Digitalisation and Entrepreneurial Ecosystems as Drivers of Energy Start-Ups: Evidence from Cross-Country Panel Data

Abstract

The accelerating energy transition and digital transformation have increased the importance of understanding the drivers of energy-related entrepreneurship and investment across countries. This study aims to investigate how digitalisation and entrepreneurial ecosystem development influence the number and funding of energy-related start-ups, with particular attention to stage-specific effects, lagged dynamics, and non-linear relationships in a cross-country panel setting. The analysis is based on panel data from the European Commission (DESI), the International Energy Agency, and StartupBlink, covering 25 countries (2017–2022) and a global sample (2019–2023), and is estimated using Poisson pseudo-maximum likelihood models with fixed effects, lagged variables, and non-linear specifications in R. The findings show that digitalisation has a limited, selective relationship with energy-related entrepreneurship, whereas entrepreneurial ecosystem development plays a more consistent role. Digital connectivity is associated mainly with improved early-stage funding conditions, whereas broader digitalisation indicators do not systematically explain start-up formation. Stronger entrepreneurial ecosystems are linked to both higher green start-up activity and a shift in investment from early-stage ventures to more mature digital energy firms. The non-linear results further suggest diminishing returns to ecosystem development in later-stage green funding, indicating potential saturation effects in highly developed ecosystems. These findings suggest that policies aimed at accelerating sustainable energy entrepreneurship should go beyond general digitalisation strategies and focus more directly on strengthening inclusive entrepreneurial ecosystems, improving access to finance across the start-up lifecycle, and preventing excessive investment concentration in already mature ventures.

1. Introduction

The global transition towards low-carbon energy systems has increased the strategic importance of innovation and entrepreneurship in the energy sector. Achieving net-zero targets requires faster deployment of both mature and emerging clean technologies, while start-ups play an important role in commercialising solutions such as energy storage, hydrogen and digital energy systems [1,2]. Innovation ecosystems and entrepreneurial ventures are also essential for scaling renewable energy capacity and supporting decentralised, flexible energy systems [3]. Thus, energy start-ups are becoming key actors in the structural transformation of global energy systems.

Digitalisation further reshapes the energy sector by enabling new business models, reducing entry barriers and improving efficiency, resilience and system integration. Technologies such as artificial intelligence, advanced data analytics and smart infrastructure are increasingly embedded in energy systems, strengthening the link between digital transformation and sustainable energy innovation [4,5]. However, despite strong policy attention to digitalisation, empirical evidence remains limited on how specific dimensions of digital development translate into start-up creation and investment flows in energy-related sectors.

Entrepreneurial ecosystems are equally important because access to capital, ecosystem connectivity, talent and institutional support influence start-up formation and scaling, especially in capital-intensive sectors such as energy [6,7]. Recent global rankings also show that start-up activity and venture capital remain concentrated in a limited number of leading ecosystems, revealing persistent disparities between advanced and emerging markets [8,9]. This suggests that ecosystem maturity affects not only innovation capacity but also investment allocation across the start-up lifecycle.

Energy-related start-ups emerge at the intersection of clean technology, sustainability transition and entrepreneurial ecosystem development. Their growth depends not only on technological availability, but also on institutional, financial, digital and market conditions that enable commercialisation and scaling [1,3,4]. Since start-up activity and venture investment remain concentrated in leading ecosystems, energy entrepreneurship requires an ecosystem-based perspective rather than a purely technology-based explanation [7,9].

Against this background, the present study focuses on two mechanisms that may shape energy-related entrepreneurship across countries: digitalisation and entrepreneurial ecosystem maturity. The paper examines whether these factors influence not only the number of energy-related start-ups, but also the allocation of funding across green and digital ventures and across early-stage and later-stage investment. This focus is important because sustainable energy entrepreneurship requires both technological opportunity and ecosystem conditions that enable start-ups to emerge, attract capital and scale.

The remainder of the paper is structured as follows. Section 2 reviews the theoretical background and formulates the hypotheses. Section 3 describes the data and methodology. Section 4 presents the empirical results. Section 5 discusses the findings, and Section 6 concludes with theoretical, empirical, methodological and policy implications.

2. Background and Literature Review

2.1. Policy, Macroeconomic and Structural Conditions of Energy Entrepreneurship

The regulatory and policy environment forms one of the key foundations for sustainability-oriented entrepreneurship. Environmental innovation is closely linked to regulatory frameworks, market-based incentives, and firm-level capabilities, which together shape firms’ capacity to respond to decarbonisation pressures [10,11]. More recent research confirms that environmental policy stringency, green finance and institutional quality can stimulate eco-innovation and sustainable business activity, particularly in sectors exposed to energy transition challenges [12,13,14]. Energy-related uncertainty, environmental regulation and reporting quality also influence how firms and investors respond to green transition pressures [15,16,17]. In addition, sector-specific policy instruments, decarbonisation strategies and energy-cost considerations shape the broader opportunity structure for sustainable entrepreneurship [18,19]. These findings suggest that energy start-ups develop within a policy-sensitive environment in which regulation can create incentives, but also uncertainty and adjustment costs.

Energy entrepreneurship is also embedded in broader macroeconomic and structural conditions. Financial development, renewable energy expansion, energy intensity and green growth trajectories influence the extent to which countries can generate and support sustainability-oriented entrepreneurial activity [18,20,21]. The interaction between finance, entrepreneurship, innovation and policy is especially relevant for sustainability transitions because start-ups require not only technological opportunities but also favourable economic and institutional conditions for scaling [22]. Human capital, labour market conditions and macroeconomic stability further affect enterprise creation and entrepreneurial performance, particularly when start-ups operate in uncertain or capital-intensive sectors [23,24,25]. Strategic potential, knowledge resources and leadership capacity also determine whether firms and regions can transform innovation opportunities into sustainable competitive advantages [26,27]. This macro-level perspective reinforces the need to analyse energy start-ups within wider socio-economic systems rather than as isolated entrepreneurial units.

2.2. Digitalisation and Energy Start-Up Development

Within this broader setting, digitalisation is commonly expected to support start-up formation and investment by improving connectivity, reducing information frictions, enabling data-driven business models and expanding access to markets. Artificial intelligence, digital platforms and data analytics can increase start-up efficiency, scalability and innovation performance, including in energy-related sectors [28,29,30]. Digital maturity and knowledge management capabilities also strengthen innovation processes and organisational performance, especially among SMEs and technology-oriented ventures [31,32]. Digital transformation further contributes to methodological innovation, supply-chain transformation and economic growth within the energy transition framework [33,34]. However, the effect of digitalisation on energy start-up formation and funding remains theoretically ambiguous, because digital capabilities alone may be insufficient without complementary entrepreneurial networks, finance, institutional support and sector-specific demand.

2.3. Entrepreneurial Ecosystems, Universities and Innovation Networks

Entrepreneurial ecosystems, therefore, provide a more comprehensive framework for explaining why some countries are more successful than others in generating and financing energy-related start-ups. Ecosystems combine access to finance, human capital, institutional support, infrastructure, knowledge flows, entrepreneurial culture and networks of public and private actors. Ecosystem quality influences start-up success through access to resources, entrepreneurial intention and self-efficacy [35,36]. University–industry collaboration, technology business incubators and innovation networks further facilitate knowledge transfer and start-up performance in technology-intensive sectors [37,38,39]. Universities are important within this ecosystem logic because research universities can provide digital entrepreneurial tools, knowledge infrastructure and innovation capacity for start-up development [40,41,42]. Broader creative-industry strategies and Open Science policies can also support the diffusion of innovation and entrepreneurial activity by strengthening knowledge exchange and openness [43,44].

The ecosystem approach is particularly important for energy-related entrepreneurship because green and digital energy ventures often face high technological uncertainty, long development cycles and substantial financing needs. Innovation ecosystems are connected to the development of the circular economy, industrial transformation, reliable infrastructure, and the financial landscape for renewable energy investments [45,46,47]. At the same time, the relationship between digitalisation and entrepreneurial ecosystems is not automatically complementary. Digital and ecosystem development may generate synergies, but they may also produce trade-offs, for example, when advanced smart-city or innovation systems concentrate resources in already mature ecosystems rather than widening entrepreneurial participation [48,49]. This means that stronger ecosystems may not only stimulate start-up creation but also shift investment allocation from early- to later-stage firms.

2.4. Financing, Firm-Level Mechanisms and Sustainability-Oriented Entrepreneurship

The firm-level and financing dimensions further explain why the impact of digitalisation and ecosystems may vary across types and stages of energy start-ups. Entrepreneurial orientation, ambidexterity, transformational leadership and organisational adaptability enhance firms’ ability to respond to uncertainty and exploit innovation opportunities [50,51,52]. Entrepreneurial experience, psychological resilience, financial perceptions and risk management also shape decision-making under instability [53,54]. Behavioural incentives, green promotions, and financial instruments can further encourage sustainable practices and the adoption of eco-innovation [55,56]. Access to finance remains especially critical because energy start-ups are often capital-intensive and exposed to regulatory, technological and market risks. Funding availability, investment practices and venture capital decisions are shaped by labelled start-ups’ financing mechanisms, renewable energy investment risks and environmental policy pressures [57,58,59]. Administrative procedures and regulatory barriers may also constrain the development of clean and digital energy ventures [60,61]. Public support, external knowledge sourcing, and governance quality further moderate the extent to which innovation potential is translated into sustainable entrepreneurial outcomes [62,63,64].

