Measuring Post-Quantum TLS Deployment Across UK Internet Sectors

Measuring Post-Quantum TLS Deployment Across UK Internet Sectors Konstantinos Loizou and Essam Ghadafi

arXiv:2608.02147v1 [cs.CR] 3 Aug 2026

School of Computing, Newcastle University, Newcastle upon Tyne, UK [email protected] [email protected]

Abstract. Post-quantum cryptography (PQC) is becoming an important component of long-term trust in Internet-facing infrastructure. Publicly observable PQC support provides evidence of externally visible deployment, but does not necessarily reflect the overall progress of an organisation’s post-quantum migration. This distinction matters when observable deployment is used as an indicator of organisational readiness or progress towards migration deadlines. We present a measurement study of observable PQC deployment across 4,665 UK organisations spanning ten sectors. We measure post-quantum key-exchange support across HTTPS and SMTP STARTTLS endpoints, attribute reachable endpoints to their underlying infrastructure providers where possible, and statistically examine protocol-, sector-, and providerlevel deployment patterns. Among reachable endpoints, 44.0% of HTTPS services supported at least one evaluated PQC key-exchange group, compared with 6.4% of SMTP services. Among organisations reachable over both protocols, HTTPS support was significantly more common than SMTP support (matched odds ratio 16.89). Although deployment varied across sectors, infrastructure provider identity was substantially more predictive than organisational sector, and observable deployment was highly concentrated among a small number of providers. Only 144 organisations supported PQC across both web and email infrastructure, highlighting an uneven and fragmented migration landscape. No postquantum certificate signatures were observed across the measured endpoints. These findings show that observable PQC deployment is currently shaped predominantly by infrastructure-provider deployment decisions and should not be interpreted as a complete measure of organisational migration readiness. Keywords: Post-quantum cryptography · TLS · Internet measurement · Trust assessment · Networked systems security

1

Introduction

Public-key cryptography underpins the security of modern Internet services, supporting authentication, key establishment, and secure communications across web, email, and cloud infrastructures [16,17,10,9]. However, the emergence of

large-scale quantum computers threatens many of the asymmetric cryptographic algorithms currently deployed in practice. In particular, Shor’s algorithm can efficiently solve the integer factorisation and discrete logarithm problems on which widely used schemes such as RSA and elliptic-curve cryptography rely [21]. The possibility of adversaries collecting encrypted traffic today for future decryption, commonly referred to as the harvest-now-decrypt-later threat, further increases the urgency of migration towards quantum-resistant alternatives. To address this challenge, significant international efforts have focused on standardising and deploying post-quantum cryptography (PQC). In 2024, the U.S. National Institute of Standards and Technology (NIST) standardised the first generation of post-quantum cryptographic algorithms, publishing ML-KEM, ML-DSA and SLH-DSA as FIPS 203, FIPS 204 and FIPS 205, respectively [14,13,15]. In parallel, hybrid key exchange mechanisms have been proposed to facilitate the gradual integration of post-quantum cryptography into TLS while maintaining compatibility with existing deployments [23]. Governments and regulatory bodies have also begun establishing migration roadmaps for the transition to quantum-resistant cryptography. In the United Kingdom, the National Cyber Security Centre (NCSC) has published a phased migration timeline recommending that organisations complete cryptographic discovery, dependency mapping, and migration planning by 2028, followed by the progressive migration of systems and services during the subsequent years [12]. Similarly, the European Commission has recommended a coordinated roadmap to support the transition to post-quantum cryptography across member states and critical sectors [4]. These initiatives highlight the growing importance of understanding the current state of PQC deployment and establishing baselines against which future migration progress can be assessed. Recent industry developments further emphasise the urgency of understanding current deployment practices. Major Internet infrastructure providers have accelerated their own migration roadmaps, with Google and Cloudflare publicly targeting 2029 for completing the migration of their infrastructures to postquantum cryptography [5,26]. Given the substantial proportion of Internet-facing services delivered through these providers, their deployment decisions are likely to have a significant influence on the observable state of PQC adoption across organisations. Consequently, understanding the contribution of infrastructure providers is important when interpreting Internet-scale measurements of organisational PQC readiness. The United Kingdom provides a particularly relevant case study because the NCSC has published one of the most comprehensive national migration roadmaps for PQC adoption, allowing observable deployment to be interpreted within a well-defined policy framework. Despite recent progress in measuring PQC deployment across Internet-facing services, several important challenges remain. Existing studies predominantly rely on popularity-ranked domain lists, provide limited visibility into sectorspecific organisational deployment, typically analyse either HTTPS or SMTP in isolation, and provide limited insight into the influence of infrastructure providers

