interpreted as a signal of how green the forward-looking project pipeline is rather than as a deterministic measure of future physical commissioning.

2.3 Sensitivity and boundary conditions

We evaluate three boundary conditions around the main monitoring framework. First, we apply an effectiveness-adjusted greening (EGC) sensitivity to assess whether nameplate capacity accounting may overstate the strength of stock-based substitution. EGC down-weights nameplate operating capacity using technology-specific capacity factors through a nine-scenario parameter scan that varies solar and wind performance assumptions; this adjustment affects SSI-based quantities but does not alter PT (SI.6). EGC is used only as a sensitivity lens and not as a replacement for generation-based analysis. Second, because project realization may diverge from forward-looking pipeline signals, we test archetype stability under three alternative PT definitions: \(\textrm{PT}_{all}\) (announced, pre-construction, and construction), \(\textrm{PT}_{mid}\) (pre-construction and construction), and \(\textrm{PT}_{hard}\) (construction only) (SI.4). These variants provide status-certainty bounds on pipeline orientation. Third, because retirement timing is only partially observed, we interpret fossil exit using bounded retirement evidence that juxtaposes observed retirements against net fossil stock contraction inferred from operating-stock differences (SI.2). For rank-based summaries, we additionally exclude denominator-driven low-baseline cases and flag small systems in Top-SSI tables (SI.1).

3 Results

3.1 Mismatch between pipeline orientation and realized stock reallocation

Figure 1 maps the 17 subregions into a diagnostic space defined by Stock Substitution Intensity (SSI, 2020–2026) and Pipeline Tilt (PT, 2026). Throughout, \(\textrm{SSI}_{cap}\) is interpreted strictly as a stock-based measure of net structural reallocation in operating capacity, not as a measure of generation displacement, dispatchable adequacy, or emissions substitution. Table 1 provides regional context by summarizing 2026 operating stock together with regional median SSI and PT.

Globally, project-pipeline composition is strongly renewable-leaning, with a median PT of 0.8275. Yet realized stock reallocation remains sharply polarized, producing a quadrant distribution of Q1 (high PT, positive SSI) = 7, Q2 (high PT, non-positive SSI) = 2, Q3 (low PT, positive SSI) = 7, and Q4 (low PT, non-positive SSI) = 1. This pattern shows that a green-oriented project pipeline does not reliably translate into realized stock substitution over the same period. Northern Africa illustrates the clearest mismatch (the Q2 archetype): despite a high PT (\(\approx 0.893\)), its SSI over 2020–2026 is slightly negative (\(\approx -0.026\)), indicating that renewable-oriented development intent can coexist with weak or absent net stock replacement in the operating base. This outlier suggests that project intent and realized stock change have not yet aligned. Possible contributing factors include delayed project realization, continued fossil retention, and broader financing, infrastructure, or governance constraints, although the present analysis does not identify their relative importance.

This polarization is robust to alternative pipeline-status definitions. Quadrant assignments are identical when comparing \(\textrm{PT}_{all}\) and \(\textrm{PT}_{mid}\) (17/17 agreement), where \(\textrm{PT}_{all}\) includes announced, pre-construction, and construction projects and \(\textrm{PT}_{mid}\) includes pre-construction and construction only. Assignments remain largely stable under the stricter \(\textrm{PT}_{hard}\) definition (13/17 unchanged), which is restricted to projects under construction; importantly, the high-PT/non-positive-SSI archetype remains present across all variants. The effectiveness-adjusted sensitivity also does not overturn this interpretation. EGC reweights nameplate operating capacity using technology-specific effectiveness factors, and the associated nine-scenario scan varies solar and wind performance assumptions. Across this scan, the capacity-based Q1 count of 7 falls to 6 under EGC, with 6/7 capacity-based Q1 members retained in all scenarios. We therefore treat EGC as a supporting sensitivity rather than a competing transition metric.

The key point is not simply that PT and SSI differ numerically, but that they capture distinct dimensions of transition monitoring that would otherwise be collapsed under conventional addition- or pipeline-based readings. Figure 1 should therefore be read as a stock-based diagnostic of how realized operating-stock change aligns, or fails to align, with forward-looking project-pipeline orientation.

