1 Introduction

Utility-scale solar deployment has expanded rapidly in recent years and now occupies a central place in energy-transition planning, infrastructure investment, and decarbonization strategy worldwide (Gielen et al. 2019; Ardani et al. 2021; O’Shaughnessy et al. 2022; Joshi et al. 2025; Meneguzzo et al. 2015; Altaye et al. 2025). As this expansion has accelerated, cross-national comparison has become increasingly prominent in both academic research and policy-oriented discourse. Yet such comparisons remain predominantly structured around aggregate indicators—most commonly installed capacity, project counts, and country rankings by scale (Global Energy Monitor 2026a; Joshi et al. 2025; Meneguzzo et al. 2015; Wang and Hong 2026). While these indicators effectively describe deployment volume, they are generally less suited to illuminating how national solar portfolios are internally organized, or how different forms of deployment may relate to the planning and implementation conditions that shape them.

Aggregate parity does not imply structural equivalence. Two countries may report similar total solar capacity while exhibiting markedly different combinations of project size and development status. One country may be anchored in a large base of already-operating small and mid-sized projects, whereas another may derive much of its apparent scale from a limited number of very large projects in announced or pre-construction stages. Similarly, one national portfolio may be dense and incremental, while another is sparse but concentrated in mega-project development (Hernandez et al. 2014, 2015; Mulvaney 2017; Sareen et al. 2025; Chen et al. 2025; Owusu-Obeng et al. 2025; Li et al. 2026; Rekik and El Alimi 2024). Similar aggregate totals can therefore mask materially different size–status compositions, with significant implications for how one interprets build-out maturity, pipeline credibility, and the relationship between present operating stock and future expansion potential.

A further complication arises from the fact that project counts and capacity weights illuminate distinct aspects of solar development. Count-based perspectives are highly sensitive to project abundance and to the prevalence of small operating installations, making them well suited to characterizing project ecology—the frequency structure of projects within a national portfolio. Capacity-based perspectives, by contrast, are more sensitive to the concentration effects of very large projects, rendering them more informative about system weight and broader structural position (Shivakumar et al. 2019; Hao and Shao 2021; Hunt et al. 2024; Clò et al. 2025). Treating these two perspectives as interchangeable risks blurring the distinction between dispersed project frequency and concentrated system infrastructure. A country may appear predominantly small-scale and operational in count terms while remaining announcement-heavy in capacity terms, because a relatively small number of large projects dominate its retained megawatts.

These distinctions are especially consequential in utility-scale solar, where deployment is shaped not only by aggregate expansion targets but also by siting constraints, land availability, interconnection conditions, permitting environments, and social acceptance (Hernandez et al. 2014; Mulvaney 2017; Chen et al. 2025; Gorman et al. 2025; Zhang et al. 2024). As a result, countries that appear similar in aggregate volume may differ substantially in the underlying structure of their realized operating stock, intermediate pipeline buildup, and forward-looking large-project ambition. A comparative framework that does not account for these dimensions risks compressing meaningful structural heterogeneity into a single volumetric scale.

Existing literature has provided important insights into renewable energy deployment drivers, policy support mechanisms, national heterogeneity, and solar growth trajectories (Marques and Fuinhas 2011; Omri and Nguyen 2014; Papież et al. 2018; Salim and Rafiq 2012; Hao and Shao 2021; Hunt et al. 2024). Comparatively little attention, however, has been devoted to characterizing countries simultaneously by project scale, development status, and dual count/capacity perspectives in a manner amenable to structural comparison. Within energy research, network-based methods have been increasingly applied to examine system interdependencies, topological properties, and resilience (Kazim et al. 2025). In the present study, bipartite projection is used primarily as a comparative and classificatory device rather than as a tool for detailed topological analysis. What remains underdeveloped is a framework capable of moving beyond aggregate rankings to identify recurring structural configurations in utility-scale solar development while preserving the distinction between project ecology and system weight.