on observable deployment. To address these challenges, this paper presents an empirical measurement study of 4,665 UK organisations spanning ten sectors. Using a sector-stratified dataset, we jointly analyse HTTPS and SMTP STARTTLS infrastructures, attribute reachable endpoints to their underlying infrastructure providers, and examine leaf certificate signature algorithms alongside post-quantum key exchange support. Collectively, these analyses provide a complementary perspective to existing Internet-wide measurement studies by focusing on organisational sectors and the role of infrastructure providers in shaping observable PQC deployment. Accordingly, this study addresses the following research questions: RQ1: How prevalent is observable PQC support across HTTPS and SMTP infrastructures? RQ2: Does observable PQC support vary across organisational sectors? RQ3: Is observable PQC support better predicted by organisational sector or infrastructure provider? RQ4: How concentrated is observable PQC deployment among infrastructure providers? Our Contribution. The contributions of this paper are as follows: – We present a sector-stratified measurement study of observable PQC deployment across 4,665 UK organisations spanning ten sectors. – We jointly measure HTTPS and SMTP STARTTLS deployment for the same organisational domains, enabling paired comparison of observable PQC support across web and email infrastructures. – We quantify differences in deployment across protocols and sectors using confidence intervals, paired hypothesis testing, and effect-size analysis. – We compare the predictive value of organisational sector and infrastructure provider using cross-validated logistic regression, including MX-host-grouped validation for SMTP to account for shared mail infrastructure. – We show that provider identity is substantially more predictive of observable PQC support than organisational sector, and that deployment is highly concentrated among a small number of infrastructure providers. – As a secondary analysis, we characterise leaf certificate signature algorithms to assess whether visible deployment has progressed beyond post-quantum key establishment. The remainder of the paper is organised as follows. Section 2 reviews background and related work. Section 3 describes the measurement methodology. Section 4 presents the empirical results. Section 5 discusses the implications of the findings for post-quantum migration and organisational readiness, and Section 6 concludes the paper.

2

Related Work

Large-scale Internet measurement has long been used to study the deployment and evolution of security protocols and public key infrastructures. Holz et al. [8]

presented one of the first comprehensive active and passive measurement studies of the X.509 Public Key Infrastructure (PKI), demonstrating the effectiveness of Internet-scale measurements for analysing certificate ecosystems and TLS deployments. More recently, Holz et al. [7] investigated the deployment of TLS 1.3 and showed that rapid protocol adoption was largely driven by a small number of cloud providers and content delivery networks. These observations highlight the importance of considering the role of infrastructure providers when interpreting Internet-wide deployment measurements. Other measurement studies have examined TLS deployment within specific sectors. For example, Chung and Vlajic analysed the TLS configurations of major banking websites using remote scanning techniques, highlighting the continued prevalence of legacy protocol versions and insecure cryptographic configurations in a critical sector [1]. Several recent studies have investigated the deployment of post-quantum cryptography across Internet-facing services. Schaumann conducted longitudinal Internet measurement studies examining PQC deployment across Internet-facing websites and later extended the analysis to SMTP infrastructures [19,20]. Complementary evidence has also been reported in industry measurement reports, such as those published by F5 Labs, which analyse PQC deployment across large populations of Internet-facing websites and examine deployment trends across infrastructure providers and geographical regions [25]. Collectively, these studies indicate increasing deployment of post-quantum TLS, although observable deployment remains highly concentrated among a relatively small number of major infrastructure providers. While preparing this manuscript, a related measurement study of Internet PQC readiness became publicly available [3]. That study analyses more than 32,000 Internet-facing domains by examining TLS protocol versions, cipher suites, key exchange mechanisms, and certificate configurations. Although complementary, the two studies differ in several important respects. Our work focuses specifically on UK organisations across 10 sectors, jointly analyses HTTPS and SMTP infrastructures, examines sector-level deployment patterns, attributes endpoints to their underlying infrastructure providers, and analyses certificate signature algorithms alongside post-quantum key exchange support. Rather than providing another Internet-wide scan, our study complements existing work by offering a sector-oriented perspective and investigating the influence of infrastructure providers on observable PQC deployment. Together, the two studies provide complementary perspectives on the current state of observable PQC deployment across Internet-facing services.

3

Methodology

This section describes the methodology used to measure observable deployment of post-quantum cryptography across UK organisations, including dataset construction, HTTPS and SMTP measurements, infrastructure-provider attribution, statistical analysis, and the ethical considerations that guided the measurement campaign.

3.1

Dataset Construction

The final measurement dataset comprises 4,665 UK organisations spanning ten sectors. To provide broadly balanced representation while avoiding over-representation of sectors containing substantially larger organisational populations, the commercialsector samples and the Government sample were limited to a maximum of 500 organisations per sector. All 165 eligible Higher Education institutions were included because this source contained fewer than 500 entries. A sector-stratified sampling strategy was adopted using publicly available data sources. Commercial organisations from the Construction, Utilities and Environmental Services, Healthcare, Legal, Retail, Technology, Telecommunications, and Finance and Insurance sectors were identified using the Moody’s FAME database [11]. The dataset was filtered to include only active organisations whose registered office was located in England, Scotland, Wales, or Northern Ireland. Organisations were assigned to sectors using their primary UK Standard Industrial Classification (SIC) code. Table 1 summarises the primary SIC codes used to construct the commercial-sector dataset. Table 1. Primary UK SIC codes used to construct the commercial-sector dataset. Sector Primary UK SIC code(s) Construction 41, 42, 43 Finance & Insurance 64, 65, 66 Healthcare 86 Legal 691 Retail 47 Technology 62, 63 Telecommunications 61 Utilities & Environmental Services 35, 36, 37, 38, 39

For each commercial sector, the first 500 eligible records returned by the corresponding FAME query were selected. The website recorded by FAME was used as the organisation’s public Internet domain. Records without a valid public domain were skipped and replaced using the next eligible record returned by the query. Government organisations were obtained from the official GOV.UK domain list [6], with the first 500 eligible non-duplicate domains retained. Higher Education institutions were obtained from the UK Learning Providers dataset [24]. An initial Police category was constructed using the UK Police API [22], together with the British Transport Police domain. However, because this category contained fewer than the minimum threshold of 50 organisations adopted for reporting sector-level statistics, it was excluded from all analyses and public reporting. Each organisation in the final dataset was represented by a single publicly accessible Internet domain corresponding to its primary organisational website.

