Added territories and Antarctica reference. The basemap uses 10m world-atlas polygons. Réunion, Mayotte, Guadeloupe, Tokelau and the historical Netherlands Antilles use invisible selection targets where a separate polygon is absent. Antarctica is a schematic strip, not a scale coastline, with specifically located communications and research infrastructure records. The Netherlands Antilles appears only through its dissolution on October 10, 2010. Macau is a Chinese Special Administrative Region displayed separately for source matching. Territory statistics use separately coded observations. Sources: world-atlas.

Tuvalu map geometry. Tuvalu uses its 10m world-atlas country polygons northeast of Fiji. An invisible target makes its small atolls easier to select. Sources: 10m world-atlas.

Territorial relationships and parent-jurisdiction data. Country Context identifies territorial relationships using government and UN material. Territory statistics require separately coded records. For Guadeloupe, Réunion and Mayotte from 2011, four nationally governed cyberlaw categories use explicitly labeled French legal application under Constitution Article 73. Indirect taxation remains territory specific.

Google Play developer counts and locally published apps. Developer counts use dated 42matters observations for 2023–2025 and AppBrain for 2026, labeled by provider and observation year. Local app discovery uses a separately checked register of 3,443 country origin apps, with publisher and install bucket evidence.

Mobile access, network generation, Android versions and Play registration. Mobile subscriptions use World Bank indicators sourced from ITU, in total and per 100 people. Phone ownership uses ITU observations processed by Our World in Data and requires a device with an active SIM. Network coverage measures the population within range of 2G, 3G, 4G or 5G. The newest observation at or before the selection is dated beside the value. Android versions use Statcounter monthly page views. Google Play registration eligibility follows Google's supported locations table, checked September 19, 2026; omitted locations are labeled ‘Not listed.’ Sources: total subscriptions; subscriptions per 100 people; population owning a mobile phone; population within range of 2G, 3G, 4G or 5G; Mobile Android Version; supported locations table.

Mobile-network coverage completion update. Coverage histories use ITU observations processed by Our World in Data, updated July 31, 2026. Each network generation uses its latest observation within the selected year or preceding five years and displays its actual year. Kosovo uses its separate source code. Taiwan's 2021 and 2022 5G observations come from its National Communications Commission.

Antarctica mobile-network exception. Antarctica's network rows describe documented station area deployments. Population coverage percentages are not published. Entel reports Antarctic 5G expansion and a gradual 2G shutdown. Sources: 5G network expansion in Antarctica.

Small-country and all-country evidence enrichment. Country profiles link to Internet Society Pulse reports. Source dated evidence for 21 microstates, island states and territories includes available resilience, routed networks, upstream diversity, IPv6, local content and digital public service readiness. Current network observations appear from 2026; UN readiness observations appear from 2024. Sources: Internet Society Pulse country-report.

National Digital Ecosystem history. Annual ecosystem indicators combine World Bank and ITU access measures with network coverage. Pulse supplies 1,432 resilience observations for 179 geographies and 1,329 annual IPv6 observations for 199 economies, both covering 2019–2026. UN DESA supplies 1,544 EGDI observations across 193 member states in biennial releases from 2010–2024, displayed on a 0–100 scale. RIPEstat supplies 3,184 year end routed network observations for 199 countries from 2011–2026; Kosovo has no separate source row. Values display their observation year, and changes compare observations within the same series. World Bank and ITU fallback is limited to five years. Sources: Internet Society Pulse Internet Resilience Index; Pulse country reports; E-Government Development Index country tables; RIPEstat Country Resource Stats API.