This study addresses that gap by drawing on phase-level data from the February 2026 Global Solar Power Tracker (GSTP) utility-scale dataset, which distinguishes operating, announced, pre-construction, and construction stages under explicit inclusion criteria (Global Energy Monitor 2026a, 2026b). Restricting the analysis to countries with at least 30 records, we define 16 discrete size–status modes derived from four capacity bins and four development statuses, and employ these modes to represent each country as a normalized size–status profile rather than a simple aggregate total. Row-normalized country projections are then constructed to enable structurally consistent cross-national comparison, with capacity serving as the primary analytical lens and project count as a complementary perspective. The study thereby develops a comparative framework for representing utility-scale solar portfolios as size–status structural configurations and examines whether cross-national deployment is better understood through these recurring configurations than through aggregate megawatt rankings alone.

2 Methods

2.1 Data source, retained analytical scope, and mode construction

This study draws on the February 2026 release of the Global Solar Power Tracker (GSTP) and restricts the analysis to the Utility-Scale (1 MW+) main sheet, which tracks utility-scale solar photovoltaic and solar thermal project phases worldwide under explicit inclusion criteria (Global Energy Monitor 2026a, 2026b). The observational unit is the phase-level record rather than the aggregated plant-level record, because large facilities are often developed in multiple phases that differ in capacity and development status. The distributed sheet is excluded because it is compiled as a country-level aggregate and is therefore not structurally comparable to the phase-level utility-scale data. The overall analytical workflow and retained scope are summarized in Fig. 1.

Fig. 1
Fig. 1
Full size image

Study design and retained analytical scope. The left panel summarizes the analytical workflow from retained-scope construction to interpretation and sensitivity assessment. The right panel reports the retained analytical scope applied throughout the main analysis

The main analysis retains four development statuses: operating, construction, pre-construction, and announced, which together represent active operating stock and forward pipeline composition. Residual statuses, including shelved, cancelled, retired, and mothballed, are excluded because the study focuses on active structural development patterns rather than full project life-cycle closure.

Countries are included in the primary analytical sample only if they contain at least 30 retained phase-level records following status cleaning. This threshold serves as a profile-stability criterion, since very sparse country profiles would be unduly sensitive to single-project additions. The retained analytical scope is fixed prior to downstream projection, community detection, and sensitivity analysis.

Each retained phase-level record is assigned to one of four capacity bins: 1–5 MW, 5–50 MW, 50–500 MW, and 500 + MW, defined as \(\:1\le\:\text{M}\text{W}<5\), \(\:5\le\:\text{M}\text{W}<50\), \(\:50\le\:\text{M}\text{W}<500\), and \(\:\text{M}\text{W}\ge\:500\). These thresholds are intended to capture substantively different implementation scales, ranging from localized utility deployments to very large projects. Combining the four retained statuses with the four capacity bins yields 16 discrete size–status modes. These modes are treated as related components of a normalized size–status profile rather than as independent predictors (Aitchison 1982). The framework is compositional in the sense that it compares countries through normalized shares, but it is not intended as a formal log-ratio compositional data analysis (CoDA) treatment.

2.2 Country profiles, projection, and community detection

For each retained country, two country-by-mode matrices are constructed. The first is a count matrix, in which each cell records the number of retained phase-level records in a given size–status mode. The second is a capacity matrix, in which each cell records retained megawatt capacity in that mode. The count view preserves sensitivity to project frequency structure, whereas the capacity view captures system-weighted structure.

Both matrices are generated in raw and row-normalized forms. If \(\:{x}_{cm}\) denotes the value for country \(\:c\) in mode \(\:m\), the row-normalized profile is defined as Eq. 1:

$$\:{p}_{cm}=\frac{{x}_{cm}}{{\sum\:}_{m}{x}_{cm}}.$$
(1)

This transformation shifts the comparison from absolute country weight to within-country mode composition, thereby reducing the disproportionate influence of high-volume countries on similarity calculations.

Country communities are identified on one-mode country projections rather than on the original country–mode bipartite graph, because the analytical target is structural similarity among countries. A raw diagnostic projection is first constructed from unnormalized country rows using dot-product similarity. The substantive analysis is then conducted on row-normalized country projections based on cosine similarity, which is well suited to comparing normalized profile vectors independently of absolute magnitude. To preserve a sparse and interpretable comparison structure, only the top \(\:k=6\) strongest country–country ties are retained per node, thereby avoiding a near-fully-connected similarity graph while retaining dominant structural affinities.