Duplicate domains across sectors were removed to ensure that each domain appeared only once. Where a null, invalid, or previously included domain was encountered, the next available eligible entry from the corresponding source was used to maintain the target sample size. This process resulted in a final analytical dataset of 4,665 organisations across ten sectors. All HTTPS and SMTP measurements reported in this paper were collected during a single measurement campaign on 30 June 2026. The dataset therefore represents a cross-sectional snapshot of observable PQC deployment at that point in time. 3.2

HTTPS & SMTP Measurements

HTTPS services were evaluated using TLS handshakes on TCP port 443. Each endpoint was probed using TLS 1.3 handshakes offering the 4 post-quantum key exchange groups listed in Table 2. Since PQC key exchange is negotiated through the TLS 1.3 supported groups extension, only endpoints successfully negotiating TLS 1.3 were evaluated for PQC capability. For each successful handshake, the negotiated TLS version, negotiated key exchange group, X.509 certificate issuer, and leaf certificate’s signature algorithm were recorded for subsequent analysis. SMTP infrastructure was evaluated independently using the STARTTLS extension on TCP port 25. For each organisation, the highest-priority Mail Exchange (MX) host was identified from its DNS MX records. As multiple organisations frequently share common mail infrastructure, duplicate MX hosts were removed prior to scanning, ensuring that each unique mail server was measured only once. Each unique MX host was subsequently probed using the hybrid key exchange groups listed in Table 2. The resulting measurements were then mapped back to all organisations sharing the corresponding mail infrastructure. Alongside PQC support, the negotiated TLS version, X.509 certificate issuer, and leaf certificate signature algorithm were recorded for each successful connection, enabling comparative analysis of HTTPS and SMTP deployment across the dataset. Table 2. TLS 1.3 post-quantum key exchange groups evaluated in this study. Key Exchange Group Classical Component PQC component X25519MLKEM768 X25519 ML-KEM-768 SecP256r1MLKEM768 secp256r1 ML-KEM-768 MLKEM1024 – ML-KEM-1024 SecP384r1MLKEM1024 secp384r1 ML-KEM-1024

The evaluated key exchange groups comprise the hybrid and pure postquantum key exchange groups supported by the OpenSSL implementation used during data collection. The hybrid groups combine a classical elliptic-curve key

exchange mechanism with ML-KEM, whereas MLKEM1024 represents a pure postquantum key exchange group. An endpoint was classified as supporting observable PQC deployment if at least one of the evaluated post-quantum key-exchange groups was successfully negotiated. 3.3

Infrastructure Provider Attribution

Following completion of the HTTPS and SMTP measurements, a separate infrastructure attribution stage was performed to identify the underlying infrastructure provider responsible for each reachable endpoint. For HTTPS services, provider attribution combined multiple complementary techniques, including DNS CNAME resolution, reverse DNS (PTR) lookups, Autonomous System (AS) mapping, WHOIS organisation-name matching and HTTP response header analysis. The implementation was based on the opensource whohosts software [18], which integrates these attribution techniques into a unified workflow. SMTP infrastructure was attributed by analysing the highest-priority MX hostname associated with each organisation. MX hostnames were matched against provider-specific patterns representing the major email service providers, including Microsoft, Google, Hostinger, Proofpoint, Mimecast, Barracuda, Cisco, Cloudflare, Apple iCloud, Amazon, Namecheap and other widely deployed hosted email services. Additional provider-specific patterns were incorporated for organisations operating multiple hosting brands, including United Internet, to improve attribution coverage. MX hostnames that could not be matched to a recognised provider were classified as Other. The resulting provider attributions were subsequently used to analyse the distribution of observable PQC deployment across major hosting and email infrastructure providers. 3.4

Statistical Analysis

Statistical analyses were performed to evaluate deployment patterns at the protocol, sector, and infrastructure-provider levels. Deployment proportions are reported with Wilson 95% confidence intervals. Differences between HTTPS and SMTP deployment were evaluated among organisations reachable over both protocols using McNemar’s test with continuity correction. The corresponding matched odds ratio and 95% confidence interval were calculated using the discordant pairs. Associations between organisational sector and observable PQC deployment were evaluated using Pearson’s chi-square test of independence, with Cramer’s V reported as a measure of effect size. These analyses used the final ten-sector dataset after exclusion of the Police category described in Section 3.1. To compare the predictive value of organisational sector and infrastructure provider, logistic regression models were fitted using three predictor sets: (i) sector only, (ii) provider only, and (iii) sector and provider combined. Categorical

predictors were one-hot encoded, and model discrimination was measured using the area under the receiver operating characteristic curve (AUC). HTTPS models were evaluated using stratified five-fold cross-validation. SMTP models were evaluated using five-fold grouped cross-validation, with organisations grouped by anonymised MX-host identifier so that organisations sharing the same mail infrastructure were not divided between training and test folds. Provider concentration was summarised by ranking infrastructure providers according to the number of PQC-supporting endpoints attributed to each provider. Concentration was then quantified using the proportions attributable to the highest-ranked provider, the three highest-ranked providers, and the Herfindahl– Hirschman Index (HHI). All analyses were implemented in Python 3.13 using pandas, NumPy, SciPy, statsmodels, and scikit-learn.