Sustainability-oriented entrepreneurship adds another layer to this argument because green start-ups are expected not only to generate economic value but also to contribute to environmental transformation. Green entrepreneurial orientation, relational capital, eco-innovation and circular economy practices support sustainable business performance and strengthen the role of entrepreneurship in the green transition [65,66,67]. In the energy sector, new service-based business models and digital solutions create additional opportunities for entrepreneurial participation, while also increasing the complexity of market dynamics and investment decisions [68]. Thus, the development of energy-related start-ups depends on the interaction of policy incentives, digital capabilities, ecosystem maturity, firm-level resources and financing conditions.

2.5. Research Gap and Hypotheses

Despite this growing body of evidence, an important research gap remains. Existing studies usually examine digitalisation, environmental policy, universities, firm-level capabilities, finance, or entrepreneurial ecosystems separately, while paying less attention to how digitalisation and ecosystem maturity jointly influence energy start-up formation and funding across countries. Moreover, previous research rarely distinguishes between green and digital energy start-ups, between early-stage and later-stage funding, or between immediate, lagged and non-linear effects. This is important because the same ecosystem conditions may stimulate start-up creation, redirect capital towards mature ventures, or produce diminishing returns in highly developed innovation environments. Therefore, a comprehensive cross-country analysis that integrates digitalisation and entrepreneurial ecosystem development is necessary to clarify how these factors shape energy-related start-up dynamics in the context of sustainable energy transition.

Based on this research gap, the study tests the following hypotheses:

H1. 

Digitalisation is significantly associated with energy-related start-up formation and funding, but its effects differ across digitalisation dimensions and entrepreneurial outcomes.

H2. 

Entrepreneurial ecosystem maturity is significantly associated with energy-related start-up formation and funding, with different effects across green and digital start-ups and across early-stage and later-stage funding.

H3. 

The effects of digitalisation and entrepreneurial ecosystem maturity on energy-related start-up dynamics are not limited to immediate linear relationships; they may also include lagged effects and non-linear patterns, including diminishing returns at higher levels of ecosystem development.

This study aims to investigate how digitalisation and entrepreneurial ecosystem development influence the number and funding of energy-related start-ups, with a particular focus on stage-specific effects, lagged dynamics, and non-linear relationships in a cross-country panel setting.

3. Materials and Methods

3.1. Data Sources and Sample Structure

The empirical analysis is based on a panel dataset combining indicators of digitalisation, entrepreneurial ecosystems, and energy-related start-up activity across countries and time. The data are drawn from several authoritative international sources. Indicators of digital development are obtained from the European Commission [69] through the Digital Economy and Society Index (DESI), which provides harmonised measures of connectivity, digital public services, human capital, and the integration of digital technologies. Data on start-up activity and funding in the energy sector are sourced from the International Energy Agency Energy Start-up Data Explorer [70], which offers detailed information on the number of start-ups and investment flows at different stages (early and later). In addition, information on the broader entrepreneurial environment is captured using the Global Startup Ecosystem Index (GSEI), compiled by StartupBlink [71,72,73,74,75] based on annual ecosystem rankings for the period 2019–2023. The combined dataset includes two complementary samples: a DESI-based panel covering 25 countries over 2017–2022 (150 observations) and an extended GSEI-based panel covering 2019–2023 (414 observations), allowing for both within-European and global comparative analysis. It should be noted that these two samples differ in both temporal and cross-sectional composition. The DESI-based sample is restricted to countries covered by the Digital Economy and Society Index and therefore mainly reflects a European digitalisation context. In contrast, the GSEI-based sample includes a broader set of countries and captures a more heterogeneous global entrepreneurial ecosystem landscape. For this reason, the second sample should not be interpreted as a direct replication of the first on a longer time horizon, but rather as a complementary comparative extension that tests whether the broader relationship between entrepreneurial ecosystem development and energy start-up dynamics remains observable in a larger, more diverse country setting. The two panels differ in their degree of balance and cross-sectional coverage. The DESI-based panel is balanced at the country–year level, covering 25 countries over six years. The GSEI-based dataset is broader and more heterogeneous, covering a larger set of countries over 2019–2023; however, the number of observations differs across model specifications because of data availability for specific dependent variables and the structure of the fixed-effects PPML estimations. Therefore, the comparison between the DESI-based and GSEI-based results should not be interpreted as a strict test of parameter stability across identical panels. Rather, the two samples provide complementary evidence from different empirical settings: a more homogeneous European digitalisation sample and a broader global entrepreneurial ecosystem sample.

3.2. Variables and Terminology

The dependent variables capture two dimensions of entrepreneurial outcomes: start-up formation and funding. To ensure terminological consistency, the study uses the term ‘energy-related start-ups’ as the general category encompassing all start-ups included in the analysis. Within this category, two subgroups are distinguished (

Table A4. Standardised terminology used in the study.

Term Used in the Article	Meaning in the Study	Corresponding Variable/Category
Energy-related start-ups	General umbrella term for all start-ups analysed in the energy domain	Green energy start-ups and digital energy start-ups
Green energy start-ups	Energy-sector start-ups linked to green, clean-energy or sustainability-oriented activities	Energy and green start-ups
Digital energy start-ups	Energy-sector start-ups linked to digital technologies, platforms, data-driven or smart-energy solutions	Energy and digital start-ups
Early-stage funding	Initial investment flows into energy-related start-ups	Early-stage funding variables
Later-stage funding	Growth or scaling investment flows into energy-related start-ups	Later-stage funding variables

Source: Authors’ own elaboration based on the terminology and variable categories used in the study.

). Green energy start-ups are energy-related ventures that the IEA classifies as green or sustainability-oriented. In contrast, digital energy start-ups are energy-related ventures that leverage digital technologies, data-driven solutions, or digital energy services. The terms “green energy start-ups” and “digital energy start-ups” are used consistently throughout the remainder of the article. Funding outcomes are further distinguished by investment stage: early-stage and later-stage. Specifically, the number of energy and green start-ups and energy and digital start-ups are treated as count variables, while funding variables include early-stage and later-stage investment volumes in both green and digital energy segments, measured in current U.S. dollars. The key explanatory variables differ across model specifications. In the first set of models, digitalisation is proxied by the four DESI subindices, while in the second set, the entrepreneurial ecosystem is captured by the GSEI. All models incorporate country- and year-fixed effects to control for unobserved heterogeneity across countries and common time shocks, thereby isolating within-country variation over time.

3.3. Indicator Limitations and Measurement Considerations

Although DESI and GSEI are suitable for cross-country comparative analysis, both indicators should be interpreted with caution. DESI is a composite measure of digital development that provides harmonised information on connectivity, digital public services, human capital, and the integration of digital technologies. However, it does not directly measure sector-specific digital capabilities in the energy sector, such as smart-grid readiness, digital energy platforms, energy data infrastructure, or the digitalisation of electricity markets. In addition, changes in index construction, weighting or data availability may affect comparability over time. Similarly, GSEI captures the general strength and visibility of entrepreneurial ecosystems. However, as a composite ranking-based indicator, it may be influenced by reporting intensity, ecosystem visibility, venture capital concentration, and the overrepresentation of globally visible start-up hubs. Therefore, DESI and GSEI are treated in this study as broad proxies for digitalisation and entrepreneurial ecosystem maturity rather than as direct measures of the specific mechanisms through which digital and ecosystem conditions influence energy-related start-ups.

3.4. Model Specification and Estimation Strategy

Given the distributional properties of the dependent variables, characterised by strong right-skewness, excess zeros, and the presence of extreme values (as documented in

Table A1. Descriptive statistics of the variables used in the analysis.

Variable	Obs.	Mean	Std. Dev.	Median	Min	Max	Skewness	Kurtosis
Energy and green start-ups (Number)	150	11.66	17.69	4.00	0.00	70.00	1.89	2.32
Energy and digital start-ups (Number)	150	6.94	9.97	3.00	0.00	45.00	2.07	3.57
Early-stage funding of energy and green start-ups (USD)	150	41,481,596.45	110,676,975.04	3,231,095.75	0.00	792,449,700.00	4.51	23.15
Later-stage funding of energy and green start-ups (USD)	150	53,404,781.14	293,768,520.65	0.00	0.00	3,397,459,000.00	10.01	109.31
Later-stage funding of energy and digital start-ups (USD)	150	39,601,584.68	285,140,628.16	0.00	0.00	3,376,590,000.00	10.81	122.34
Early-stage funding of energy and digital start-ups (USD)	150	29,901,183.57	80,571,153.92	3,127,198.12	0.00	795,792,600.00	6.32	52.86
Digital Economy and Society Index—Connectivity	150	37.84	12.77	36.01	12.67	77.09	0.68	0.13
Digital Economy and Society Index—Digital Public Services	150	59.47	14.65	59.56	23.98	91.18	−0.08	−0.73
Digital Economy and Society Index—Human Capital	150	46.56	9.06	45.87	29.96	71.39	0.40	−0.40
Digital Economy and Society Index—Integration of Digital Technology	150	30.17	10.27	29.90	10.12	59.09	0.42	−0.04

Note: The sample comprises 150 country–year observations from 25 countries for the period 2017–2022. Funding variables are expressed in current U.S. dollars. Source: authors’ calculations in R.

and

Table A3. Descriptive statistics of variables with GSEI.