Communities are detected using the Louvain algorithm (Blondel et al. 2008) as implemented in NetworkX (Hagberg et al. 2008), with weighted graph input, seed = 42, and resolution = 1.0. The row-normalized capacity projection serves as the primary analytical path, while the row-normalized count projection is retained as a complementary lens. The baseline resolution parameter is held constant across the main analyses and corresponding sensitivity checks to ensure comparability.

2.3 Structural validation and sensitivity analysis

To evaluate whether the detected community structure reflects meaningful organization beyond a trivial artefact of network degree configuration, the observed country projections are compared against a degree-preserving null baseline (Maslov and Sneppen 2002). For each null replicate, the observed projection is rewired using nx.double_edge_swap; the original edge weights are then permuted and reassigned to the rewired topology; Louvain detection is rerun under identical settings; and modularity is recorded. The null analysis uses 40 replicates, with replicate seeds following the deterministic schedule \(\:42+i\).

To assess how structural differentiation is distributed across the 16-mode architecture, a pre-specified multi-metric mode-distinction audit is applied to all predefined size–status modes. The audit integrates country coverage, concentration among top-contributing countries, cross-community separation, leave-one-mode-out community disruption, edge-contribution share, and an overall salience score. Final mode labels—discriminative, background, and head-dominated—are assigned on the basis of the joint pattern of salience, concentration, and cross-community contribution rather than any single metric in isolation.

A predefined compression rule is applied to the count path to allow higher-order grouping without overwriting the underlying algorithmic partition. Candidate community pairs are considered for compression only when they exhibit very high similarity in both full-profile and discriminative-mode structure, along with consistent dominant-mode logic. Under this rule, a single pair—communities 3 and 6—is retained in the higher-order ecology summary layer.

Sensitivity analysis is conducted under two alternative binning schemes and one alternative country-inclusion criterion. The alternative binning schemes comprise a low-end split (1–5, 5–20, 20–100, 100 + MW) and a giga split (1–10, 10–100, 100–1000, 1000 + MW); the alternative country filter requires a total retained capacity of \(\:\ge\:500\) MW. Sensitivity is evaluated in terms of continuity in the broad structural differentiation logic rather than exact membership invariance, using community correspondence diagnostics including the adjusted Rand index (Hubert and Arabie 1985).

Finally, the analysis is explicitly scoped as a cross-sectional structural comparison of the February 2026 GSTP release. Although project timing fields may be referenced descriptively, they are not treated as direct evidence of realized transition speed, realized conversion probability, or causal sequence.

3 Results

3.1 From aggregate volume to structural space

Fig. 2
Fig. 2
Full size image

Structural normalization contrast. Row normalization substantially reduces head-country dominance in both the count and capacity projections, shifting the country space from raw volumetric ranking toward within-portfolio structural comparison

Table 1 Retained analytical scope and sample summary for the baseline main-paper specification
Full size table

The retained analytical scope is fixed for the main analysis and underlies all subsequent country-level comparisons.

We first establish the retained analytical scope and then show why structural comparison requires normalization. As summarized in Table 1, the February 2026 GSTP utility-scale main sheet was reduced to a retained analytical base comprising 94 countries, 98,942 phase-level records, and 3,573,038.8 MW of retained capacity following status cleaning and country-threshold application. Only operating, construction, pre-construction, and announced records were retained, and each was assigned to one of 16 predefined size–status modes. This retained scope defines the empirical boundary of the study and makes explicit that the analysis concerns cross-sectional differentiation in utility-scale stock-plus-pipeline structure rather than realized transition outcomes.

Within this retained base, raw country projections are strongly dominated by a small number of high-volume countries. Figure 2 shows that in the count view, the top five countries account for 0.614 of total projection strength in the raw graph, falling to 0.084 after row normalization. In the capacity view, the corresponding top-five share declines from 0.665 to 0.093. The same pattern is evident at the single-country level: the largest-country share decreases from 0.170 to 0.018 in the count projection and from 0.333 to 0.020 in the capacity projection. In practical terms, the raw graph functions as a ranking space dominated by large countries, whereas the row-normalized graph enables comparison of internal portfolio structure.

The contrast is especially pronounced in the capacity projection. In the raw graph, the strongest ties cluster around the largest-capacity countries, indicating that absolute system size largely determines network visibility. In the row-normalized graph, by contrast, countries organize according to similarity in size–status composition rather than aggregate scale. Row normalization thus shifts the analytical object from absolute country weight to within-country structural profile, making cross-national structural comparison both feasible and interpretable.