3.5

Ethical Considerations

The study collected only publicly observable TLS handshake metadata from Internet-facing HTTPS and SMTP services. To minimise operational impact, the measurements employed conservative scanning parameters, including connection timeouts, inter-connection delays and rate-limiting mechanisms. Each endpoint received only the minimum number of standards-compliant TLS handshakes required to determine protocol reachability, supported post-quantum key exchange groups and certificate characteristics, and all connections were terminated cleanly following protocol completion. The measurement methodology was designed in accordance with the ethical principles outlined in the Menlo Report [2] by minimising potential disruption while collecting only the metadata necessary to assess observable deployment of post-quantum cryptography. No attempts were made to authenticate to services, exploit vulnerabilities, enumerate additional resources, or otherwise interfere with the operation of the measured systems. A minimum sector size of 50 organisations was applied for public reporting, and the Police category was excluded because it did not meet this threshold. Organisation domains and MX hostnames were replaced with stable pseudonymous identifiers for statistical analysis. Only aggregate statistics are reported in this paper. The original scan data containing organisation identifiers are not included in the published results and are retained only in a secure, access-controlled environment for verification purposes.

4

Results

This section presents the results of our empirical measurement study of postquantum cryptographic deployment across UK organisations.

4.1

Measurement Coverage

The final analytical dataset comprised 4,665 UK organisations spanning ten sectors. During the measurement campaign, 4,063 organisations (87.10%) were reachable over HTTPS, while 3,858 organisations (82.70%) were reachable over SMTP. The SMTP dataset contained 2,615 distinct resolvable MX-host identifiers. Among the SMTP-reachable organisations used in the predictive analysis, the 3,858 observations corresponded to 2,408 distinct MX-host groups. Adoption statistics are reported at the organisation level to enable comparison with HTTPS and sector-level analysis, while the grouped cross-validation procedure described in Section 3.4 accounts for organisations sharing the same mail infrastructure. Table 3. Summary of measurement coverage. Metric HTTPS SMTP Organisations in analytical dataset 4,665 4,665 Reachable endpoints 4,063 3,858 Reachability (%) 87.10 82.70 Distinct resolvable MX hosts – 2,615

4.2

Observable PQC Deployment Across Protocols and Sectors

Table 4 presents the distribution of the evaluated HTTPS post-quantum keyexchange groups across the ten measured sectors. The X25519MLKEM768 hybrid group was by far the most widely deployed, with 1,787 HTTPS endpoints supporting this configuration. Government exhibited the highest number of deployments (255), followed by Retail (230), Technology (212), and Legal (188). Support for the remaining groups was considerably less common. A total of 161 HTTPS endpoints supported SecP256r1MLKEM768, while the pure post-quantum MLKEM1024 group was observed at only three endpoints. The SecP384r1MLKEM1024 group was supported by 59 endpoints. Overall, 1,788 of the 4,063 reachable HTTPS endpoints supported at least one evaluated postquantum key-exchange group. Observable HTTPS deployment was therefore overwhelmingly centred on X25519MLKEM768, while alternative hybrid configurations and pure post-quantum key exchange remained rarely deployed. SMTP exhibited substantially lower PQC deployment and algorithm diversity than HTTPS. Across the ten sectors, 246 SMTP endpoints supported at least one evaluated post-quantum key-exchange group. Support was almost entirely limited to X25519MLKEM768; only one SMTP endpoint, in the University sector, also supported SecP256r1MLKEM768. Technology exhibited the highest number of PQC-supporting SMTP endpoints (72), followed by Retail (57) and Telecommunications (30). Observable SMTP PQC deployment therefore remained considerably less prevalent and less diverse than HTTPS deployment.

Table 4. Support for the evaluated HTTPS post-quantum key exchange groups by sector. The total number of reachable HTTPS endpoints in each sector is shown in parentheses. “Env. Serv.” denotes Environmental Services. X25519 SecP256r1 MLKEM SecP384r1 Sector MLKEM768 MLKEM768 1024 MLKEM1024 Construction (445) 145 13 0 7 Finance & Insurance (425) 177 25 0 1 Government (455) 255 1 0 1 Healthcare (445) 186 11 0 3 Legal (430) 188 13 1 12 Retail (454) 230 28 0 4 Technology (437) 212 31 1 6 Telecommunications (407) 186 10 1 1 University (148) 39 5 0 4 Utilities & Env. Serv. (417) 169 24 0 20 Overall (4063) 1787 161 3 59

Table 5 presents the number of organisations supporting at least one evaluated post-quantum key-exchange group across both HTTPS and SMTP. Although 1,788 HTTPS endpoints and 246 SMTP endpoints supported PQC, only 144 organisations supported PQC across both services. Technology (42 organisations) and Retail (40) exhibited the highest levels of dual-service deployment, followed by Legal (14) and Telecommunications (12). Deployment across multiple Internet-facing services therefore remained comparatively limited. Figure 1 compares observable PQC support over HTTPS and SMTP across the ten sectors. HTTPS deployment consistently exceeded SMTP deployment across all sectors, although the magnitude of this difference varied considerably. SMTP deployment remained much more limited, with Technology, Retail, and Telecommunications exhibiting the largest numbers of PQC-supporting endpoints. To address RQ1, we compared the prevalence of observable PQC deployment across the two protocols. Among reachable endpoints, 1,788 of 4,063 HTTPS services supported at least one evaluated post-quantum key-exchange group, corresponding to 44.0% (95% Wilson CI: 42.5–45.5%). For SMTP, 246 of 3,858 reachable services supported PQC, corresponding to 6.4% (95% Wilson CI: 5.6– 7.2%). Restricting the paired analysis to the 3,510 organisations reachable over both protocols, 144 supported PQC over both services, 1,419 supported PQC only over HTTPS, 84 supported PQC only over SMTP, and 1,863 supported neither. McNemar’s test confirmed that observable PQC support was significantly more common over HTTPS than SMTP (χ2 = 1184.00, p < 0.001). The matched odds ratio was 16.89 (95% CI: 13.56–21.05). To address RQ2, Pearson’s chi-square tests indicated statistically significant associations between organisational sector and observable PQC deployment for both HTTPS (χ2 (9) = 85.01, p < 0.001) and SMTP (χ2 (9) = 165.10, p <

