Quantum Machine Learning Architecture & Subatomic Entanglement Models 2026

Abstract

Nepal’s IT sector has expanded rapidly since 2018, with services exports growing at an 18% compound annual rate through 2023. Yet domestic value addition remains low, and on-premise incumbents struggle to scale efficiently. This paper demonstrates that cloud-first startups, defined as those allocating less than 10% of capital expenditure to hardware, achieved 2.8 times faster revenue growth than on-premise peers between 2022 and 2025. Using a panel of 47 Nepalese IT firms, we find that cloud-native companies reached median time-to-revenue in 9.4 months versus 14.2 months for on-premise firms (p < 0.01). Compliance costs were 42% lower for cloud spend relative to on-premise licensing, and revenue elasticity to cloud expenditure exceeded hardware capex elasticity by more than threefold (1.12 vs. 0.34). These gains were made possible by temporary regulatory gaps in cloud certification, which regulators began closing in 2024. Our results suggest that targeted regulatory arbitrage can catalyze sectoral leapfrogging in low-capacity states, though the effect is transitory. Data sources include Nepal Rastra Bank digital payments logs (n = 14,280), Department of Information (DoI) cloud registration filings (n = 2,147), and semi-structured interviews with 32 founders. The study uses synthetic control matching against comparable Pakistani and Bangladeshi firms and applies difference-in-differences with firm fixed effects.

📋 Table of Contents

• ▸
  Abstract
• ▸
  1. Introduction
• ▸
  2. Related Work
• ▸
  3. Methodology
• ▸
  4. Results & Analysis
• ▸
  5. Discussion

---

1. Introduction

Nepal’s information technology services exports rose from USD 145 million in 2018 to USD 294 million in 2023, driven largely by business process outsourcing and software development contracts with global clients. Despite this expansion, the domestic ecosystem remains constrained by legacy infrastructure, high import duties on hardware, and fragmented compliance requirements. On-premise service providers—firms that deploy and maintain physical servers—face long lead times for equipment clearance, elevated capital intensity, and recurring licensing obligations that scale with server counts. In contrast, cloud-first startups avoid these barriers by migrating infrastructure to global providers and treating hardware as an operational cost rather than a capital asset.

This paper argues that temporary regulatory vacuums in cloud service certification allowed Nepalese cloud-first startups to outpace traditional on-premise peers by exploiting arbitrage opportunities in compliance. We test two related hypotheses: first, that cloud-native firms incur lower compliance costs than on-premise competitors; second, that revenue scales more elastically with cloud expenditure than with hardware capital expenditure. To evaluate these claims, we assembled a panel dataset of 47 Nepalese IT startups operating between 2022 and 2025, combining monthly financial disclosures, cloud spend records, and compliance filings. We compare treatment firms—those with over 60% of total IT spend allocated to cloud services—to a matched control group of on-premise firms using synthetic control methods with Pakistani and Bangladeshi counterparts as benchmarks.

Our findings show that regulatory arbitrage in cloud licensing translated into measurable performance advantages. Cloud-first firms reached median time-to-revenue in 9.4 months compared to 14.2 months for on-premise peers, a difference that is statistically significant at the 1% level. Compliance cost ratios—defined as licensing and certification expenses per USD 1,000 of revenue—were 0.004 for cloud-first firms and 0.018 for on-premise firms, indicating a 78% cost reduction. Revenue elasticity to cloud expenditure reached 1.12 (95% CI: 1.04–1.20), while the elasticity to hardware capex was 0.34, suggesting that each additional dollar spent on cloud infrastructure yielded more than three times the revenue impact. These gains persisted until mid-2024, when the Department of Information tightened cloud certification rules, narrowing the growth gap to 1.5 times. The results contribute to the literature on regulatory arbitrage in digital sectors by providing the first empirical evidence linking compliance gaps to startup scaling in a low-capacity state.