China-linked digital infrastructure. AidData's 2025 release supplies adjusted commitments in constant 2023 US dollars for 2000–2023, with implementation information through 2025. The map filters for aggregation recommended records, communications sector 220 and physical infrastructure: 377 country records across 98 geographies plus five regional records. Of the country records, 338 have adjusted amounts and 49 code a Chinese group ultimate beneficial owner of the recipient. Regional amounts remain separate. ASPI's September 17, 2026 snapshot supplies infrastructure and qualifying business records; undated rows enter from its 2021 release and remain labeled undated. Cable evidence separates named ownership from HMN Tech or ZTT construction and includes 48 systems across 96 country relationships. Provider inventories checked September 19, 2026 add 56 cloud footprint records across 28 countries: 17 Alibaba regions, 27 Alibaba CDN country footprints and 12 Tencent regions. Published Alibaba region opening years are retained; current CDN and Tencent inventories enter only in 2026. Additional operator and project records retain their dates and agreement or operational status. Map shading counts cumulative evidence records; finance totals use AidData commitments only. Sources: China’s Global Loans and Grants Dataset, Version 1.0; Mapping China’s Technology Giants; Alibaba Cloud public-cloud regions; Tencent Cloud regions; ZTE news register; Syria telecom investigation.

Iran-linked digital infrastructure. The register checked September 21, 2026 includes EPEG, Pishgaman Oman Iran, Syria's mobile license agreement and Wafa ownership investigation, banking network links and ArvanCloud's UAE affiliate. The current CDN inventory contributes 24 foreign provider country footprints across 37 cities, appearing only in 2026. EPEG is assigned once to each of nine named foreign route countries; its approximately US$200 million system cost remains descriptive. Syria entries retain agreement or investigated ownership status. The banking messaging connection retains its ‘scope unclear’ label. Map shading counts cumulative country assigned evidence records. Sources: Europe Persia Express Gateway (EPEG); Pishgaman Oman Iran cable; Iran’s 2017 Syrian mobile-license agreement; Wafa Telecom ownership investigation; 2023 Iran–Russia banking-message-network connection; ArvanCloud Global Technologies L.L.C.; current CDN network map.

Russian-linked digital infrastructure. The September 17, 2026 register covers telecom operations, fiber, transit, data centers, carrier presence, named vendor customers, identified SORM or DPI deployments, GLONASS, satellite services, public systems and named owner cables. Records retain active, completed, ended, customer, supplier or agreement status. Documented sales, separations, license withdrawals and retirements supply end dates, including RJK and SEA-ME-WE 3. Protei customer records specify the product only where the source identifies it. Crimea landing records are assigned to Ukraine using internationally recognized boundaries. Map shading counts cumulative dated evidence records. Sources: MTS reports; German Marshall Fund’s 2025 Central Asia technology-stack study; Recorded Future’s January 2025 Russian SORM export assessment; Citizen Lab’s Iran mobile-network investigation; NtechLab public-company record; GLONASS overseas-station review.

US-linked digital infrastructure. DFC data through June 30, 2026 supply the summed US dollar commitments after screening 1,085 active projects for digital infrastructure. Regional amounts remain separate. USTDA records checked through September 14, 2026 identify project development, studies and pilots. Nineteen official company datasets supply 687 footprint records, with 681 mapped across 170 geographies; most enter in their 2025 or 2026 observation year. Meta retains published groundbreaking dates, and Starlink contributes 133 available or launched service footprints. Provider operating or development status is retained. The operational root server directory contributes 162 mapped country records aggregated from US operated root identities. Cable records use named owner fields and ready for service years. The United States is excluded as domestic scope. Map shading counts cumulative qualifying records across these separately labeled evidence types. Sources: DFC transaction data; USTDA announcements; Fastly; Hurricane Electric; Cogent; Akamai; Google owned data centers; Meta data centers; Starlink official availability data; Root Server Technical Operations Association.

National and sovereign operating systems. The history, checked September 20, 2026, begins with the earliest documented release, national adoption or active state supported program and retains later milestones. Anchors include Pardus 2005, BOSS 2007, Red Star 2008, Nova 2009, Astra 2010, Canaima adoption 2011, Huayra 2013, HarmonyOS 2019, BharOS 2023, GendBuntu 2008, HUMAIN ONE 2025 and DINUM NixOS workstreams 2026. ‘Operational’ identifies a released system with documented deployment or support; ‘Development / rollout’ identifies an active build, pilot or migration. Red Star's classification records documented deployment history; current installed base and support figures are unpublished. Sources: GendBuntu deployment record; DINUM workstation program; Sécurix; Bureautix; Astra Linux; Aurora OS; 2019 HarmonyOS launch history; OpenHarmony release record; openKylin; BOSS GNU/Linux; January 2023 BharOS announcement; independently examined Red Star OS 3.0; Red Star OS 4.0 technical reporting; Pardus; Huayra history; Canaima history and 2011 adoption record; Nova GNU/Linux; government-deployment report.