Table 5. Number of organisations supporting at least one of the evaluated postquantum key exchange groups across both HTTPS and SMTP services, grouped by sector. Sector Domains Construction 5 Finance & Insurance 11 Government 2 Healthcare 7 Legal 14 Retail 40 Technology 42 Telecommunications 12 University 2 Utilities & Environmental Services 9 Overall 144

0.001). The corresponding effect sizes were modest, with Cramer’s V = 0.145 for HTTPS and V = 0.207 for SMTP. Although sector was associated with observable deployment, the relatively small effect sizes indicate that sector alone provides limited discrimination between PQC-supporting and non-supporting endpoints. Overall, these findings demonstrate that observable PQC deployment is substantially more prevalent for HTTPS than SMTP. Although observable deployment varied significantly across organisational sectors, the modest effect sizes suggest that organisational sector alone provides limited explanatory power, motivating the provider-level analysis presented in the following section. 4.3

Deployment Across Infrastructure Providers

Tables 6 and 7 summarise the major infrastructure providers ranked by the number of reachable endpoints attributed to each provider. HTTPS deployment was concentrated among a small number of providers. Cloudflare hosted 1,312 reachable HTTPS endpoints, of which 1,247 (95.05%) supported at least one evaluated post-quantum key-exchange group. Fastly also exhibited a high deployment rate, with 78 of 81 endpoints (96.30%) supporting PQC. In contrast, several providers hosting substantial numbers of reachable endpoints exhibited little or no observable deployment. The SMTP ecosystem exhibited a different profile. Microsoft hosted 1,714 reachable SMTP endpoints, none of which supported the evaluated PQC groups. By contrast, all 242 Google-hosted SMTP endpoints supported PQC and accounted for almost all observable SMTP deployment. To address RQ3, Table 8 compares logistic regression models using organisational sector only, infrastructure provider only, and both predictors. For HTTPS, the provider-only model achieved an AUC of 0.957, substantially outperforming the sector-only model (AUC = 0.573). Combining sector and provider produced only a marginal improvement (AUC = 0.962).

Construction

11

Finance & Insurance

13

Government

145 177 255

6

Healthcare

17

Legal

21

187 188

Retail

230

57

Technology

212

72

Telecommunications

186

30

University

39

6

Utilities & Env. Services

13

0

HTTPS SMTP

169

50

100

150

Number of organisations

200

250

Fig. 1. Number of organisations supporting at least one of the evaluated post-quantum key exchange groups over HTTPS and SMTP across the 10 UK sectors included in the analytical dataset.

For SMTP, five-fold cross-validation was grouped by anonymised MX-host identifier. The provider-only model achieved an AUC of 0.994, whereas the sector-only model achieved an AUC of 0.156. Combining sector and provider produced an AUC of 0.996. For comparison, ordinary row-wise stratified crossvalidation produced an SMTP sector-only AUC of 0.699; this result does not account for organisations sharing the same mail infrastructure and is therefore not used as the primary estimate. To address RQ4, we quantified the concentration of observable deployment among providers. Cloudflare accounted for 69.7% of all PQC-supporting HTTPS endpoints, while Cloudflare, Amazon AWS, and Fastly together accounted for 86.9%. SMTP deployment was even more concentrated: Google accounted for 98.4% of PQC-supporting SMTP endpoints, and the three largest provider categories accounted for all observed SMTP deployment. The concentration of infrastructure providers across all reachable endpoints was also reflected by the Herfindahl–Hirschman Index (HHI), which was 1,378 for HTTPS and 2,797 for SMTP. These values indicate a substantially more concentrated provider landscape for SMTP than for HTTPS. These findings show that infrastructure-provider identity provides substantially greater predictive power than organisational sector and that observable PQC deployment is currently concentrated among a small number of providers.

Table 6. Top 10 HTTPS infrastructure providers ranked by the number of reachable endpoints. Percentages indicate the proportion of reachable endpoints supporting at least one evaluated post-quantum key-exchange group. Provider Reachable PQC PQC (%) Cloudflare 1312 1247 95.05 Amazon AWS 606 228 37.62 Microsoft 261 4 1.53 Google 176 12 6.82 Akamai 143 2 1.40 20i 138 0 0.00 Krystal 121 0 0.00 DigitalOcean 116 2 1.72 Fastly 81 78 96.30 Ionos 72 1 1.39 Table 7. Top 10 SMTP infrastructure provider categories ranked by the number of reachable endpoints. Eleven provider categories are shown because of a tie for the tenth position. Percentages are calculated using only reachable SMTP endpoints attributed to each provider category. Provider Reachable PQC PQC (%) Microsoft 1714 0 0.00 Other Providers 739 3 0.41 Mimecast 650 0 0.00 Proofpoint 368 0 0.00 Google 242 242 100.00 Barracuda 91 0 0.00 Cisco 25 0 0.00 Cloudflare 10 0 0.00 Amazon 5 0 0.00 Namecheap 4 0 0.00 Zoho Mail 4 0 0.00

4.4

Leaf Certificate Signing Algorithm

In addition to evaluating post-quantum key exchange support, we analysed the signature algorithm used by the leaf certificate presented during successful TLS handshakes. This enables us to assess whether observable PQC deployment has progressed beyond key establishment to certificate-based authentication. Table 9 summarises the leaf certificate signature algorithms observed in the certificates presented by all reachable HTTPS and SMTP endpoints. RSA-based signatures dominate both protocols, with SHA-256 with RSA accounting for the majority of certificates. ECDSA is also widely deployed for HTTPS but are comparatively uncommon for SMTP. No post-quantum certificate signature algorithms were observed, indicating that, although hybrid postquantum key exchange has begun to emerge, certificate-based authentication remains entirely dependent on classical cryptographic signature algorithms.