Variable	Obs.	Mean	Std. Dev.	Median	Min	Max	Skewness	Kurtosis
Global Startup Ecosystem Index	414	8.49	17.23	5.05	0.10	198.08	7.93	76.00
Energy and green start-ups (Number)	414	10.68	35.45	1.00	0.00	383.00	7.59	67.95
Energy and digital start-ups (Number)	414	5.81	19.37	1.00	0.00	224.00	8.11	77.12
Early-stage funding of energy and green start-ups (USD)	414	135,674,886.82	674,564,200.00	367,199.45	0.00	6,885,232,000.00	7.43	59.01
Later-stage funding of energy and green start-ups (USD)	414	170,847,928.01	1,053,372,000.00	0.00	0.00	15,486,840,000.00	9.90	119.48
Later-stage funding of energy and digital start-ups (USD)	414	72,883,055.74	475,346,000.00	0.00	0.00	5,842,002,000.00	9.57	100.43
Early-stage funding of energy and digital start-ups (USD)	414	58,129,253.41	272,133,700.00	278,134.59	0.00	3,460,145,000.00	8.45	82.57

Notes: The sample includes 414 country–year observations for the period 2019–2023. Funding variables are expressed in current U.S. dollars. Skewness and kurtosis indicate strong deviations from normality and the presence of extreme values. Source: authors’ calculations in R.

), the analysis employs Poisson pseudo-maximum likelihood (PPML) estimators with fixed effects. This approach is well-suited for both count data and non-negative continuous variables with skewed distributions, as it is robust to heteroskedasticity and does not require log-transformation of the dependent variables. Standard errors are clustered at the country level to account for serial correlation and within-country dependence. The baseline specification can be expressed as:

$$E(y_{it}\mid X_{it},{\mu }_i,{\lambda }_t)=exp(\beta X_{it}+{\mu }_i+{\lambda }_t)$$

where $y_{it}$ denotes the number of start-ups or funding level in the country $i$ and year $t$, $X_{it}$ represents the set of explanatory variables (DESI subindices or GSEI), ${\mu }_i$ captures country fixed effects, and ${\lambda }_t$ denotes year fixed effects.

3.5. Robustness, Scaling and Endogeneity Considerations

For funding models, the dependent variables were checked for scaling effects because investment volumes are measured in current U.S. dollars and contain very large values. To improve numerical readability and avoid misleadingly large likelihood values, funding variables were rescaled and reported in millions of U.S. dollars in the funding estimations. This transformation changes the scale of the dependent variable and the magnitude of likelihood-based statistics. However, it does not alter the substantive interpretation of the estimated semi-elasticities for the explanatory variables. The results were also checked against the original USD scale to ensure that the direction and significance of the main coefficients remained stable.

To account for potential dynamic effects and mitigate concerns of reverse causality, lagged specifications are estimated by introducing one-year lagged explanatory variables. This allows the analysis to capture the delayed impacts of digitalisation and ecosystem development on entrepreneurial outcomes. Furthermore, non-linear effects are examined by including squared terms of centred explanatory variables, enabling the identification of potential threshold effects, diminishing returns, or saturation dynamics. The turning point of the quadratic relationship is calculated where both linear and squared terms are statistically significant, providing an estimate of the level at which the marginal effect changes sign. To assess the stability of the results, the empirical strategy includes several robustness-oriented specifications. First, baseline PPML models with country- and year-fixed effects are estimated for each outcome variable. Second, one-year lagged explanatory variables are introduced to examine whether the results are sensitive to temporal ordering and to reduce, although not eliminate, concerns about simultaneity and reverse causality. Third, centred quadratic terms are included to test whether the estimated relationships are linear or exhibit threshold effects, diminishing returns, or saturation dynamics. These specifications allow evaluation of whether the main findings remain consistent across contemporaneous, lagged, and non-linear model forms.

All empirical estimations, data transformations, and statistical calculations are performed using the R statistical environment, primarily employing packages such as fixest for fixed-effects Poisson estimation and dplyr for data preparation. This ensures full reproducibility, transparency, and consistency of the analytical procedures.

This multi-step empirical strategy, combining baseline, lagged, and non-linear specifications, ensures a comprehensive assessment of the relationship between digital development, entrepreneurial ecosystems, and energy-related start-up dynamics, while maintaining robustness to the specific statistical challenges inherent in the data. A further methodological issue concerns potential endogeneity. Reverse causality may arise because stronger entrepreneurial ecosystems can attract higher levels of start-up funding. At the same time, large funding inflows may also improve the measured strength, visibility and ranking position of an ecosystem. A similar issue may arise in the context of digitalisation, as more innovative, start-up-intensive countries may invest more actively in digital infrastructure and digital public services. Country fixed effects reduce bias from time-invariant country characteristics, while year fixed effects account for common shocks. Lagged specifications also help to establish temporal ordering. However, these steps do not provide full causal identification. Therefore, the estimated coefficients should be interpreted as conditional associations rather than definitive causal effects.

4. Results

4.1. Descriptive Statistics and Distributional Properties

The Results section presents the empirical findings in a descriptive and synthesis-oriented manner, while broader theoretical interpretation is developed in the Discussion section. Throughout the section, the estimates are interpreted as conditional associations rather than causal effects.

The descriptive statistics (Table A1, Appendix A) indicate that the panel comprises 150 country–year observations covering the period 2017–2022, which is consistent with a sample of 25 countries observed over six years. Across the dependent variables, the entrepreneurial activity variables display substantial cross-country and intertemporal heterogeneity. On average, the number of energy and green start-ups is 11.66, whereas the corresponding figure for energy and digital start-ups is lower, at 6.94. However, the medians are only 4 and 3, respectively, which suggests that the distributions are strongly right-skewed and that a relatively small number of country–year observations account for a disproportionately large share of start-up activity. This interpretation is confirmed by the positive skewness coefficients of 1.89 and 2.07, as well as by the relatively high maximum values of 70 and 45. Thus, although start-up activity is present across the sample, it is concentrated in a limited number of more entrepreneurial and innovation-intensive contexts.

A similar but even more pronounced pattern emerges for the funding variables. Mean funding values are sizeable, ranging from USD 29.90 million for early-stage funding of energy and digital start-ups to USD 53.40 million for later-stage funding of energy and green start-ups. However, the medians are far lower and, in two cases, equal to zero. In particular, the median values for later-stage funding of both energy and green and energy and digital start-ups are zero, indicating that in at least half of the observations, no later-stage investment was recorded. The very large standard deviations, extreme maximum values, and exceptionally high skewness and kurtosis statistics point to the presence of a small number of very large funding deals. For example, later-stage funding for energy and digital start-ups reaches a maximum of USD 3.37 billion, with skewness and kurtosis of 10.81 and 122.34, respectively. Overall, these statistics imply that funding is highly unevenly distributed across countries and years, with rare but very large investment episodes driving the means upwards. Methodologically, this supports the use of estimators robust to non-normality, zero values, and over-dispersion, such as Poisson pseudo-maximum likelihood for funding and count-based models for start-up numbers.

The DESI subindices exhibit a more regular and substantially less distorted distribution than the entrepreneurial outcome variables. Among them, Digital Public Services records the highest average value (59.47), followed by Human Capital (46.56), Connectivity (37.84), and Integration of Digital Technology (30.17). Their standard deviations, ranging from 9.06 to 14.65, suggest meaningful but not excessive variation across the sample, which is appropriate for comparative panel analysis. In addition, their skewness values remain close to zero, especially for Digital Public Services (−0.08), indicating near-symmetric distributions. The absence of extreme kurtosis further suggests that severe outliers do not dominate the DESI dimensions in the same way as the start-ups and funding variables. Substantively, this means that the digital development indicators provide a relatively stable and well-behaved explanatory framework, while the dependent variables capture highly unequal entrepreneurial and financial outcomes. This contrast reinforces the empirical relevance of examining whether more advanced digital ecosystems are systematically associated with stronger start-up formation and funding performance in the energy domain.

The descriptive statistics of the extended dataset, including the GSEI, are presented in Table A3. The sample comprises 414 country–year observations over the period 2019–2023, providing a broader, more heterogeneous dataset than the DESI-based analysis. The GSEI exhibits substantial variability across countries, with a mean value of 8.49 and a median of 5.05, indicating that most countries have relatively modest ecosystem development. In contrast, a limited number of highly advanced ecosystems drive the upper tail of the distribution. This is further confirmed by the extremely high skewness (7.93) and kurtosis (76.00), suggesting the presence of a few global innovation hubs with exceptionally strong start-up ecosystems. Such distributional characteristics highlight the highly uneven global landscape of entrepreneurial capacity.

A similar pattern of strong right-skewness and concentration is observed for the start-ups variables. The average number of energy and green start-ups is 10.68, and for energy and digital start-ups 5.81; however, the medians for both variables are equal to 1, indicating that in most country–year observations, only very limited start-up activity is present. The extremely high maximum values (383 and 224, respectively), together with skewness above 7 and kurtosis exceeding 65, confirm that entrepreneurial activity is heavily concentrated in a small number of countries. This suggests that global energy-related entrepreneurship is dominated by a limited set of highly innovative ecosystems, while most countries remain at an early stage of development.