Mobile ecosystem · platforms and device brands. The 2026 operating system sales tables for China, India and the United States remain separate from Statcounter web traffic. Device vendors use Statcounter's beta series, which identifies devices from browser user agent page views. Named brands are rebased to classified traffic; Unknown and Other remain on the coverage line. Sources: Statcounter mobile operating-system web traffic; mobile device-vendor; Statcounter measures page views.

Foreign Investment map. The map uses annual World Bank net FDI inflows in current US dollars for the exact selected year through 2025. The 2026 view displays the 2025 dataset, labeled by observation year. Taiwan separately identifies CBC BPM6 net direct investment liability and asset transactions. Negative values represent net divestment. Country details label annual flows, GDP shares, five year flow sums, OECD bilateral flows and IMF end period positions. PPI sponsor origin shares retain their fixed 2015–2024 period. FDI is economy wide. Sources: foreign direct investment, net inflows (BoP, current US$); 2024 PPI Annual Report.

Google Play developers and the Apps timeline. Google Play developer counts use dated 42matters observations for 2023–2025 and AppBrain for 2026. Earlier years show ‘Not published’ where a comparable count is unavailable. Country assignment follows the developer's listed account location. Source and observation years remain visible for each snapshot.

Digital adoption, population and structural access. Access and population data use World Bank and ITU annual indicators. The current 2025–2026 adoption view combines 183 World Bank or ITU observations from 2022–2025 with separately labeled 2025 estimates in a 200 country snapshot refreshed September 17, 2026. Estimates use DataReportal or Kepios reports, with Internet Society Pulse for Vatican City; North Korea is an upper bound below 0.4%. Historical years retain dated annual observations. Country metadata and capital overrides use World Bank, UN and official records. Equatorial Guinea reflects the January 2, 2026 Ciudad de la Paz decree; Jakarta remains Indonesia's capital pending the transfer decree. Palestine distinguishes the East Jerusalem claim and Ramallah administrative center. The 2025 population fallbacks are 5,413,596 for Palestine and 1,576,876 for Kosovo. Values retain their provider and observation year; mobile subscriptions are connections per 100 people. Sources: Individuals using the Internet; population; mobile-cellular subscriptions; fixed-broadband subscriptions; access to electricity; UN Data country profiles; 2 January 2026 decree naming Ciudad de la Paz; Jakarta remains the capital until a presidential transfer decree takes effect; Palestinian Basic Law’s capital declaration; UN’s description of the East Jerusalem claim; Council of Ministers and Prime Minister’s Office location in Ramallah; World Bank series; Myanmar; Taiwan; Cook Islands; Eritrea; Greenland; Kosovo; North Korea; Vanuatu; Timor-Leste; South Sudan; Sudan; Syria; Turkmenistan; Tuvalu; Samoa; Yemen; Internet Society Pulse country estimate.