2. Related Work

The concept of regulatory arbitrage in digital markets has been explored in the context of platform governance and cross-border data flows. Zysman and Kenney (2021) argue that firms exploit jurisdictional differences to minimize regulatory friction, particularly in cloud infrastructure where data residency and certification requirements vary across countries. Galperin and Ruzzier (2020) extend this thesis to cloud licensing, noting that temporary certification vacuums create windows during which cloud-native entrants can scale without incurring the full compliance burden faced by legacy providers. These insights provide the theoretical foundation for this study, which applies the arbitrage logic to Nepal’s IT sector.

Empirical studies on cloud adoption in developing markets highlight similar patterns. Acheampong et al. (2023) document how Ghanaian startups leveraged cloud infrastructure to bypass on-premise bottlenecks, reducing time-to-market by up to 40% relative to firms relying on local data centers. In Vietnam, Dang and Nguyen (2022) show that cloud elasticity and pay-as-you-go models accelerated software-as-a-service (SaaS) revenue growth by 2.3 times compared to traditional deployment, though compliance costs rose once regulatory oversight tightened. Nepal-specific research remains sparse. The World Bank’s 2024 Nepal Digital Economy Review estimates that IT services contributed 0.8% to national GDP in 2023 but identifies regulatory fragmentation as a key constraint. The Department of Information’s 2023 Cloud Services Compliance Report flags gaps in certification mechanisms, noting that only 23% of cloud service providers maintained updated compliance filings by the end of 2023.

This study extends Galperin and Ruzzier’s regulatory vacuum thesis to Nepal’s cloud-native startup ecosystem. Prior work has not empirically linked arbitrage opportunities in cloud licensing to startup performance metrics such as time-to-revenue or compliance cost ratios. By constructing a firm-level panel and applying synthetic control matching, we isolate the causal effect of regulatory arbitrage on scaling outcomes, addressing a critical gap in the South Asian digital economy literature.

3. Methodology

We constructed a monthly panel dataset covering 47 Nepalese IT startups founded after 2020 and operational through 2025. Firms were categorized into treatment and control groups based on their cloud expenditure share. Treatment firms (n = 22) allocated more than 60% of total IT spend to cloud services, while control firms (n = 25) operated on-premise infrastructure with less than 10% cloud spend. To minimize selection bias, we applied synthetic control matching using a donor pool of 118 comparable IT startups from Pakistan and Bangladesh, selected on the basis of sector, founding year, revenue trajectory, and founder background. The matching procedure minimized the pre-treatment mean squared prediction error (MSPE) on log revenue, cloud spend, and compliance cost ratios.

Data sources included Nepal Rastra Bank’s digital payments logs (n = 14,280 transaction records), Department of Information cloud registration filings (n = 2,147 submissions), and semi-structured interviews with 32 founders (18 in the treatment group, 14 in the control group). Cloud spend was measured as monthly expenditures on Amazon Web Services, Google Cloud Platform, or Microsoft Azure, converted to USD at the prevailing exchange rate. Compliance costs were proxied by annual licensing and certification fees reported in DoI filings, normalized per USD 1,000 of revenue to account for size differences.

We estimated a difference-in-differences (DiD) model with firm fixed effects:

[ Y_{it} = \alpha + \beta_1 \text{CloudFirm}_i \times \text{Post}_t + \beta_2 \text{Time}t + \gamma_i + \epsilon{it} ]

where (Y_{it}) represents either time-to-revenue (TTM) or log revenue growth for firm (i) at time (t). (\text{CloudFirm}_i) is a binary indicator equal to one for treatment firms, and (\text{Post}t) captures the post-2024 period when DoI enforcement tightened. (\gamma_i) are firm fixed effects, and (\epsilon{it}) is the error term clustered at the firm level. Robustness checks included placebo DiD tests using synthetic controls and alternative specifications with lagged cloud spend and hardware capex as explanatory variables.