The funding variables display even more pronounced asymmetry and dispersion. Mean funding levels are relatively high, ranging from approximately USD 58 million for early-stage digital energy ventures to USD 171 million for later-stage green investments. However, the medians are extremely low and equal to zero for both later-stage variables. This indicates that more than half of the observations report no later-stage funding. The discrepancy between mean and median values, combined with very large standard deviations and extreme maximum values (up to USD 15.5 billion), reveals that funding is highly concentrated in rare but exceptionally large investment events. The very high skewness (up to 9.90) and kurtosis (above 100 in some cases) further confirm the presence of heavy-tailed distributions and substantial outliers.

The statistics reported in Table A3 indicate that both entrepreneurial activity and funding in the energy sector are characterised by strong inequality across countries and time. The distributions of all key variables deviate substantially from normality, with a high prevalence of zeros and extreme values. These features provide strong empirical justification for the use of Poisson pseudo-maximum likelihood estimators with fixed effects in subsequent analysis, as such methods are robust to skewness, heteroskedasticity, and the presence of zero observations.

4.2. DESI and Energy-Related Start-Up Formation

The results of the Poisson fixed-effects estimations are reported in

Table 1. Impact of DESI subindices on energy-related start-up formation (Poisson FE estimates).

Variables	(1) Energy and Green Start-Ups	(2) Energy and Digital Start-Ups
DESI Connectivity	−0.0063 (0.0104)	0.0035 (0.0119)
DESI Digital Public Services	−0.0140 (0.0415)	−0.0286 (0.0428)
DESI Human Capital	−0.0871 * (0.0307)	−0.0464 (0.0540)
DESI Integration of Digital Technology	0.0032 (0.0219)	−0.0263 (0.0276)
Country FE	Yes	Yes
Year FE	Yes	Yes
Observations	150	150
Log-Likelihood	−313.0	−282.9
Adjusted Pseudo R2	0.799	0.687

Notes: The sample covers 25 countries over 2017–2022. The dependent variables are counts of energy and green start-ups and energy and digital start-ups. PPML fixed-effects estimates are reported. Country and year fixed effects are included. Clustered standard errors at the country level are reported in parentheses. Significance codes: * p < 0.1. Source: authors’ calculations in R 4.4.0 (2024-04-24).

. The model for energy and green start-ups demonstrates that, among the four DESI subindices, only human capital exhibits a statistically significant effect. Specifically, the coefficient for human capital is negative and significant at the 1% level (β = −0.087, p < 0.01), indicating that higher levels of digital human capital are associated with fewer energy and green start-ups. In semi-elasticity terms, this suggests that a one-point increase in the human capital index is associated with an approximate decrease of about 8.3% in the expected number of such start-ups, holding other factors constant. This finding may reflect a structural shift in more digitally advanced labour markets, where highly skilled human capital is absorbed by more mature or less environmentally oriented sectors, or where more technologically intensive forms of innovation replace green entrepreneurial activity. A more detailed interpretation is that digital human capital may generate a sectoral allocation effect rather than a direct increase in green start-up formation. In countries with stronger digital skills, highly qualified workers may be absorbed by established technology firms, financial services, consulting, software industries or large energy incumbents, where wages, career stability and scaling opportunities are higher than in nascent green ventures. This can create a crowding-out effect: the same human capital that strengthens the general digital economy may reduce the relative supply of entrepreneurial labour available for early-stage green start-ups. The result may also reflect the structure of advanced labour markets, where digital skills support productivity and innovation inside existing organisations rather than the creation of new green firms. Therefore, the negative coefficient should not be interpreted as evidence that human capital is harmful, but as an indication that digital skills alone do not automatically translate into green entrepreneurship unless they are matched with sector-specific incentives, clean-tech finance and ecosystem support.

In contrast, the remaining DESI components, connectivity, digital public services, and integration of digital technology, do not show statistically significant effects in the green start-ups model. Although the coefficients for connectivity and public services are negative, and integration is slightly positive, all effects are statistically indistinguishable from zero. This suggests that, once country-specific and time-specific effects are controlled for, these dimensions of digital development do not exert an independent influence on the formation of green energy start-ups within the observed period. Nevertheless, the relatively high goodness-of-fit measures (Adjusted Pseudo R² = 0.799; Squared Correlation = 0.959) indicate that the overall model explains a substantial share of variation in start-up activity, largely driven by unobserved country-level heterogeneity and time dynamics.

Turning to energy and digital start-ups, the results reveal no statistically significant relationship between any of the DESI subindices and start-up formation. All estimated coefficients are insignificant and relatively small in magnitude, suggesting that variations in connectivity, digital public services, human capital, and digital integration do not systematically explain differences in the number of energy-related digital start-ups across countries and years. The signs of the coefficients are mixed: connectivity shows a weak positive association, while the remaining indices are negative; however, none of these effects reaches conventional levels of statistical significance. This may indicate that digital entrepreneurship in the energy domain is more strongly influenced by factors such as market conditions, venture capital ecosystems, and regulatory frameworks than by broad digitalisation measures captured by DESI.

The findings presented in Table 1 suggest that the relationship between digital development and energy-related entrepreneurial activity is limited and highly selective. While digital human capital appears to play a statistically significant (yet counterintuitive) role in shaping green start-up formation, the broader digital infrastructure and service environment does not exhibit a robust direct impact. For digital energy start-ups, no meaningful association with DESI subindices is identified, implying that the drivers of such ventures may lie beyond general digital economy indicators. Table 1 indicates that broad digitalisation indicators have limited explanatory power for energy start-up formation, even after controlling for country- and year-fixed effects. The negative human-capital coefficient suggests a possible sectoral allocation mechanism, whereby digitally skilled labour may be absorbed by established technology-intensive sectors rather than nascent green ventures. A fuller theoretical interpretation of this pattern is provided in Section 5.

4.3. DESI and Start-Up Funding Outcomes

The results of the PPML fixed-effects estimations for funding are presented in

Table 2. Impact of DESI subindices on funding of energy-related start-ups (PPML FE estimates).

Variables	(1) Green Early Funding	(2) Green Late Funding	(3) Digital Early Funding	(4) Digital Late Funding
DESI Connectivity	0.0498 * (0.0178)	0.0672 (0.0696)	0.0463 (0.0318)	0.3045 * (0.1575)
DESI Digital Public Services	0.0033 (0.0548)	−0.1683 (0.1462)	0.1023 (0.0915)	−0.0668 (0.3545)
DESI Human Capital	0.0014 (0.1048)	0.0967 (0.2104)	0.1333 (0.1033)	0.4347 (0.5317)
DESI Integration of Digital Technology	−0.0614 (0.0568)	0.1368 (0.1153)	−0.0166 (0.0580)	0.1954 (0.1313)
Country FE	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes
Observations	144	90	144	84
Countries	24	15	24	14
Adjusted Pseudo R2	0.924	0.901	0.854	0.912

Notes: PPML fixed-effects estimates are reported. The dependent variables are funding volumes of energy-related start-ups. To improve numerical readability, funding variables should be expressed in million current U.S. dollars in the reported funding models. Country and year fixed effects are included in all specifications. Clustered standard errors at the country level are reported in parentheses. The number of countries differs across models because of missing or unusable observations for some funding outcomes under fixed-effects PPML estimation. The log-likelihood row is omitted because likelihood values are scale-sensitive and were extremely large when funding variables were measured in current U.S. dollars. Significance codes: * p < 0.1. Source: authors’ calculations in R.

. For early-stage funding of energy and green start-ups, the findings reveal a statistically significant and positive effect of digital connectivity. The estimated coefficient (β = 0.0498, p < 0.01) indicates that improvements in digital infrastructure are associated with higher levels of early-stage investment. In semi-elasticity terms, a one-point increase in the connectivity index corresponds to an approximate 5.1% increase in expected funding, holding other factors constant. This suggests that well-developed digital infrastructure, associated with access to financing at early stages, likely reduces information frictions and improves investor–start-up matching. In contrast, the remaining DESI dimensions, digital public services, human capital, and integration of digital technology do not exhibit statistically significant effects, indicating that their independent contribution to early-stage green funding is limited once country and time effects are controlled for.

In Table 2, the log-likelihood values are not reported because the funding variables are measured on a very large monetary scale, and likelihood statistics are therefore scale-sensitive; the interpretation instead focuses on coefficient signs, statistical significance, fixed-effects structure, sample size, and pseudo R².

For later-stage funding of energy and green start-ups, no statistically significant relationships are identified between DESI subindices and funding levels. Although the coefficients vary in sign, connectivity and integration are positive, while public services are negative; none of these effects reaches conventional levels of significance. This lack of statistical significance may be partly explained by the smaller sample size (90 observations across 15 countries) and the highly uneven distribution of later-stage investments, which are characterised by infrequent but very large deals. Consequently, later-stage green funding appears to be driven more by idiosyncratic factors, such as large-scale investment cycles or country-specific financial ecosystems, rather than by general levels of digitalisation.

A similar pattern is observed for early-stage funding of energy and digital start-ups, in which none of the DESI subindices shows statistically significant effects. Although all coefficients except integration are positive, suggesting a potentially favourable association between digital development and early-stage digital energy investments, these relationships are not robust. This implies that, within the analysed period, variations in digital infrastructure, public services, and human capital do not systematically translate into higher early-stage funding for digital energy ventures. Other determinants, such as venture capital availability, market maturity, or sector-specific dynamics, are likely to play a more decisive role.