Fig. 1
Fig. 1
Full size image

Quadrant scatter of Stock Substitution Intensity (SSI) over 2020–2026 versus Pipeline Tilt (PT) in 2026. The x-axis display range is fixed to [−0.25, 0.25]; subregions outside this range are shown as off-scale boundary markers. Point size uses square-root scaling of baseline operating stock. Reference lines indicate SSI = 0 and the median PT, yielding four quadrants in the stock–pipeline space. An effectiveness-adjusted greening (EGC) overlay is shown for selected subregions (arrows indicate repositioning of SSI while PT is unchanged), indicating that nameplate-capacity accounting may modestly overstate stock-based substitution in near-threshold cases. PT is a pipeline orientation metric; sensitivity to alternative pipeline-status definitions is reported in SI.4, and EGC sensitivity in SI.6

Table 1 Regional summary of 2026 operating stock and transition metrics. Operating capacity is reported in GW. Median SSI refers to 2020–2026 stock-based reallocation in operating capacity; median PT refers to the 2026 renewable share of pipeline capacity
Full size table

3.2 Fast renewable growth is often expansion-led, not displacement-led

Figure 2 traces the spatial evolution of \(\textrm{SSI}_{cap}\) across 2010–2015, 2015–2020, and 2020–2026, revealing pronounced heterogeneity and a distinct shift toward positive net stock reallocation in several large systems by the core window. On their own, however, positive SSI values do not establish that a transition is displacement-led, because this framework tracks stock-level reallocation in operating capacity rather than changes in generation, utilization, or system flexibility.

Table 2 shows why. Among the Top-SSI subregions, the expansion-led share—calculated as the proportion of these leading subregions for which the net change in fossil operating capacity is non-negative (\(\Delta \textrm{Fos} \ge 0\))—is 100% in 2010–2015, 100% in 2015–2020, and 80% in 2020–2026. Thus, even the highest-SSI cases are usually additive rather than displacement-led: renewable build-out is occurring alongside non-declining fossil operating stock rather than through systematic fossil contraction. Western Europe and Eastern Asia exemplify this pattern in 2020–2026, both combining high positive SSI with positive \(\Delta \textrm{Fos}\), whereas Australia and New Zealand is one of the few Top-SSI cases with a slight fossil decline.

This pattern is not confined to the Top-SSI subset. Across all 17 subregions in 2020–2026, 14 register a positive SSI, but 12 of those 14 remain expansion-led under the criterion \(\Delta \textrm{Fos}\ge 0\). In other words, a positive SSI score often occurs without observable fossil contraction. Retaining fossil capacity is not necessarily unjustified in the short run, because some systems may continue to rely on thermal assets for reserve margins, peak demand, or broader energy-security purposes even as fossil generation declines. However, this highlights that renewable growth and physical stock contraction remain distinct system dynamics. The stock-based framework therefore complements, rather than replaces, generation-based assessments by explicitly monitoring whether the operating asset base itself is contracting.

Fig. 2
Fig. 2
Full size image

Stock Substitution Intensity (SSI) by subregion shown as a three-panel vertical layout for 2010–2015, 2015–2020, and 2020–2026, with a shared diverging color scale centered at SSI = 0. SSI is computed as the baseline-normalized net change in operating renewable versus fossil capacity (excluding nuclear). Low-baseline and extreme-SSI cases are outlined and excluded from color-limit calibration

Table 2 Top-SSI subregions by interval with component decomposition. \(\Delta\)Ren and \(\Delta\)Fos are changes in operating capacity (GW) over the same interval. Expansion-led indicates \(\Delta\)Fos \(\ge 0\). Small-system subregions (baseline operating stock \(<100\) MW) are excluded from this main-text table for interpretability
Full size table

3.3 What conventional transition metrics miss: fossil persistence and transition differentiation

Table 3 compares conventional transition readings with the stock–pipeline diagnostic for six representative subregions. Conventional indicators answer narrower questions: renewable additions identify build-out, fossil decline identifies one dimension of exit, and PT identifies forward-looking orientation. The stock–pipeline framework adds value by separating transition patterns that conventional metrics might otherwise collapse.