Table 8. Cross-validated AUCs of logistic regression models predicting observable PQC support. Protocol Sector only Provider only Sector + Provider HTTPS 0.573 0.957 0.962 SMTP 0.156 0.994 0.996 Table 9. Leaf certificate signature algorithms observed across reachable HTTPS and SMTP endpoints. Signature Algorithm HTTPS SMTP SHA-256 with RSA 2561 3824 ECDSA with SHA-256 940 0 ECDSA with SHA-384 540 24 Other RSA 22 10 Total certificates 4063 3858

5

Discussion

The results presented in this study provide an empirical view of the current state of observable PQC deployment across UK Internet-facing organisations. Although PQC support is becoming increasingly visible, particularly for HTTPS infrastructures, deployment remains highly uneven across services, sectors and infrastructure providers. Consequently, observable protocol support should be interpreted as evidence of externally visible infrastructure capability rather than a comprehensive measure of organisational migration readiness. 5.1

Interpretation of the Findings

The results demonstrate a substantial disparity between observable PQC deployment over HTTPS and SMTP. Of the reachable endpoints, 1,788 supported at least one evaluated PQC group over HTTPS, compared with 246 over SMTP, and only 144 organisations supported PQC across both services. The paired analysis confirmed that observable support was approximately 16.9 times more likely over HTTPS than SMTP. Evaluating a single protocol in isolation may therefore overstate an organisation’s broader level of externally visible post-quantum readiness. Observable deployment should therefore be interpreted as evidence of externally visible capability rather than evidence that an organisation has completed PQC migration. Observable deployment was overwhelmingly centred on X25519MLKEM768. The remaining evaluated hybrid groups exhibited limited deployment, while the pure post-quantum MLKEM1024 group was observed at only three HTTPS endpoints and was not observed over SMTP. Current deployment therefore remains focused on hybrid migration mechanisms that preserve a classical component. Sector was significantly associated with observable deployment for both protocols, but the corresponding effect sizes were modest. Cross-validated mod-

elling provided a stronger distinction: infrastructure-provider identity substantially outperformed organisational sector as a predictor, and adding sector to a provider-based model produced only marginal improvement. Raw sector differences should therefore not be interpreted primarily as evidence of sectorspecific migration programmes; they appear to reflect, to a substantial extent, the provider ecosystems used by organisations in those sectors. Observable deployment was also highly concentrated. Cloudflare accounted for 69.7% of PQC-supporting HTTPS endpoints, while Google accounted for 98.4% of PQC-supporting SMTP endpoints. Several other providers hosting substantial numbers of reachable services exhibited little or no observable support. These findings indicate that present Internet-facing deployment is strongly shaped by provider-level configuration decisions rather than uniform organisationspecific adoption of post-quantum cryptography. 5.2

Comparison with Previous Work

The findings broadly agree with previous Internet-scale measurement studies reporting increasing deployment of post-quantum TLS across Internet-facing services. Schaumann’s longitudinal measurements similarly observed that deployment is concentrated among organisations using major cloud and content delivery providers, while industry reports published by F5 Labs also identified Cloudflare as one of the principal drivers of observable HTTPS PQC deployment. Our results reinforce these observations by demonstrating that deployment remains highly concentrated among a relatively small number of infrastructure providers, particularly for HTTPS, while also showing that observable SMTP deployment remains considerably more limited. During the preparation of this manuscript, Dubey and Varshney [3] published a large-scale measurement study examining PQC readiness across more than 32,000 Internet-facing domains. While both studies investigate observable deployment of post-quantum cryptography, the two works differ substantially in scope and methodology and should be considered complementary. Their dataset is drawn primarily from the Tranco global domain popularity ranking, supplemented with a set of Indian banking-sector domains, rather than using a nationally or sector-stratified sampling methodology. In contrast, our work employs sector-stratified sampling using UK Standard Industrial Classification (SIC) codes across the full analytical dataset, jointly analyses HTTPS and SMTP infrastructures, attributes endpoints to their underlying infrastructure providers, and evaluates deployment patterns using formal statistical testing and cross-validated predictive modelling. A further methodological distinction is that their measurements primarily analyse negotiated TLS parameters obtained using browser-based clients, reflecting the interaction between client capabilities and server configuration, whereas our measurements probe server-side protocol support directly. These represent complementary perspectives on observable PQC deployment, and together the two studies provide a broader picture of the current state of Internet-wide post-quantum readiness.