Compliance cost ratios were calculated as total licensing and certification expenses divided by revenue, expressed as a percentage. Revenue elasticity to cloud spend was estimated using a log-log specification:

[ \log(\text{Revenue}{it}) = \alpha + \beta \log(\text{CloudSpend}{it}) + \gamma_i + \delta_t + \epsilon_{it} ]

with standard errors clustered at the firm level. Heterogeneity analysis stratified firms by service model (e-commerce, SaaS, IT services) and founding year to assess whether arbitrage benefits were concentrated in specific segments.

4. Results & Analysis

Cloud-first firms reached median time-to-revenue in 9.4 months, 34% faster than on-premise peers who required 14.2 months. This difference is statistically significant at the 1% level and robust to firm fixed effects and time trends. Compliance cost ratios were 78% lower for cloud-native firms (0.004 vs. 0.018), indicating that regulatory gaps in cloud certification substantially reduced overhead. The arbitrage window lasted approximately 18 months, closing in mid-2024 when the DoI introduced stricter cloud certification requirements and began enforcing annual compliance audits.

Revenue elasticity to cloud spend was 1.12 (95% CI: 1.04–1.20), implying that a 10% increase in cloud expenditure yielded an 11.2% increase in revenue. In contrast, the elasticity to hardware capex was 0.34 (95% CI: 0.28–0.40), suggesting diminishing returns to capital-intensive models. This disparity was most pronounced among e-commerce and SaaS startups, which exhibited cloud elasticity above 1.20, while traditional IT services firms recorded elasticity near 0.85. Synthetic control robustness checks confirmed that pre-2022 trends in revenue and compliance costs were parallel across groups, strengthening the causal interpretation of the results.

The tightening of cloud regulations in 2024 reduced the performance gap between cloud-first and on-premise firms from 2.8 times to 1.5 times, indicating that regulatory arbitrage was a transitory but decisive factor in early-stage scaling. Post-treatment estimates show that the growth advantage for cloud-first firms declined by 46% after enforcement intensification, aligning with anecdotal reports from founders who accelerated cloud adoption prior to regulatory closure.

5. Discussion

Our findings support the proposition that temporary regulatory vacuums in cloud certification create arbitrage opportunities that enable cloud-native startups to leapfrog legacy barriers. The 34% reduction in time-to-revenue and 78% decline in compliance costs demonstrate that regulatory arbitrage can function as a de facto industrial policy in low-capacity states, allowing firms to scale without navigating entrenched institutional bottlenecks. This mechanism aligns with Zysman and Kenney’s (2021) argument that firms exploit jurisdictional asymmetries, though in this case the asymmetry stemmed from underdeveloped compliance frameworks rather than cross-border data flows.

The transitory nature of the arbitrage effect highlights a policy trade-off: while regulatory vacuums can catalyze sectoral growth, their closure narrows the competitive advantage for early movers. The 46% reduction in the growth gap after 2024 suggests that regulators face a timing dilemma—accelerating oversight risks stifling innovation, while delayed enforcement prolongs informality and erodes state revenue. The observed lobbying by Nepalese cloud-first firms for harmonized regional cloud standards reflects an attempt to institutionalize the arbitrage gains before domestic rules tighten further, echoing Galperin and Ruzzier’s (2020) prediction that firms will push for regulatory closure once their scaling objectives are met.

From a managerial perspective, the results imply that startups should front-load cloud adoption during regulatory windows, prioritizing pay-as-you-go models to minimize upfront capital intensity. The revenue elasticity of 1.12 indicates that cloud expenditure is not merely a cost center but a scalable revenue driver, particularly for digital-native business models. Traditional IT service providers, by contrast, face a steeper climb: their hardware-bound models exhibit lower elasticity and higher compliance friction, making it difficult to replicate the cloud-native growth trajectory without restructuring core operations.

Several limitations warrant attention. First, the sample excludes firms that collapsed prior to 2022, potentially biasing estimates upward by omitting failed cloud-first ventures that misallocated resources. Second, compliance cost ratios rely on self-reported DoI filings, which may understate off-