The clearest example of divergence is Northern Africa. Under conventional readings, it appears promising because renewable growth is positive and PT is above the median; under the stock–pipeline diagnostic, however, it is a high-PT but slightly negative-SSI case, indicating a green-oriented pipeline without realized stock replacement. Western Europe and Eastern Asia would both be labeled strong progress cases under conventional metrics, but the framework shows that both remained predominantly additive in 2020–2026 rather than clearly fossil-displacing. Australia and New Zealand stands out as a rarer replacement-like case, combining a high SSI with a slight fossil decline. North America and Eastern Europe reveal a different separation: realized stock change and forward-looking pipeline orientation need not move together.

Interpretation of fossil exit remains constrained by incomplete and heterogeneous observability of retirement timing across regions and fossil types. Because retirement timing is only partially visible, we interpret exit using bounded retirement evidence that juxtaposes observed retirements (capacity with recorded retirement years) against net fossil stock contraction inferred from operating-stock differences. Within this bounded framework, the observable subset of Q1 subregions shows renewable additions exceeding observed fossil retirements by approximately 5:1, consistent with an add-green-dominant pattern. Supporting stock-reconstruction diagnostics reinforce this observability challenge: in the 2026 reconciliation, missing start years account for 198,348.93 MW of operating-status capacity, and units with a retired year \(\le 2026\) but a status still marked as operating account for 73,765.60 MW, together contributing to a 3.16% gap between the status-only and year-filtered operating totals.

Taken together, these results show that conventional transition metrics can miss fossil persistence when they collapse distinct patterns into a single narrative of progress. The stock–pipeline diagnostic does not simply identify whether renewable capacity is growing; it tracks whether that growth is additive or replacement-linked, whether future pipeline orientation aligns with realized stock change, and whether apparent exit is supported by bounded retirement evidence. It is therefore intended as a stock-based complement to generation- or emissions-based assessments, not as a substitute for them.

Table 3 Conventional transition readings versus the stock–pipeline diagnostic for six representative subregions
Full size table

4 Discussion

4.1 Why additive transition matters for transition monitoring

A central implication of our results is that renewable additions can coexist with fossil persistence, so capacity-based transition optics can bias transition monitoring toward visible expansion rather than fossil exit. Across the core 2020–2026 window, most subregions with positive SSI remain expansion-led, and even the highest-SSI cases are usually characterized by renewable growth layered onto non-declining fossil operating stock rather than systematic fossil contraction. In this sense, headline greening often reflects additive expansion rather than structural replacement. This matters because long-lived fossil infrastructure can sustain path dependence and carbon lock-in even under rapid renewable build-out (Seto et al. 2016; Bertram et al. 2015). Consistent with the committed-emissions literature, existing and planned infrastructure can therefore continue to embed substantial future emissions trajectories when fossil exit lags behind visible low-carbon additions (Steven et al. 2010; Tong et al. 2019; Pfeiffer et al. 2018).

Our contribution is not to infer emissions outcomes directly from operating-stock patterns, but to clarify a monitoring problem with immediate governance relevance: transition narratives that emphasize additions alone can become systematically optimistic if they do not also track fossil persistence, retirement transparency, and the conditions under which operating stocks actually contract. The effectiveness-adjusted sensitivity reinforces this point without overturning the main archetype structure. Although EGC is only a sensitivity lens, it suggests that nameplate-capacity accounting may modestly overstate substitution in near-threshold cases, especially outside the high-PT regime. Capacity-based metrics remain useful for tracking deployment, but they should not be read as self-sufficient evidence of functional replacement.

4.2 What current metrics miss

A broader implication of the stock–pipeline framework is that commonly used transition indicators answer different questions and should not be treated as interchangeable. Renewable additions identify build-out; pipeline tilt captures forward-looking development orientation; realized stock reallocation captures net change in the operating base. The comparative results show that these signals often diverge. Northern Africa, for example, appears promising under conventional readings because renewable growth is positive and pipeline orientation is strongly green, yet realized stock substitution remains slightly negative. Western Europe and Eastern Asia would both be labeled strong progress cases under conventional metrics, but the framework shows that both remain predominantly additive rather than clearly fossil-displacing in 2020–2026. Australia and New Zealand stands out as a rarer replacement-like case, while North America and Eastern Europe show that positive stock change need not coincide with strong forward pipeline momentum.