In contrast, the model for later-stage funding of energy and digital start-ups provides weak evidence of a positive association with digital connectivity. The coefficient for connectivity is positive and marginally significant at the 10% level (β = 0.305, p < 0.1), suggesting that more advanced digital infrastructure may support larger-scale investments at later stages. The implied effect size is substantial: a one-point increase in connectivity is associated with an approximate 35.6% increase in expected funding. However, given the borderline significance and relatively small sample (84 observations across 14 countries), this result should be interpreted with caution. The remaining DESI components are not statistically significant, although the positive coefficients for human capital and integration may indicate potential, yet imprecisely estimated, relationships.

Table 2 suggests that digital connectivity is more relevant for funding mobilisation than for start-up creation. This may reflect lower information frictions, better investor–start-ups communication and improved visibility of young ventures. However, the weak effects of other DESI dimensions indicate that energy start-up funding depends on more sector-specific conditions, such as venture capital depth, clean-tech policy support, technology readiness, and regulatory certainty. These mechanisms are discussed further in Section 5.

4.4. Lagged and Non-Linear DESI Specifications

The results of the lagged specifications are reported in

Table 3. Lagged effects of DESI subindices on start-ups and funding (PPML FE estimates).

Variables	(1) Green Start-Ups	(2) Digital Start-Ups	(3) Green Early Funding	(4) Green Late Funding	(5) Digital Early Funding	(6) Digital Late Funding
Lagged DESI Connectivity	−0.0063 (0.0104)	0.0035 (0.0118)	0.0498 * (0.0178)	0.0672 (0.0696)	0.0463 (0.0318)	0.3045 * (0.1575)
Lagged DESI Digital Public Services	−0.0140 (0.0415)	−0.0286 (0.0428)	0.0033 (0.0548)	−0.1683 (0.1462)	0.1023 (0.0915)	−0.0668 (0.3545)
Lagged DESI Human Capital	−0.0871 * (0.0307)	−0.0464 (0.0540)	0.0014 (0.1048)	0.0967 (0.2104)	0.1333 (0.1033)	0.4347 (0.5317)
Lagged DESI Integration of Digital Technology	0.0032 (0.0219)	−0.0263 (0.0276)	−0.0614 (0.0568)	0.1368 (0.1153)	−0.0166 (0.0580)	0.1954 (0.1313)
Country FE	Yes	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes	Yes
Observations	150	150	144	90	144	84
Pseudo R2	0.818	0.720	0.924	0.901	0.854	0.912

Notes: Clustered standard errors (by country) are reported in parentheses. Significance codes: * p < 0.1. Source: authors’ calculations in R.

, where the DESI subindices are introduced with a one-year lag to account for the delayed effects of digital development on entrepreneurial activity and funding. For start-up formation, the findings remain largely consistent with the baseline models. In particular, lagged human capital continues to exhibit a negative and statistically significant effect on the number of energy and green start-ups (β = −0.087, p < 0.01), suggesting that higher levels of digital skills in the previous period are associated with a lower number of such ventures in the current period. In semi-elasticity terms, a one-point increase in the human capital index in year t − 1 corresponds to an approximate 8.3% decrease in green start-up activity in year t. This persistent negative effect reinforces the interpretation that more advanced digital labour markets may reallocate entrepreneurial effort away from green start-ups towards other, potentially more technology-intensive sectors. In contrast, none of the lagged DESI subindices shows statistically significant effects on energy and digital start-ups, indicating that digital development does not exert a delayed influence on this segment of entrepreneurship.

Turning to funding, the lagged results highlight a more nuanced pattern. For early-stage funding of energy and green start-ups, lagged connectivity remains positive and statistically significant at the 1% level (β = 0.0498, p < 0.01). This implies that improvements in digital infrastructure in the previous year are associated with an approximate 5.1% increase in early-stage funding in the current year. This finding is robust and consistent with the baseline model, providing strong evidence that digital connectivity is associated with early-stage investment processes, likely by improving information flows and reducing transaction costs. By contrast, the remaining DESI dimensions (public services, human capital, and integration) remain statistically insignificant, suggesting that their effects on early-stage green funding are neither immediate nor delayed.

For later-stage funding of energy and green start-ups, no statistically significant lagged effects are observed, consistent with earlier findings and likely reflecting the episodic and highly concentrated nature of large-scale investments. Similarly, for early-stage funding of energy and digital start-ups, none of the lagged DESI variables is statistically significant. However, the positive signs for connectivity, public services, and human capital may indicate weak underlying relationships that are not precisely estimated. In the case of later-stage funding of energy and digital start-ups, lagged connectivity shows a marginally significant positive effect at the 10% level (β = 0.305, p < 0.1), suggesting that improvements in digital infrastructure may translate into larger investment volumes with a delay. The implied magnitude is substantial: a one-point increase in connectivity is associated with an approximate 35.6% increase in expected funding, though this result should be interpreted cautiously given the relatively small sample size.

The evidence presented in Table 3 confirms that the impact of digitalisation on energy-related entrepreneurship and funding is both limited and dimension-specific. The only robust and consistent finding is the positive role of digital connectivity in supporting early-stage green funding, including with a temporal lag. At the same time, the negative association between human capital and green start-up formation persists even after lagging, suggesting a structural rather than a short-term relationship. Other dimensions of digital development do not exhibit significant delayed effects, indicating that broader digital ecosystem improvements do not automatically translate into increased entrepreneurial activity or investment in the energy sector.

The DESI-based results suggest that general digital development is not a sufficient standalone driver of energy-related entrepreneurship. The weak and mostly insignificant effects of digital public services, human capital, and the integration of digital technology may indicate that broad national digitalisation indicators do not capture the sector-specific digital capabilities needed for energy start-up creation, such as smart-grid infrastructure, energy data platforms, digital energy market design, or investor access to specialised clean-tech information. In addition, the inclusion of country- and year-fixed effects means the estimates rely primarily on within-country changes over time rather than on cross-country differences in digital maturity. Therefore, the absence of significant DESI effects should not be interpreted as evidence that digitalisation is irrelevant, but rather as evidence that general digitalisation indicators have limited explanatory power once structural country differences and common time shocks are controlled for.

The results of the non-linear specifications are reported in

Table A2. Non-linear effects of DESI subindices on start-ups and funding (PPML FE estimates).

Variables	(1) Green Start-Ups	(2) Digital Start-Ups	(3) Green Early Funding	(4) Green Late Funding	(5) Digital Early Funding	(6) Digital Late Funding
Connectivity (centred)	−0.0197 (0.0129)	0.0047 (0.0136)	−0.0072 (0.0397)	0.0189 (0.0634)	−0.0078 (0.0601)	0.2217 (0.1458)
Connectivity2	0.0005 (0.0004)	−0.00008 (0.0003)	0.0013 (0.0010)	0.0011 (0.0020)	0.0013 (0.0015)	−0.0004 (0.0059)
Digital Public Services (centred)	0.0451 (0.0511)	0.0257 (0.0596)	−0.0241 (0.1197)	−0.0057 (0.3969)	0.1380 (0.1608)	−1.334 (1.212)
Digital Public Services2	−0.0011 * (0.0006)	−0.0009 (0.0006)	−0.00004 (0.0011)	0.0098 (0.0065)	0.0007 (0.0016)	0.0362 (0.0274)
Human Capital (centred)	−0.0977 (0.0614)	−0.0364 (0.0722)	−0.1581 (0.1459)	0.3859 (0.5870)	0.1718 (0.2061)	−0.5499 (0.5322)
Human Capital2	0.0020 (0.0019)	0.0025 (0.0028)	0.0064 (0.0091)	−0.0597 * (0.0322)	0.0003 (0.0096)	−0.0625 (0.0476)
Integration (centred)	−0.0274 (0.0446)	−0.0339 (0.0532)	−0.0428 (0.1359)	−0.0666 (0.4364)	0.1585 (0.1491)	−1.118 (0.8930)
Integration2	0.0006 (0.0012)	0.00006 (0.0012)	−0.0021 (0.0031)	0.0068 (0.0065)	−0.0046 (0.0039)	0.0084 (0.0096)
Country FE	Yes	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes	Yes
Observations	150	150	144	90	144	84
Pseudo R2	0.820	0.722	0.928	0.915	0.863	0.933

Notes: The symbol “²” denotes the squared term of the corresponding explanatory variable and is used to test non-linear effects. Clustered standard errors (by country) are reported in parentheses. Significance codes: * p < 0.1. Source: authors’ calculations in R.

, where quadratic terms of the centred DESI subindices are introduced to test for potential non-linear relationships. Overall, the findings provide limited evidence of systematic non-linear effects across both start-up formation and funding variables. For energy and green start-ups as well as energy and digital start-ups, none of the linear or squared terms reaches conventional levels of statistical significance. Although some coefficients suggest potential curvature, for instance, a negative linear and positive squared term for connectivity in the green start-ups model, which would be consistent with a U-shaped pattern, these effects are not statistically robust. Similarly, for digital public services, the squared term is marginally significant (at the 10% level) and negative in the green start-ups model, indicating a weak indication of an inverted U-shaped relationship; however, the lack of significance of the linear term limits the strength of this interpretation. Overall, the evidence suggests that the relationship between digital development and start-up formation remains largely linear or statistically indistinguishable from zero after accounting for nonlinearity.