3.2 Capacity-based camps capture system-level differentiation

Fig. 3
Fig. 3
Full size image

Capacity-based national camps identified from the row-normalized capacity projection. The six camps represent distinct system-weighted configurations of utility-scale solar development

Table 2 Summary of the six capacity camps and their dominant structural modes
Full size table

Applying community detection to the row-normalized capacity projection yields six distinct national camps, indicating that global utility-scale solar development does not collapse into a single system-weighted profile (Fig. 3; Table 2). Instead, countries separate into structurally distinct groupings characterized by different combinations of operating stock, pre-construction buildup, and announcement-heavy forward structure.

Camp 1: 50–500 pre-construction comprises 24 countries and is anchored by the pre-construction | 50–500 mode, which contributes approximately 37% of the camp profile. Camp 2: 500 + pre-construction / operating mixed comprises 18 countries and is dominated by pre-construction | 500+; it is also the single heaviest system-weighted camp, accounting for 2,293,741.2 MW, or 64.2% of total retained capacity. Camp 3: 50–500 announcement comprises 17 countries and is defined by the announced | 50–500 mode, which contributes approximately 47% of the camp profile. Together, these three camps represent distinct pipeline-dominated capacity profiles rather than a common forward structure.

Two further camps are primarily operating-based. Camp 4: 1–5 operating comprises 15 countries and is dominated by the operating | 1–5 mode, with its top three modes collectively accounting for 87.6% of the camp profile. Camp 6: 5–50 operating comprises 7 countries and is dominated by the operating | 5–50 mode, which contributes approximately 53% of the camp profile. Camp 5: 500 + announcement occupies a distinct large-announcement position, comprising 13 countries and dominated by the announced | 500 + mode, with the top three modes jointly accounting for 70.8% of camp structure. Taken together, the six camps indicate that countries occupy distinct system-weighted structural positions rather than points along a single developmental gradient.

3.3 Count-based ecologies expose fragmented project environments

Fig. 4
Fig. 4
Full size image

Count-based higher-order ecologies identified from the row-normalized count projection. The five ecologies describe project-frequency environments rather than system-weighted national groupings

Viewed through project counts rather than capacity weight, the same retained country space yields a distinct but equally informative structure. The count path is more fragmented than the capacity path, yet this fragmentation is analytically instructive because it captures variation in project-frequency environments that aggregate capacity measures cannot resolve (Fig. 4).

Ecology A: Broad small-operating comprises 25 countries and is dominated by the operating | 1–5 mode, corresponding to source community 1 in the original count partition. Ecology B: Mixed small/mid-operating comprises 23 countries and is dominated by the operating | 5–50 mode, corresponding to source community 2. Ecology C: High-purity small-operating comprises 22 countries and represents the most internally coherent small-operating configuration; it is formed by compressing source communities 3 and 6, the sole strongly supported merge pair from the count compression audit, with a profile similarity of approximately 0.983.

The remaining two ecologies capture more mixed or transition-oriented project-frequency environments. Ecology D: Pipeline-transition comprises 14 countries and is defined by a pre-construction and announced mixed profile centered on the 5–50 MW layer, corresponding to source community 4. Ecology E: Small-operating mixed comprises 10 countries and is characterized by a blended operating | 1–5 and operating | 5–50 structure, corresponding to source community 5. The count path thus reveals that while many countries share a small-operating base, this base is internally differentiated by profile purity, scale mixing, and pipeline transition orientation rather than constituting a single undifferentiated ecology.

3.4 Cross-view correspondence highlights partial alignment rather than equivalence

Fig. 5
Fig. 5
Full size image

Cross-view correspondence between count ecologies and capacity camps. The mapping reveals partial alignment rather than one-to-one equivalence between project ecology and system-weighted structure

Figure 5 maps the relationship between count-based ecologies and capacity-based camps. The dominant pattern is one of partial alignment rather than equivalence: project ecology and system-weighted structure are related but do not define identical country groupings. Alignment is not, however, uniformly weak. Certain operating-heavy groups exhibit stronger cross-view correspondence, particularly where small- and mid-scale operating profiles remain prominent in both project frequency and retained capacity.