The analytical value of the framework, therefore, is not simply that it introduces additional metrics. Rather, it separates dimensions that are routinely conflated in transition assessment: renewable build-out, fossil persistence, and forward-looking pipeline orientation. This matters because conflation can obscure fossil persistence, especially when green investment grows faster than fossil assets retire. Interpreting transition through a single metric risks collapsing qualitatively different patterns into the same narrative of success.

4.3 Drivers of fossil persistence: capacity, demand growth, and energy security

An important reason why additive transition persists is the non-equivalence between installed capacity and dispatched generation. Variable renewable resources such as wind and solar typically operate at lower capacity factors than fossil thermal assets and, without sufficient storage or flexibility resources, do not provide the same dispatchable service on a one-to-one nameplate basis. As a result, retained fossil capacity may remain important for reserve margins, peak-demand coverage, or broader energy-security purposes even as renewable penetration rises.

In many systems, higher renewable penetration may therefore reduce fossil utilization without triggering immediate fossil retirement. Thermal plants may shift into backup or peaking roles, remaining physically active in the operating stock even if their generation share declines. From a system perspective, this is not necessarily undesirable in the short run where retained thermal assets continue to support adequacy, reliability, or energy security. From a transition-monitoring perspective, however, it means that visible renewable growth does not necessarily imply rapid contraction of the fossil asset base. Rapid stock contraction without sufficient flexibility resources could create reliability and adequacy challenges in some systems, while prolonged retention of underutilized fossil assets may also sustain longer-run stranding and lock-in risks (Yangsiyu et al. 2022). Our framework is designed to capture this stock-level persistence directly. The EGC sensitivity further indicates that nameplate-capacity metrics can modestly overstate the practical strength of substitution, but it does not overturn the broader finding that renewable build-out and fossil exit often proceed on different timelines.

This distinction may become more important under conditions of rapid load growth. Growing electricity demand associated with data centers, AI workloads, electrification, and other emerging loads can make additive transition patterns more likely, because renewable additions may be absorbed by demand growth without forcing immediate fossil stock contraction. Where storage, transmission, or other flexibility resources remain limited, these pressures may further reinforce the short-run retention of fossil operating capacity.

These mechanisms should be interpreted as plausible explanations rather than as causal findings from the present analysis. The current framework does not identify the relative contribution of storage limitations, grid constraints, market design, contractual commitments, demand growth, or political economy factors. Nor does it evaluate parallel changes in renewable or fossil generation shares. Rather than substituting for generation-based assessments, the present framework is scoped to isolate the stock dimension of transition, showing why monitoring operating-stock change remains necessary even when renewable generation and capacity additions are substantial.

4.4 Governance implications and phase-out monitoring

The results also point to a critical governance challenge: fossil persistence is not only substantively important, but unevenly observable. With retirement timing only partially documented across regions and fossil types, additions are structurally more visible than exit. This observability asymmetry creates a structural challenge for phase-out monitoring, potentially biasing transition assessment toward growth narratives and away from auditable fossil exit. Our bounded retirement framework is designed as a conservative response to this problem: rather than imputing unobserved decommissioning, it juxtaposes observed retirements with stock-implied contraction and thereby communicates exit evidence under incomplete information. The underlying stock-reconstruction audits reinforce this asymmetry. In the 2026 reconciliation, missing start years account for 198,348.93 MW of operating-status capacity and units with retirement years \(\le 2026\) but still labeled as operating account for 73,765.60 MW, together contributing to a 3.16% gap between status-only and year-filtered operating totals. These discrepancies are large enough to matter for how physical exit is interpreted in climate mitigation governance and phase-out monitoring.

If additive transition is common, then renewable deployment policy alone is insufficient as a marker of credible structural transition. Transition governance must explicitly address retirement transparency, stranded-asset exposure, and the institutional design of policy-managed fossil exit. Emerging work on coal phase-out, compensation mechanisms, and stranded-asset concentration suggests that retirement is mediated by contracts, ownership structures, and explicit policy instruments rather than occurring automatically in response to renewable growth (Jenkins 2014; Rozenberg et al. 2020; Angelika von Dulong et al. 2023; Robert Fofrich Navarro et al. 2026; Erickson et al. 2018; Lazarus and van Asselt 2018; van Asselt 2021a). Clean investment and fossil exit must therefore be treated as linked but distinct governance priorities.