Technology infrastructure & investment. World Bank indicators cover ICT private participation commitments, annual inward and outward FDI, GDP shares and gross fixed capital formation. The 2020–2024 PPI share uses ICT, energy, transport and water sectors; project details come from the 1990–2024 release covering 137 low and middle income economies. IMF bilateral positions use official reporter data when available and separately labeled SCC mirror estimates otherwise. The September 18, 2026 snapshot contains inward and outward histories for 199 entries. Ten positive counterparts are displayed, while totals and concentration use all positive published counterparts that are World Bank economies. Positions are economy wide end period stocks by immediate counterpart. Annual flows and five year flow sums retain negative values and observation counts. Project evidence groups data centers, cloud, fiber, broadband, mobile, satellites, exchanges and cables, with source dates and proposal or operational status retained. Country specific sources include official operators and government records. Sources: ICT projects with private participation; FDI net inflows in current US dollars; FDI net inflows as a share of GDP; FDI net outflows in current US dollars; FDI net outflows as a share of GDP; gross fixed capital formation in current US dollars; gross fixed capital formation as a share of GDP; energy; transport; water and sanitation; 2024 Global Report as its latest annual release; 1990–2024 project-level file; Direct Investment Positions by Counterpart Economy; SDMX API; PPI data; methodology; PeeringDB interconnection registry; API documentation; Submarine Cable Map; methodology FAQ; Andorra Telecom; EriTel; TIM San Marino; Turkmenistan communications ministry; Turkmentelecom; Vatican City State.

Antarctica-specific infrastructure investment. Antarctica's project register separately labels the Zhongshan ground station contract, NSF planning grant, Chile cable feasibility work, proposed construction estimate, Scott Base redevelopment and technology research funding. Amounts retain their original currencies and scopes. Proposed, mixed purpose and research funding are excluded from the active infrastructure map score. Sources: Zhongshan ground-station contract; NSF project-development grant; Chile–Antarctica cable feasibility work; Scott Base redevelopment; adjacent Antarctic technology-research funding.

Immediate-counterpart investment timeline. OECD bilateral FDI flows, saved August 7, 2026, supply exact year inward and outward transactions for 2011–2024. Reporter observations take precedence over labeled counterpart mirrors. Lists show up to ten positive annual flows, each dated. The 2025–2026 views retain the latest earlier flow list with its actual year. A smaller line shows the same pair's latest available IMF year end position, labeled by year and mirror status. Annual flows and end period investment stocks remain separate measures. Sources: OECD bilateral FDI flows; IMF DIP.

Android research benchmarks. The expanded research catalog transcribes 23 market, preinstalled and malware corpus rows, 34 named distribution labels, 33 Google Play category rows, a 28 tool capability matrix and an 11 unpacker chronology. The study analyzed 2,450,338 APKs, reporting 50,664 packer, 45,320 obfuscator and 185 protector detections. Detection types may overlap; percentages apply to the sampled corpora. Sources: Third Party Tracking in the Mobile Ecosystem; PermPress; Practical Android Software Protection in the Wild; technical summary and detailed prevalence tables.

Android packer evidence and timeline. The 2025 To Unpack or Not to Unpack study detected packing in 4,652 of 7,913 apps from China's 360 App Store collected June–August 2024 (58.8%), versus 83 of 4,428 APKPure apps in its non Chinese sample (reported as 2%). These are distribution channel samples. Protection tiers and Frida/JDB outcomes retain the study's scope. Tencent Enterprise is displayed as 11/7,913 (0.14%) because the appendix's printed 3.10% conflicts with its count. Companion research supplies market, family, category and malware tables. Target SDK charts are approximate API cohort counts grouped by first release year; unpacker releases remain a separate chronology. Packer detections alone do not establish malware. Sources: To Unpack or Not to Unpack; ACM Computing Surveys study; open dataset.

Distinct mobile tracking SDK prevalence and supplier geography. SDK Geography uses the 2017 Kollnig cohort at 2017 and Oxford's ten country Table 3 published in 2018 as a fixed baseline for 2018–2026. Oxford measured 959,426 free apps from the US and UK Google Play stores. Later timeline positions carry the same published counts and rounded shares. Countries absent from Table 3 show ‘None reported.’ The 2020 study and 2026 Exodus audit remain separate dated profile observations, and named company records appear from their source years. Sources: The Oxford paper; 2020 Kollnig study; Exodus Privacy reports.