The strongest one-to-one alignment occurs where Ecology C: High-purity small-operating maps onto Camp 4: 1–5 operating: of the 22 countries in Ecology C, 14 also fall within Camp 4. This tighter correspondence is consistent with cases in which operating small-project profiles dominate both the count-based representation and a substantial share of the capacity-based representation. More commonly, however, count ecologies disperse across multiple capacity camps. The clearest example is Ecology A: Broad small-operating, whose 25 countries distribute across Camp 1: 50–500 pre-construction (8 countries), Camp 2: 500 + pre-construction / operating mixed (7), Camp 3: 50–500 announcement (6), and Camp 5: 500 + announcement (4).

A comparable dispersion pattern characterizes Ecology B: Mixed small/mid-operating, whose 23 countries spread across Camp 6: 5–50 operating (6), Camp 1 (5), Camp 5 (5), Camp 3 (4), and Camp 2 (3). Ecology D: Pipeline-transition maps predominantly to Camp 1 (6), Camp 2 (3), Camp 3 (3), and Camp 5 (2). The cross-view pattern thus combines localized areas of tighter correspondence with broader dispersion, indicating that project ecology and system weight capture related but non-interchangeable dimensions of national solar portfolio structure.

3.5 A limited subset of modes drives cross-national polarization

Fig. 6
Fig. 6
Full size image

Mode-distinction audit across the count and capacity views. A limited subset of size–status modes accounts for the majority of cross-national differentiating signal

The existence of camps does not in itself explain what differentiates them. Figure 6 therefore shifts the analysis from country groupings to mode-level structure, asking which size–status modes carry the greatest separation signal. The answer is highly uneven: cross-national differentiation is concentrated in a limited subset of modes rather than distributed evenly across all 16.

The shared discriminative core consists of four modes: operating | 1–5, operating | 5–50, announced | 50–500, and pre-construction | 50–500. These modes occupy the highest ranks in the cross-view priority ordering and define the principal axes along which countries separate in both the count and capacity spaces. Two additional modes are view-specific: pre-construction | 5–50 is discriminative exclusively in the count view, whereas announced | 500 + is discriminative exclusively in the capacity view.

Several further modes are more appropriately characterized as head-dominated rather than broadly discriminative, including construction | 500+, pre-construction | 1–5, operating | 500+, announced | 1–5, and construction | 1–5. The concentration of signal is also apparent quantitatively: in the count view, the combined \(\:{\eta\:}^{2}\) of the top three modes is approximately 4.28 times that of the bottom five; in the capacity view, the corresponding ratio rises to approximately 9.65. The bulk of the differentiating signal is therefore carried by a compact structural core centered on operating modes and mid-scale pipeline modes, with large-announcement modes providing additional capacity-specific separation.

3.6 Structural patterns remain visible under alternative specifications

Table 3 Degree-preserving null-model comparison for the baseline count and capacity projections
Full size table
Table 4 Sensitivity summary under alternative binning schemes and country-inclusion rules
Full size table

Tables 3 and 4 summarize the null-model and sensitivity results; Figure S1 in the Supplementary Information displays the full null distributions. Both analytical paths show community structure well above the degree-preserving null baseline: capacity modularity is 0.684 compared with a null mean of 0.314 (\(\:z=37.1\)), and count modularity is 0.618 compared with 0.293 (\(\:z=33.3\)). No null replicate reached the observed modularity in either path, indicating that the identified groupings are not explained by degree configuration alone.

Sensitivity analysis assesses continuity in the broader structural logic rather than exact camp membership. Under alternative binning schemes, the adjusted Rand index (ARI) ranges from 0.49 to 0.54 across both views. These values indicate moderate reassignment at the membership level, particularly in the count view, but do not imply dissolution of the main structural axes. The capacity low-end split specification shows the largest member-level sensitivity, with top-20 tier stability declining to 0.40. In the same specification, however, status-level stability remains 1.0 and top-5 camp-signature overlap reaches 0.80, suggesting that broader structural identity remains identifiable even as detailed placement shifts.