5.3

Implications for Organisational Readiness

The findings have several implications for assessing progress towards national and international post-quantum migration roadmaps. The migration guidance published by the UK National Cyber Security Centre (NCSC) and the European Commission emphasises activities such as cryptographic discovery, dependency identification, migration planning and risk assessment before large-scale deployment of post-quantum cryptography. Observable protocol support alone cannot demonstrate that these activities have been completed, nor can it provide evidence that an organisation has comprehensively migrated its cryptographic infrastructure. At the same time, the results demonstrate the important role played by infrastructure providers in accelerating Internet-wide deployment. Organisations relying on providers that have enabled hybrid post-quantum key exchange may benefit from improved protection of their externally visible services without necessarily modifying their own public-facing applications or infrastructure. Consequently, infrastructure providers are likely to play a significant role in the early phases of post-quantum migration by enabling widespread deployment across large numbers of organisations. The certificate analysis provides an additional perspective on migration readiness. Although hybrid post-quantum key exchange was observed across a substantial number of HTTPS deployments, no post-quantum certificate signature algorithms were identified within the measured dataset. This suggests that current observable deployments remain focused on protecting key establishment, while certificate-based authentication continues to rely entirely on classical cryptographic algorithms. Observable deployment should therefore be interpreted as an indicator of externally visible infrastructure capability rather than as evidence of a complete transition to post-quantum cryptography. Future Internet-scale measurement studies should continue to combine protocol measurements with infrastructure-provider attribution and additional deployment characteristics to provide a more comprehensive assessment of postquantum migration progress and to distinguish provider-driven deployment from organisation-driven migration. 5.4

Limitations

Several limitations should be considered when interpreting the results. First, the study focuses exclusively on UK organisations and therefore may not reflect deployment characteristics in other geographical regions. Second, the measurements consider only publicly observable HTTPS and SMTP services, excluding internally deployed systems and other communication protocols. Third, although the provider-attribution methodology successfully identifies the major hosting and email infrastructure providers responsible for much of the observed deployment, it cannot guarantee complete attribution for every endpoint. Consequently, the provider statistics presented in this paper should be interpreted as an approximation of infrastructure concentration rather than a definitive attribution of organisational responsibility.

Fourth, the study evaluates only the post-quantum key-exchange groups supported by the measurement platform and OpenSSL implementation used during data collection. As post-quantum standards and implementations continue to evolve, future studies should consider additional algorithms and deployment configurations as they become supported by mainstream TLS implementations. The commercial-sector samples are not probability samples of all UK organisations. For each FAME query, the first 500 eligible records returned by the database were retained, with null or duplicate domains replaced by subsequent eligible records. The results should therefore be interpreted as measurements of the constructed sector-stratified dataset rather than unbiased population estimates for every UK organisation in each sector. A minimum public-reporting threshold of 50 organisations was applied, resulting in the exclusion of the Police sector from the reported analyses. The paper therefore makes no claims about observable PQC deployment within that sector. In addition, all measurements were collected on 30 June 2026 and represent a single cross-sectional snapshot. The results cannot establish deployment trends, growth rates, or persistence over time. The sector-level SMTP analysis treats organisations as independent observations. Because multiple organisations may share common mail infrastructure, the reported significance levels should be interpreted with appropriate caution. However, the corresponding effect size remained modest, and the subsequent provider-level analyses explicitly examined the influence of shared infrastructure. Future iterations of the measurement framework could further improve transparency by providing a publicly accessible project page or contact mechanism through which organisations can identify the source of measurement traffic and request exclusion from future measurement campaigns. Despite these limitations, this study provides one of the first empirical sectorlevel assessments of observable post-quantum cryptographic deployment across UK organisations. By jointly analysing HTTPS and SMTP deployment, infrastructure-provider concentration, and certificate signature algorithms, it establishes an empirical baseline against which future progress towards post-quantum migration can be assessed and highlights the importance of considering infrastructureprovider influence when interpreting Internet-scale deployment measurements.

6

Conclusion

This paper presented an empirical measurement study of observable post-quantum cryptographic deployment across 4,665 UK organisations spanning ten sectors. Using a sector-stratified dataset, we jointly analysed HTTPS and SMTP STARTTLS infrastructures, attributed reachable endpoints to their infrastructure providers, and examined leaf certificate signature algorithms. Among reachable endpoints, 44.0% of HTTPS services supported at least one evaluated post-quantum key-exchange group, compared with 6.4% of SMTP services. Among organisations reachable over both protocols, observable PQC

support was approximately 16.9 times more likely over HTTPS than SMTP, and only 144 organisations supported PQC across both services. Deployment was overwhelmingly centred on the X25519MLKEM768 hybrid group. Although observable deployment differed significantly across organisational sectors, the corresponding effect sizes were modest. Cross-validated modelling showed that infrastructure-provider identity was substantially more predictive than organisational sector, achieving provider-only AUCs of 0.957 for HTTPS and 0.994 for SMTP. Deployment was also highly concentrated, with the largest provider accounting for 69.7% of PQC-supporting HTTPS endpoints and 98.4% of PQC-supporting SMTP endpoints. No post-quantum leaf-certificate signature algorithms were observed. These findings show that externally visible PQC support currently reflects infrastructure-provider deployment decisions to a substantial extent and should not be interpreted as a comprehensive measure of organisational migration readiness. Furthermore, the absence of post-quantum certificate signatures suggests that observable deployment remains focused on key establishment rather than complete cryptographic migration. The measurements reported in this paper provide an empirical baseline for future longitudinal studies of post-quantum deployment across UK Internet-facing services.