To operationalize these insights, we propose that a minimum transition-monitoring dashboard for climate mitigation governance should separately track five distinct elements: (1) renewable capacity additions; (2) fossil operating-stock change; (3) recorded retirements alongside retirement transparency metrics; (4) project-pipeline composition; and (5) project-pipeline scale under explicit status-certainty bounds. By shifting from single-indicator assessments to a multidimensional dashboard, policymakers can more robustly monitor whether visible greening is translating into the structural fossil phase-out required to meet climate goals.

5 Limitations

Several boundaries contextualize our findings. First, this is strictly a stock-based capacity analysis: the SSI framework tracks structural reallocation in the physical asset base, not delivered generation, dispatchable adequacy, or emissions substitution. Because variable renewables and fossil thermal assets differ in both capacity factors and system functions, a positive SSI should not be interpreted as proof of functional energy replacement. Second, retirement-year observability remains incomplete and systematically heterogeneous across regions and fossil types. This limits inference on fossil-exit dynamics to bounded descriptive evidence rather than full retirement accounting. Third, our macro-subregion design prioritizes global comparability and auditability, but necessarily smooths subnational heterogeneity in grid constraints, market design, policy implementation, and asset age structure. Fourth, project-pipeline orientation remains sensitive to execution uncertainty; although the archetype structure is robust across status variants, PT should be interpreted as a development-orientation signal rather than a deterministic commitment measure.

More broadly, this study does not attempt causal identification. The framework is designed to diagnose stock-level transition dynamics and to clarify how commonly used indicators can obscure fossil persistence, not to estimate the causal effects of specific policies, ownership structures, or market conditions on fossil exit. Specifically, the framework does not quantify actual changes in the generation mix, nor does it trace the evolution of renewable generation shares. It therefore cannot determine the actual utilisation of retained fossil assets or establish whether stock persistence corresponds to frequent dispatch, reserve use, or largely infrequent operation. Furthermore, the present methodology does not explicitly isolate demand-driven system expansion from replacement-driven asset reallocation, including load growth associated with data centers, AI workloads, electrification, and other emerging demands, a distinction that would require integrating capacity data with demand-side and load-growth information. Future research should examine how younger fossil fleets, stranded-asset exposure, storage constraints, grid flexibility, and compensation mechanisms shape the feasibility of policy-managed phase-out.

6 Conclusion

The global power transition is often characterized by additive growth rather than clear structural fossil replacement. By establishing an auditable, stock-based diagnostic framework, we show that rapid net greening in operating capacity frequently coexists with fossil persistence. Over the core 2020–2026 window, most subregions achieving positive stock reallocation remained expansion-led: renewable additions were layered onto non-declining fossil operating capacity rather than driving systematic fossil contraction. Furthermore, while forward-looking project pipelines reflect strong green ambitions globally, realized stock reallocation remains sharply polarized, underscoring a persistent mismatch between developmental intent and physical operating-stock outcomes.

Taken together, these results show that monitoring renewable build-out alone is insufficient for assessing stock-level transition progress. Retained fossil capacity is not necessarily unjustified in the short run where thermal assets continue to support adequacy and energy security; however, renewable additions alone cannot show whether the operating asset base itself is contracting. This reinforces an important interpretive boundary: stock-based reallocation in operating capacity should not be read as equivalent to direct changes in generation, dispatchable adequacy, or emissions outcomes. Nor does the present framework trace renewable generation shares or provide a direct empirical comparison between SSI and generation-based indicators. Instead, it is intended as a complement to generation-based assessments by isolating the stock dimension of transition. Nameplate-capacity measures may additionally modestly overstate the practical strength of substitution in some settings, particularly near-threshold cases. To support credible climate mitigation governance, transition monitoring should explicitly track asset exit alongside asset entry. Retirement transparency and policy-managed fossil phase-out should therefore be treated as core components of credible transition monitoring and governance rather than as secondary concerns to renewable deployment alone.