For funding variables, the results are likewise characterised by a general absence of robust non-linear patterns. In the case of early-stage funding of energy and green start-ups, none of the quadratic terms is statistically significant, indicating that the previously identified positive effect of connectivity does not exhibit meaningful curvature. For later-stage funding of energy and green start-ups, there is weak evidence of non-linearity in the human capital dimension, where the squared term is marginally significant and negative (p < 0.1). Combined with a positive linear coefficient, this suggests a potential inverted U-shaped relationship, implying that increases in digital human capital may initially support funding. Still, the marginal effect diminishes at higher levels. However, given the relatively small sample size and marginal significance, this result should be interpreted cautiously.

For digital energy funding, neither early-stage nor later-stage models reveal statistically significant nonlinear effects across the DESI subindices. Although some coefficients, such as the negative linear and positive squared terms for public services or integration, hint at possible curvature, these patterns are not statistically supported. In the later-stage digital funding model, the magnitude of several coefficients is large, particularly for public services and integration. Still, the corresponding standard errors are substantial, reflecting high variability and limited precision in the estimates. This is consistent with the earlier observation that funding variables, especially at later stages, are highly volatile and driven by a small number of large investment events.

The results presented in Table A2 suggest that introducing quadratic terms does not materially alter the main conclusions of the analysis. There is no strong or consistent evidence that the impact of digitalisation on energy-related entrepreneurship and funding follows a non-linear pattern. Instead, the effects of DESI subindices appear either linear or statistically insignificant across most specifications. These findings reinforce the robustness of the baseline models and indicate that, within the observed range of digital development, threshold effects or diminishing returns are not a dominant feature of the relationship.

4.5. GSEI and Energy-Related Start-Up Formation and Funding

Since the GSEI-based analysis relies on a broader and more heterogeneous country sample than the DESI-based analysis, the results should be interpreted as a complementary global extension rather than a direct country-by-country robustness check. Similarities between the DESI- and GSEI-based results indicate qualitative consistency across different empirical settings. In contrast, differences may reflect period effects, sample expansion, country composition or the conceptual differences between the indicators. The preceding DESI results suggest that digitalisation acts mainly as a selective enabling condition: digital infrastructure may support specific investment processes, especially early-stage green funding, but broader digitalisation does not automatically generate new energy ventures or sustained funding flows.

The estimation results for the GSEI are presented in

Table 4. Impact of GSEI on energy-related start-ups and funding (PPML FE estimates).

Variables	(1) Green Start-Ups	(2) Digital Start-Ups	(3) Green Early Funding	(4) Green Late Funding	(5) Digital Early Funding	(6) Digital Late Funding
GSEI	0.0016 (0.0006)	0.0006 (0.0006)	−0.0037 * (0.0009)	−0.0006 (0.0026)	−0.0035 * (0.0014)	0.0073 * (0.0035)
Country FE	Yes	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes	Yes
Observations	374	350	319	165	310	130
Pseudo R2	0.911	0.854	0.968	0.948	0.933	0.927

Notes: The sample covers the GSEI-based panel for 2019–2023. The number of countries differs across models because of differences in data availability and the fixed-effects PPML estimation sample. Funding variables are expressed in million current U.S. dollars. PPML fixed-effects estimates are reported. Country and year fixed effects are included. Clustered standard errors at the country level are reported in parentheses. Significance codes: * p < 0.1. Source: authors’ calculations in R.

and provide more consistent and economically meaningful patterns than those from the digitalisation-based specifications. For start-up formation, the results indicate a positive and statistically significant relationship between the strength of the entrepreneurial ecosystem and the number of energy and green start-ups. The coefficient for GSEI is positive and significant at the 5% level (β = 0.0016, p < 0.05), indicating that improvements in the ecosystem are associated with increased green entrepreneurial activity. In semi-elasticity terms, a one-unit increase in GSEI corresponds to an approximate 0.16% increase in the expected number of green start-ups. Although the magnitude appears modest, it is important to note the large variation in GSEI across countries, which implies potentially substantial cumulative effects. In contrast, no statistically significant effect is found for energy and digital start-ups, suggesting that ecosystem strength does not systematically translate into higher entry rates in these segments.

For funding variables, the results reveal a differentiated pattern across stages and investment types. In the case of early-stage funding for energy and green start-ups, GSEI exhibits a negative, highly significant effect (β = −0.0037, p < 0.01), indicating that stronger ecosystems are associated with lower levels of early-stage green investment. A similar negative relationship is observed for early-stage funding of energy and digital start-ups (β = −0.0035, p < 0.1). This counterintuitive result may reflect the structural characteristics of more advanced ecosystems, in which investment tends to concentrate in later-stage ventures rather than in seed or early-stage financing. In such environments, mature start-ups dominate funding flows, potentially crowding out early-stage opportunities.

The results for later-stage funding support this interpretation. While the coefficient for later-stage funding of energy and green start-ups is not statistically significant, the effect becomes positive and marginally significant for later-stage funding of energy and digital start-ups (β = 0.0073, p < 0.1). This suggests that stronger entrepreneurial ecosystems are associated with larger-scale investments at more advanced stages, particularly in digitally oriented energy ventures. The magnitude of the coefficient implies that a one-unit increase in GSEI is associated with approximately a 0.73% increase in expected later-stage funding, indicating a meaningful amplification effect at higher levels of ecosystem development.

The findings reported in Table 4 suggest that the entrepreneurial ecosystem plays a more direct and stage-specific role in shaping investment dynamics than in influencing start-up entry. While stronger ecosystems support the formation of green start-ups, their primary impact appears to be the reallocation of funding towards later-stage ventures, particularly in the digital energy domain. This highlights the importance of distinguishing between early-stage and growth-stage financing when analysing the effects of ecosystem development.

4.6. Lagged and Non-Linear GSEI Specifications

The results of the lagged specifications for the GSEI are reported in

Table 5. Lagged effects of GSEI on energy-related start-ups and funding (PPML FE estimates).

Variables	(1) Green Start-Ups	(2) Digital Start-Ups	(3) Green Early Funding	(4) Green Late Funding	(5) Digital Early Funding	(6) Digital Late Funding
Lagged GSEI	0.0016 (0.0006)	0.0006 (0.0006)	−0.0037 * (0.0009)	−0.0006 (0.0026)	−0.0035 * (0.0014)	0.0073 * (0.0035)
Country FE	Yes	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes	Yes
Observations	374	350	319	165	310	130
Pseudo R2	0.911	0.854	0.968	0.948	0.933	0.927

Notes: The sample covers the GSEI-based panel for 2019–2023. The number of countries differs across models because of differences in data availability and the fixed-effects PPML estimation sample. Funding variables are expressed in million current U.S. dollars. PPML fixed-effects estimates are reported. Country and year fixed effects are included. Clustered standard errors at the country level are reported in parentheses. Significance codes: * p < 0.1. Source: authors’ calculations in R.

. Overall, the findings closely mirror the baseline estimates, indicating that the effects of the entrepreneurial ecosystem are not only contemporaneous but also persistent over time. For start-up formation, lagged GSEI has a positive and statistically significant effect on the number of energy and green start-ups (β = 0.0016, p < 0.05). This indicates that improvements in the entrepreneurial ecosystem in the previous year are associated with an increase in green start-up activity in the current period. In semi-elasticity terms, a one-unit increase in GSEI in year t − 1 corresponds to approximately a 0.16% increase in the expected number of green start-ups in year t. In contrast, no statistically significant relationship is observed for energy and digital start-ups, suggesting that ecosystem effects on this segment remain weak or indirect even when accounting for temporal dynamics.

For funding variables, the lagged results reveal a clear and consistent pattern of stage-specific effects. Lagged GSEI is negatively and highly significantly associated with early-stage funding of energy and green start-ups (β = −0.0037, p < 0.01), and also negatively related to early-stage funding of energy and digital start-ups (β = −0.0035, p < 0.1). These findings indicate that more developed entrepreneurial ecosystems tend to allocate relatively fewer resources to early-stage investments in the subsequent period. This may reflect a structural shift in mature ecosystems, in which capital is increasingly directed toward scaling established ventures rather than supporting nascent firms.

The results for later-stage funding reinforce this interpretation. While no significant effect is found for later-stage funding of energy and green start-ups, lagged GSEI shows a positive, marginally significant effect on later-stage funding of energy and digital start-ups (β = 0.0073, p < 0.1). This suggests that stronger ecosystems are associated with larger-scale investments at more advanced stages, with effects materialising over time. The magnitude implies that a one-unit increase in GSEI in the previous year is associated with an approximate 0.73% increase in later-stage funding for digital energy ventures in the current year.

The results presented in Table 5 suggest that the entrepreneurial ecosystem exerts a persistent and structurally differentiated influence on energy-related innovation. While it supports the formation of green start-ups over time, its primary impact lies in shaping the allocation of financial resources, shifting investment from early-stage to later-stage ventures, particularly in digitally oriented segments. The consistency between contemporaneous and lagged results strengthens the robustness of the findings, but the estimates should still be interpreted as conditional associations rather than definitive causal effects. Substantively, the lagged models suggest that ecosystem maturity is associated less with the immediate expansion of all energy start-up activity and more with a structural reallocation of financial resources. More developed ecosystems appear to support green start-up formation while directing capital increasingly towards later-stage and digitally oriented ventures. This pattern is consistent with the logic of ecosystem maturation, in which investors become more selective and prioritise scalable firms over nascent ventures.