Expanding the analytical universe to 124 countries with retained capacity \(\:\ge\:500\) MW (99,338 rows; 3.66 TW) yields ARI values of 0.617 for the capacity view and 0.490 for the count view, with top-20 dominant profile stability equal to 1.0 in both paths. This pattern again indicates that exact country membership can be sensitive to specification choices, while the main operating-heavy, mid-scale pipeline, and large-announcement configurations remain recognizable under varied assumptions about country inclusion and capacity binning.

Taken together, the sensitivity checks support continuity in the principal structural axes rather than exact camp-membership invariance. Accordingly, these findings should be read as evidence of broad structural continuity, not as evidence of invariant country-level classification across specifications.

4 Discussion

4.1 Multiple structural configurations in global utility-scale solar development

Cross-national energy-transition comparisons often place countries along a single scale inferred from aggregate deployment totals or capacity rankings (Gielen et al. 2019; Joshi et al. 2025). The structural patterns identified here challenge that simplification. Rather than representing different positions on one aggregate scale, the retained country space separates into distinct structural configurations defined by different combinations of operating stock, pre-construction buildup, and announcement-heavy capacity structure.

This distinction matters because countries with similar aggregate solar totals can occupy materially different structural positions. Operating-heavy countries are anchored in realized infrastructure, pre-construction-heavy countries in more advanced but still unrealized pipelines, and announcement-heavy countries in projected capacity. These configurations are not simply earlier and later points on an aggregate scale. They represent different arrangements of present infrastructure and future expansion, likely associated with different combinations of land-use, grid, finance, and policy conditions (Hernandez et al. 2014; Ardani et al. 2021).

The main implication is that aggregate rankings remain informative but incomplete. Total installed megawatts can indicate scale, yet they do not show how that scale is composed or whether it rests mainly on realized operating depth or forward pipeline structure. The present framework therefore complements volumetric indicators with a structural lens that makes these differences visible and comparable across countries.

4.2 Project ecology, system weight, and the interpretation of projected capacity

A central implication of this study is that capacity-based camps and count-based ecologies should be read together rather than treated as interchangeable summaries of the same national profile. The capacity path captures system weight: it emphasizes where retained megawatts are concentrated and highlights the structural importance of very large projects and forward pipelines. The count path captures project ecology: it is more sensitive to the frequency environment in which projects are distributed, especially within the small-operating world. Their partial divergence is therefore substantively informative rather than a mere methodological artifact (Joshi et al. 2025; Hernandez et al. 2014; Mulvaney 2017; Chen et al. 2025; Zhang et al. 2024).

This divergence shows that project abundance does not necessarily translate into system weight. A country may sustain a dense small-project operating ecology while also appearing pipeline-heavy or announcement-heavy in the capacity view if a limited number of large projects dominate retained megawatts. Conversely, a country may appear system-heavy because of a few very large projects while lacking a comparably dense small-operating base. This interpretation is consistent with evidence that distributed and utility-scale photovoltaic expansion can follow distinct spatial and institutional logics, and that utility-scale deployment is often shaped by implementation constraints not captured by simple project counts alone (Clò et al. 2025; Gorman et al. 2025; Chen et al. 2025).

The same logic applies to announcement-heavy and pipeline-heavy configurations. High capacity totals may reflect large forward pipelines rather than built operating depth, and the distance between announcement and realization may vary substantially across countries depending on land availability, interconnection capacity, financing conditions, permitting regimes, and institutional coordination (Hernandez et al. 2015; Sareen et al. 2025; Owusu-Obeng et al. 2025; Li et al. 2026; Rekik and El Alimi 2024). A structural reading therefore helps distinguish realized development from projected capacity without assuming that projected capacity will necessarily translate into operating infrastructure.

4.3 Structural typologies as a comparative lens

Beyond serving as descriptive categories, the structural typologies identified here provide a useful lens for comparative benchmarking and planning interpretation. Aggregate rankings can identify which countries are large, fast-growing, or pipeline-heavy, but they are less informative about whether expansion is anchored by a dense operating base, an intermediate pre-construction pipeline, or a relatively small number of very large proposed projects. By locating countries within recurring size–status configurations, the present framework makes these latent structural differences visible and comparable.

This perspective is especially useful for interpreting whether present project ecology and projected capacity structure are broadly aligned. Countries that appear small-operating in count terms but announcement-heavy or large-pipeline in capacity terms may occupy a different implementation position from countries whose count and capacity profiles are more closely aligned. Such contrasts may reflect differences in siting conditions, land availability, interconnection readiness, project-development institutions, or social acceptance, rather than simple differences in total solar volume (Chen et al. 2025; Gorman et al. 2025; Zhang et al. 2024).