References 1. Chung, J., Vlajic, N.: Survey of remote tls vulnerability scanning tools and snapshot of tls use in banking sector. In: 2022 IEEE Conference on Communications and Network Security (CNS). pp. 1–2. IEEE (2022). https://doi.org/10.1109/ CNS56114.2022.9947230 2. Dittrich, D., Kenneally, E.: The menlo report: Ethical principles guiding information and communication technology research. Tech. Rep. TR-2012-001, U.S. Department of Homeland Security (2012), https: //www.caida.org/catalog/papers/2012_menlo_report_actual_formatted/ menlo_report_actual_formatted.pdf 3. Dubey, V.M., Varshney, G.: Measurement study of post-quantum readiness of internet. arXiv preprint arXiv:2606.16473 (2026). https://doi.org/10.48550/arXiv. 2606.16473, https://arxiv.org/abs/2606.16473 4. European Commission: Recommendation on a coordinated roadmap for the transition to post-quantum cryptography. European Commission (2024), https:// digital-strategy.ec.europa.eu 5. Google: Quantum frontiers may be closer than they appear. https: //blog.google/innovation-and-ai/technology/safety-security/ cryptography-migration-timeline/ (2026), google Security Blog. Accessed: 2 July 2026 6. Government Digital Service, Central Digital and Data Office: List of .gov.uk domain names (Jan 2025), https://www.gov.uk/government/publications/ list-of-gov-uk-domain-names, accessed: 7 April 2026 7. Holz, R., Amann, J., Razaghpanah, A., Vallina-Rodriguez, N.: The era of TLS 1.3: Measuring deployment and use with active and passive methods. arXiv preprint arXiv:1907.12762 (2019), https://arxiv.org/abs/1907.12762

8. Holz, R., Braun, L., Kammenhuber, N., Carle, G.: The SSL landscape: A thorough analysis of the X.509 PKI using active and passive measurements. In: Proceedings of the 2011 ACM SIGCOMM Conference on Internet Measurement Conference (IMC). pp. 427–444. ACM (2011). https://doi.org/10.1145/2068816.2068859 9. Koblitz, N.: Elliptic curve cryptosystems. Mathematics of Computation 48(177), 203–209 (1987) 10. Miller, V.S.: Use of elliptic curves in cryptography. In: Williams, H.C. (ed.) Advances in Cryptology — CRYPTO ’85 Proceedings. pp. 417–426. Springer Berlin Heidelberg, Berlin, Heidelberg (1986) 11. Moody’s Analytics: FAME (Financial Analysis Made Easy). Online database (2026), https://fame-r1.bvdinfo.com/, accessed: 6 March 2026 12. National Cyber Security Centre: Timelines for migration to post-quantum cryptography (2025), https://www.ncsc.gov.uk/guidance/pqc-migration-timelines 13. National Institute of Standards and Technology: Module-Lattice-Based Digital Signature Standard (ML-DSA). Federal Information Processing Standards Publication 204, National Institute of Standards and Technology (Aug 2024). https: //doi.org/10.6028/NIST.FIPS.204 14. National Institute of Standards and Technology: Module-Lattice-Based KeyEncapsulation Mechanism Standard (ML-KEM). Federal Information Processing Standards Publication 203, National Institute of Standards and Technology (Aug 2024). https://doi.org/10.6028/NIST.FIPS.203 15. National Institute of Standards and Technology: Stateless Hash-Based Digital Signature Standard (SLH-DSA). Federal Information Processing Standards Publication 205, National Institute of Standards and Technology (Aug 2024). https: //doi.org/10.6028/NIST.FIPS.205 16. Rescorla, E.: The Transport Layer Security (TLS) Protocol Version 1.3. Tech. Rep. 8446, Internet Engineering Task Force (Aug 2018). https://doi.org/10. 17487/RFC8446, https://www.rfc-editor.org/info/rfc8446 17. Rivest, R.L., Shamir, A., Adleman, L.: A method for obtaining digital signatures and public-key cryptosystems. Commun. ACM 21(2), 120–126 (Feb 1978). https: //doi.org/10.1145/359340.359342, https://doi.org/10.1145/359340.359342 18. Schaumann, J.: whohosts. GitHub repository, https://github.com/jschauma/ whohosts, commit 8196458. Accessed: 3 July 2026 19. Schaumann, J.: Sites using PQC (february 2026). Signs of Triviality Blog (Feb 2026), https://www.netmeister.org/blog/pqc-use-2026-02.html, accessed: 7 April 2026 20. Schaumann, J.: Use of PQC in smtp starttls. Signs of Triviality Blog (Jan 2026), https://www.netmeister.org/blog/smtp-pqc.html, accessed: 7 April 2026 21. Shor, P.W.: Polynomial-time algorithms for prime factorization and discrete logarithms on a quantum computer. SIAM Journal on Computing 26(5), 1484– 1509 (1997). https://doi.org/10.1137/S0097539795293172, https://doi.org/ 10.1137/S0097539795293172 22. Single Online Home National Digital Team: List of forces. data.police.uk API Documentation, https://data.police.uk/docs/method/forces/, accessed: 16 February 2026 23. Stebila, D., Kwiatkowski, K., et al.: Hybrid key exchange in TLS 1.3. IETF Internet-Draft (2025) 24. UK Learning Providers: All UK universities. Online database (2026), https:// learning-provider.data.ac.uk/all.html, accessed: 16 February 2026

25. Warburton, D., Heath, M.: The state of post-quantum cryptography (pqc) on the web. https://www.f5.com/labs/articles/the-state-of-pqc-on-the-web (Jun 2025), f5 Labs Report, Accessed: 2 July 2026 26. Westerbaan, B.: Cloudflare targets 2029 for full post-quantum security. https: //blog.cloudflare.com/post-quantum-roadmap/ (Apr 2026), cloudflare Blog, Accessed: 2 July 2026