The results of the non-linear specifications for the GSEI are reported in

Table 6. Non-linear effects of GSEI on energy-related start-ups and funding (PPML FE estimates).

Variables	(1) Green Start-Ups	(2) Digital Start-Ups	(3) Green Early Funding	(4) Green Late Funding	(5) Digital Early Funding	(6) Digital Late Funding
GSEI (centred)	0.0021 (0.0016)	0.0011 (0.0017)	−0.0081 (0.0050)	0.0140 (0.0059)	−0.0101 * (0.0054)	−0.0108 (0.0078)
GSEI2	−0.000003 (0.0000065)	−0.000002 (0.0000066)	0.000017 (0.0000186)	−0.0000556 (0.0000243)	0.000025 (0.0000237)	0.0000695 * (0.0000416)
Country FE	Yes	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes	Yes
Observations	374	350	319	165	310	130
Pseudo R2	0.911	0.854	0.968	0.950	0.933	0.931

Notes: The sample covers the GSEI-based panel for 2019–2023. The number of countries differs across models because of differences in data availability and the fixed-effects PPML estimation sample. Funding variables are expressed in million current U.S. dollars. PPML fixed-effects estimates are reported. Country and year fixed effects are included. Clustered standard errors at the country level are reported in parentheses. Significance codes: * p < 0.1. Source: authors’ calculations in R.

. Overall, the evidence suggests that non-linear effects are limited and largely confined to specific funding dimensions, while start-up formation remains predominantly linear. For both energy and green start-ups and energy and digital start-ups, neither the linear nor the quadratic terms of GSEI are statistically significant. Although the signs of the coefficients (positive linear and negative squared terms) may suggest a weak inverted U-shaped pattern, the lack of statistical significance indicates no robust evidence of non-linear ecosystem effects on start-up entry. This implies that the relationship between ecosystem development and start-up formation is either weak or approximately linear within the observed range of GSEI.

These non-linear findings indicate that ecosystem development has different meanings at different stages of maturity. At lower and intermediate levels of ecosystem development, improvements in GSEI may reduce market frictions, strengthen investor networks, and increase firms’ capacity to scale. At very high levels of ecosystem maturity, however, additional ecosystem strength may generate diminishing marginal benefits because funding becomes concentrated in already established firms, competition for capital increases, and investors may shift towards alternative high-growth sectors. Therefore, the main contribution of the non-linear results is not only the identification of a turning point, but also the demonstration that ecosystem quality can produce both enabling and concentration effects. Although graphical presentation of the marginal effects would be useful, the non-linear interpretation is supported numerically by the positive linear and negative squared GSEI terms for later-stage green funding, as well as by the estimated turning point located within the observed GSEI range. This indicates that the inverted U-shaped relationship is not only a formal statistical result but also an economically interpretable pattern within the empirical sample.

For early-stage funding variables, the results similarly do not provide strong support for non-linear effects. In the case of early-stage funding of energy and green start-ups, both the linear and squared terms are statistically insignificant. However, the negative linear coefficient is consistent with earlier findings that early-stage funding declines as ecosystems mature. For early-stage funding of energy and digital start-ups, there is weak evidence of a negative linear effect (marginally significant at the 10% level). Still, the squared term remains insignificant, suggesting a primarily linear relationship rather than non-linearity. These results reinforce the interpretation that more developed ecosystems tend to shift investment away from early-stage ventures without exhibiting clear threshold effects.

In contrast, some evidence of non-linearity emerges for later-stage funding of energy and green start-ups. The linear term is positive and significant (β = 0.0140, p < 0.05), while the squared term is negative and also statistically significant (β = −0.0000556, p < 0.05), indicating an inverted U-shaped relationship. This suggests that improvements in the entrepreneurial ecosystem initially stimulate later-stage green funding, but the marginal effect diminishes as the ecosystem develops. In other words, beyond a certain threshold, additional improvements in GSEI yield decreasing returns on funding growth. This pattern is consistent with the notion of ecosystem saturation, where highly developed markets reach a stage of maturity with slower incremental gains.

For later-stage funding of energy and digital start-ups, the squared term is marginally significant and positive (p < 0.1), while the linear term is negative but insignificant. This combination may indicate a weak U-shaped relationship, implying that ecosystem development initially suppresses funding but becomes beneficial beyond a certain threshold. However, given the lack of significance of the linear term and the relatively small sample size, this result should be interpreted cautiously. Overall, the findings in Table 6 suggest that non-linear effects of ecosystem development are not pervasive but may emerge in specific contexts, particularly in later-stage funding, where diminishing or threshold effects are more likely.

The estimated turning point for the non-linear relationship between the GSEI and later-stage funding of energy and green start-ups provides important insights into the nature of ecosystem effects. Based on the quadratic specification, the turning point is calculated at approximately 133.9 points on the GSEI scale, after transforming back from the centred variable. This value lies well within the observed data range (0.104 to 198.08), confirming that the estimated non-linear relationship is economically meaningful and empirically relevant.

The results indicate an inverted U-shaped relationship between ecosystem development and later-stage green funding. At lower and intermediate levels of GSEI, improvements in the entrepreneurial ecosystem are associated with increases in later-stage funding. This reflects the role of stronger ecosystems in facilitating access to capital, improving investor networks, and supporting the scaling of energy-related ventures. However, once the ecosystem reaches a high level of development, around the 133.9-point threshold, the marginal effect becomes negative. This suggests that further improvements in already highly developed ecosystems yield diminishing returns in terms of additional funding for green start-ups.

Substantively, this pattern may reflect saturation effects in the ecosystem. In highly advanced innovation hubs, investment activity may become concentrated in a limited number of large-scale or highly mature ventures, reducing the relative expansion of funding across the broader set of green start-ups. Additionally, competition for capital, sectoral specialisation, or a shift towards alternative high-growth industries may limit further increases in green investment. Therefore, while ecosystem development is crucial for stimulating funding at earlier stages, its impact becomes less pronounced and potentially constrained at the highest levels of ecosystem maturity.

The presence of a statistically significant turning point within the observed data range strengthens the validity of the non-linear specification. It highlights the importance of considering threshold effects when analysing the relationship between entrepreneurial ecosystems and green investment dynamics.

4.7. Synthesis of Empirical Patterns

To avoid interpreting the results solely as separate coefficient estimates,

Table 7. Synthesis of the main empirical patterns and analytical interpretation.

Empirical Pattern	Analytical Interpretation	Implication for Hypotheses
DESI indicators show limited and selective effects on start-up formation and funding.	General digitalisation does not automatically translate into energy start-up creation; sector-specific digital capabilities may matter more than broad national digital indicators.	Partially supports H1.
Digital connectivity is positively associated with early-stage green funding.	Connectivity may reduce information frictions and improve investor–start-up matching at early investment stages.	Supports the funding dimension of H1.
GSEI is positively associated with the formation of green start-ups.	Entrepreneurial ecosystem maturity supports venture creation in green energy segments.	Supports H2.
GSEI is negatively associated with early-stage funding but positively associated with later-stage digital funding.	Mature ecosystems may reallocate capital from nascent ventures to more established, scalable firms.	Supports the stage-specific part of H2.
Lagged models are broadly consistent with baseline models.	The main patterns are not purely contemporaneous, although they should still be interpreted as associations rather than causal effects.	Supports temporal robustness under H3.
Nonlinear effects are limited in DESI but present in later-stage green funding in the GSEI models.	Digitalisation primarily acts as an enabling condition, whereas ecosystem maturity may lead to saturation and diminishing returns.	Supports the non-linear part of H3.

Source: Authors’ own synthesis based on the empirical results reported in Table 1, Table 2, Table 3, Table 4, Table 5 and Table 6.

summarises the main empirical patterns, their analytical interpretations, and their relationships to the hypotheses. This synthesis shows that the findings are not simply a collection of significant and insignificant coefficients, but point to a broader mechanism: digitalisation provides selective enabling conditions, whereas entrepreneurial ecosystem maturity shapes both start-up formation and the allocation of funding across development stages.

5. Discussion

The findings of this study provide a nuanced perspective on the role of digitalisation in shaping energy-related entrepreneurship, suggesting that its impact is more limited and context-dependent than often assumed in the literature. While prior research highlights the importance of digital technologies for enhancing innovation capacity and start-up performance [28,30,33], the empirical results indicate that most DESI dimensions do not exert statistically significant effects on start-up formation or funding. The only robust effect emerges for human capital, which is negatively associated with green start-up formation (β = −0.087, p < 0.01), suggesting potential saturation or structural mismatches in highly developed digital labour markets. This finding contrasts with studies emphasising the positive role of digital maturity [32] and knowledge capabilities [31]. It implies that digitalisation alone is insufficient to stimulate entrepreneurial activity without complementary ecosystem conditions. In particular, the negative human-capital effect may indicate that digitally skilled labour is allocated to established high-productivity sectors or incumbent firms, producing a crowding-out mechanism that weakens the formation of new green ventures unless targeted clean-tech entrepreneurship incentives are in place. This result is important because it challenges a common assumption in the digital entrepreneurship literature: that higher levels of digital skills automatically support the creation of start-ups. In the context of energy-related entrepreneurship, the findings suggest a more conditional mechanism. Digital human capital may strengthen innovation capacity in the general economy, but it does not necessarily flow into green start-up formation. Instead, skilled labour may be absorbed by established technology-intensive sectors, incumbent energy firms, financial services or software industries. This divergence from studies emphasising the positive role of digital maturity indicates that the effect of human capital depends on sectoral allocation, labour-market incentives and the presence of specialised clean-tech entrepreneurial opportunities.