At the same time, this framework does not directly measure site-specific land-use impacts, permitting delays, grid integration capacity, or causal policy drivers. Its value lies in comparative diagnosis: it helps identify whether national utility-scale solar portfolios appear structurally coherent, pipeline-heavy, or strongly dependent on future-oriented large-project concentration. Recent literature suggests that large-project expansion is often sensitive to spatial and institutional constraints that aggregate metrics alone cannot reveal (Clò et al. 2025; Gorman et al. 2025; Joshi et al. 2025; Hunt et al. 2024). In this sense, the framework is best understood as a planning-relevant comparative structural lens that complements, rather than replaces, direct measures of environmental performance or realization probability.

5 Limitations and future extensions

While this study establishes a size–status framework for cross-national solar comparison, its interpretive boundaries should be stated clearly. First, the analysis relies on a single cross-sectional snapshot of the February 2026 GSTP release and excludes residual statuses such as cancelled, shelved, retired, and mothballed projects. It therefore captures active operating stock plus forward pipeline structure at one point in time, but does not directly encode historical project mortality, pipeline attrition, retirement dynamics, or realized conversion probability. A country that appears announcement-heavy in the present framework may or may not ultimately convert that profile into operating capacity; the resulting map should therefore be interpreted as a representation of active stock-plus-pipeline structure rather than as a direct indicator of pipeline quality or realization likelihood.

Second, the analysis is conducted at the phase level rather than the plant level. This provides a more granular representation of development status and project scaling, especially where large facilities are developed across multiple phases with different capacities or statuses, but it also means that multi-phase projects may contribute multiple records to the analytical base. This consideration is especially relevant for the count-based ecology path, where heavily phased utility-scale developments may contribute more to apparent project density than they would under a plant-level representation. In addition, the framework characterizes deployment structure rather than solar production conditions or technological performance. It does not directly incorporate solar irradiance, climate, latitude, seasonality, storage integration, panel efficiency, or grid-integration capacity, all of which may shape the production value and implementation feasibility of utility-scale solar portfolios.

Third, the sensitivity results should be interpreted in terms of continuity in broad structural logic rather than exact camp-membership invariance across every alternative specification. When bin boundaries or country-inclusion rules are modified, some edge-case countries may shift between adjacent camps or undergo tier-level reshuffling. This is best understood as an expected feature of comparison in a continuous structural space, provided that the higher-order axes of differentiation remain visible and interpretable across specifications. These boundaries point toward future work extending the country–mode framework to successive tracker releases as panel data, enabling analysis of structural change, pipeline realization, attrition, and movement across operating-heavy, pre-construction-heavy, and announcement-heavy configurations over time.

6 Conclusion

This study develops a size–status framework for characterizing global utility-scale solar development beyond aggregate megawatts alone. It shows that cross-national comparison is more informative when grounded in internal portfolio structure than in installed capacity or project counts in isolation. Viewed through this structural lens, global utility-scale solar development does not collapse into a single aggregate continuum, but differentiates across two related yet non-equivalent dimensions: system weight and project ecology. The capacity view identifies six system-weighted national camps, while the count view identifies five project-frequency ecologies, with only partial alignment between them. This differentiation is driven disproportionately by a limited subset of key size–status modes rather than distributed evenly across all 16 modes, and the broad structural logic remains coherent and interpretable under alternative specifications.

The central contribution of this paper is therefore not to re-rank countries by aggregate solar scale, but to identify the distinct structural configurations through which countries are organizing utility-scale solar development and to show why those configurations cannot be reduced to any single volumetric metric. Aggregate megawatt totals can indicate scale, but they do not reveal whether that scale is anchored in realized operating depth, intermediate pipeline buildup, or large-scale announced capacity concentrated in a limited number of projects. By making these structural differences explicit and comparable, the framework provides a comparative diagnostic lens for interpreting how similar aggregate solar totals may correspond to materially different project-frequency structures and system-weighted profiles. Future work can link these structural profiles to direct measures of land use, permitting, grid integration, policy context, and project realization.