In contrast, the results strongly support the literature emphasising the central role of entrepreneurial ecosystems in driving start-up dynamics. The positive effect of ecosystem strength on green start-up formation (β = 0.0016, p < 0.05) aligns with evidence highlighting the importance of institutional support, networks, and access to resources for entrepreneurial success [35,36,38]. At the same time, the observed negative relationship between ecosystem development and early-stage funding (β = −0.0037, p < 0.01) is consistent with findings on capital concentration and increasing selectivity in mature ecosystems [58,59]. This suggests that as ecosystems evolve, financial resources become more focused on scaling established ventures rather than supporting nascent firms, reflecting structural shifts in investment behaviour. Moreover, the positive effect of digital energy start-up ecosystem maturity on later-stage funding (β = 0.0073, p < 0.1) supports the argument that ecosystem maturity enhances scaling opportunities and enables access to larger investment rounds [37,47]. The negative association between ecosystem maturity and early-stage funding also requires critical interpretation. While entrepreneurial ecosystem theory often assumes that stronger ecosystems improve access to finance for all start-ups, the results suggest that mature ecosystems may become more selective and capital-concentrating. In such environments, investors may prefer ventures with proven scalability, stronger market traction and lower technological uncertainty, which can reduce the relative availability of seed and early-stage finance for younger green and digital energy firms. Therefore, ecosystem maturity should not be understood only as an unconditional advantage. It may simultaneously support scaling and create barriers for nascent ventures if financial resources become concentrated around already visible or later-stage firms.

The non-linear results further extend the existing literature by demonstrating that the relationship between ecosystem development and funding is not monotonic but characterised by threshold effects. The identified inverted U-shaped relationship for later-stage green funding, with a turning point at approximately 133.9 GSEI points, indicates diminishing returns in highly developed ecosystems. It also complements studies emphasising the importance of balanced policy frameworks and efficient allocation of resources in supporting sustainable innovation [13,14]. The results suggest that while digitalisation provides necessary technological infrastructure, it is the quality and maturity of entrepreneurial ecosystems that primarily determine both the quantity and structure of energy-related start-up activity and investment. The interpretation of parameter stability is further complicated by the fact that both analysed periods overlap with the COVID-19 pandemic. COVID-related disruptions may have affected start-up formation, venture capital allocation, digitalisation processes, and energy-sector investment by altering uncertainty, public support, investor risk appetite, and sectoral demand. Although year fixed effects partially account for common annual shocks, they cannot fully capture heterogeneous country-level responses to the pandemic or differences in recovery trajectories. Thus, the observed similarities or differences across specifications may reflect a combination of underlying economic relationships, period-specific shocks, and changes in sample composition.

The theoretical contribution of the study lies in showing that digitalisation and entrepreneurial ecosystems operate through different mechanisms in energy-related entrepreneurship. Digitalisation appears mainly as a selective enabling condition: connectivity can support funding by reducing information frictions, but broader digital indicators do not automatically generate start-up formation. Entrepreneurial ecosystem maturity, by contrast, acts as an allocation mechanism, supporting green start-up formation and later-stage digital funding. However, it may also redirect capital away from early-stage ventures. The non-linear evidence further extends ecosystem theory by showing that ecosystem development can produce diminishing returns at high levels of maturity. Thus, the study contributes to sustainability entrepreneurship research by moving beyond a simple “more digitalisation and stronger ecosystems are always better” interpretation and showing that their effects are conditional, stage-specific and potentially asymmetric.

Despite providing new cross-country evidence on the role of digitalisation and entrepreneurial ecosystems in shaping energy-related start-ups and their funding, this study has several limitations. First, the analysis relies on aggregated country-level data, which may obscure heterogeneity across regions, cities and specific technological niches within the energy sector. Second, the time span remains relatively short, limiting the ability to capture longer-term structural changes in digitalisation, entrepreneurial ecosystems, and energy investment. Third, DESI and GSEI are composite indicators. While they are useful for cross-country comparison, they do not capture all sector-specific mechanisms relevant to energy start-ups. They may be affected by changes in methodology, data availability, ranking visibility or venture-capital concentration. Fourth, the two empirical samples differ in both period coverage and country composition, which limits direct comparability between the DESI- and GSEI-based estimates. A further limitation is that the graphical presentation of the results is limited. In particular, marginal-effects plots for the non-linear GSEI specifications and histograms of funding variables would further improve the visual interpretation of the estimated relationships and the distributional properties of the data. Future versions of the analysis could include graphical diagnostics. Finally, potential endogeneity cannot be fully excluded. Stronger ecosystems may attract investment, but large investment flows may also strengthen ecosystem visibility and performance. Although country and year fixed effects, lagged specifications and non-linear models improve the robustness of the analysis, the empirical design should be interpreted as an observational panel-data approach rather than a causal identification strategy. Future research could address these limitations by using firm-level or regional data, longer time series, instrumental variables, quasi-experimental designs or alternative measures of sector-specific digital and ecosystem conditions.

6. Conclusions

The study set out to examine how digital development and the strength of entrepreneurial ecosystems shape both the formation and financing of energy-related start-ups, with particular attention to differences across investment stages, time dynamics, and potential non-linear effects. In other words, the research aimed to identify whether general digitalisation or broader ecosystem conditions play a more decisive role in fostering energy entrepreneurship and attracting investment.

The empirical analysis relies on a cross-country panel dataset combining information from the European Commission (DESI indicators), the International Energy Agency (energy start-up activity and funding), and StartupBlink (Global Startup Ecosystem Index). Two complementary samples were analysed, covering European countries and a broader global set of economies over the period 2017–2023. The study employs Poisson pseudo-maximum likelihood models with country- and year-fixed effects, complemented by lagged and quadratic specifications to capture delayed and nonlinear relationships. All estimations were conducted in the R statistical environment with clustered standard errors at the country level.

The results indicate that digitalisation, as measured by DESI, has limited and inconsistent effects on energy-related entrepreneurship, with only human capital showing a negative and statistically significant association with green start-up formation (β = −0.087, p < 0.01). In contrast, the entrepreneurial ecosystem index shows a positive and significant effect on green start-ups (β = 0.0016, p < 0.05), suggesting that stronger ecosystems are associated with higher entrepreneurial activity. At the same time, GSEI is negatively associated with early-stage funding for both green (β = −0.0037, p < 0.01) and digital (β = −0.0035, p < 0.1) energy start-ups, indicating a shift away from seed-stage investment in more developed ecosystems. Conversely, later-stage funding of digital energy start-ups increases with ecosystem strength (β = 0.0073, p < 0.1), highlighting a concentration of capital in more mature ventures. Finally, non-linear analysis reveals an inverted U-shaped relationship for later-stage green funding, with a turning point at approximately 133.9 points of GSEI, beyond which marginal gains diminish, confirming the presence of ecosystem saturation effects.

From a policy perspective, the results suggest that support for energy entrepreneurship should be more targeted across the start-up lifecycle. First, because digital connectivity is associated with early-stage green funding, governments should prioritise digital infrastructure that directly improves investor–start-ups interaction, including clean-tech data platforms, digital due diligence tools, open energy data systems, and smart-grid information infrastructure. Second, because broad DESI indicators do not consistently explain start-up formation, digitalisation policies should be complemented by sector-specific instruments, such as clean-tech incubators, regulatory sandboxes, pilot project grants, and specialised support for energy data and grid-related business models. Third, because stronger ecosystems are associated with a shift from early-stage to later-stage funding, policymakers should protect the seed-financing segment through public co-investment schemes, green venture funds, early-stage guarantees and targeted support for pre-commercial technologies. Fourth, because the non-linear results suggest diminishing returns in highly mature ecosystems, policy should focus not only on strengthening leading hubs but also on widening ecosystem inclusiveness by supporting peripheral regions, university-based entrepreneurship, smaller clean-tech firms and cross-border innovation networks.

The study makes three main contributions. Theoretically, it advances the literature on sustainability-oriented entrepreneurship by showing that digitalisation and entrepreneurial ecosystem maturity operate through different mechanisms. Digitalisation primarily functions as a selective enabling condition. At the same time, ecosystem maturity serves as a stage-specific allocation mechanism that shapes both start-up formation and the distribution of funding across early and later stages. Empirically, the study provides cross-country evidence distinguishing between green and digital energy start-ups, early-stage and later-stage funding, contemporaneous and lagged effects, and linear and non-linear relationships. Methodologically, the study contributes by applying PPML fixed-effects models to highly skewed, zero-heavy start-ups and funding data and by complementing baseline estimates with lagged and quadratic specifications to test the stability and non-linearity of the relationships.

The findings show that sustainable energy entrepreneurship cannot be accelerated by digitalisation alone; it requires inclusive ecosystems, balanced financing across the start-up lifecycle and policy instruments that prevent mature innovation hubs from concentrating investment at the expense of emerging green and digital ventures.