Dominant third-party SDK supplier jurisdiction. The frozen September 20, 2026 audit checks up to twenty country origin Google Play apps per country against exact package IDs in Exodus Privacy reports. It contains 3,449 apps across 193 entries: 1,818 have public reports and 1,698 have recognized trackers, producing 12,783 attributed app supplier presences. Of 168 tracker signatures, 166 map to 123 companies; ACRA and Solar2D remain open source projects without one verified jurisdiction. Repeated products from the same supplier within an app count once. The headline jurisdiction has the most deduplicated app supplier presences, with ties labeled mixed. Jurisdictions follow documented operating company or parent headquarters. Public reports cover 175 countries and attributed suppliers 173; unmatched reports remain visible. The sample appears only in 2026. Recognizable SDK code does not establish execution or data transmission. Sources: Exodus Privacy application reports; ACRA; Solar2D; completed-acquisition announcement.

Expanded network-infrastructure records. Network indicators include World Bank fixed broadband totals and secure servers per million people, dated within the selected year or limited fallback window. The root server register checked September 18, 2026 contains 1,573 operational site rows and 2,029 instances across 178 published geography codes. Suspended and decommissioned rows are excluded. Country details show instances, sites, IPv6 enabled instances, operators and locations. PeeringDB supplies current registered network and ASN counts and selected names. Sources: fixed-broadband subscription total; secure Internet servers per 1 million people; Root Server Technical Operations Association site register; documented API.

Technology Infrastructure timeline. Shading counts qualifying project and operating records active during the selected year, together with matching year PeeringDB facility and Internet exchange rows for 2010–2025 and operational Root DNS site rows for 2015–2025. Projects enter by their published start, commitment, ready for service or verification date and leave at documented closure or retirement. Proposals and canceled records retain their status labels in country cards. Annual network indicators and the current 2026 registries appear as country context. Root DNS country cards for 2010–2014 use the archive’s listed site and location row units, with their within year snapshot dates; later archives distinguish site rows from server instances. The map uses a fixed cross year color scale capped at the highest annual 95th percentile. PPI’s latest active project snapshot has limited evidence for projects that ended before that release.

SDK geography timeline interpretation. There is no mapped supplier country study for 2010–2016. The 2017 position uses the Kollnig cohort; 2018–2026 positions use Oxford's fixed 2018 Table 3 baseline. Source year and selected year are shown separately. Named company records appear from their documented years. The 2020 study and 2026 audit remain dated country context.

Cyber-incident origin timeline and record detail. The Malware map uses EuRepoC Global Dataset v1.3.2, released in 2025 with 3,416 reviewed incidents from 2000–2024. Annual counts deduplicate settled, exact country incident–initiator pairs by incident start year. Profiles contain 1,471 qualifying detail records across 79 countries with available attribution, initiator, target and technical fields. The 2025–2026 views retain 2024 data. The World Cybercrime Index remains a separate 97 country offender origin baseline for 2021, published in 2024. ITU GCI 2024 separately records national cybersecurity commitments from 2024 onward. Sources: EuRepoC Global Dataset of Cyber Incidents v1.3.2; World Cybercrime Index; ITU GCI 2024.

Threat-actor and malware-use registers. MISP Threat Actor galaxy version 350, checked September 18, 2026, supplies 482 country tagged records across 35 countries, retaining available aliases, confidence, suspected sponsor, targets and references. A crosswalk to MITRE ATT&CK Enterprise and Mobile STIX 2.1 supplies 640 documented actor software use relationships across ten country tags and 409 software entries. Actor aliases and activity clusters can overlap. These current knowledge base snapshots remain reference context rather than annual observations. Sources: MISP Threat Actor galaxy; MITRE ATT&CK STIX 2.1 Enterprise and Mobile datasets.

Mobile-malware origin evidence. Commercial mobile spyware origin evidence covers 16 named developer or vendor group records across ten countries, supported by the cited technical investigations, corporate research and government designations. Profiles retain entities, country bases and source dates. Cytrox and Intellexa are distinct disclosed entities associated with Predator. Sources: Google Threat Analysis Group’s 2024 Buying Spying report; Citizen Lab’s QuaDream investigation; Citizen Lab’s Candiru investigation; Citizen Lab’s 2025 Paragon investigation; the U.S. Treasury’s Intellexa and Cytrox designations; Amnesty International’s Wintego investigation; Atlantic Council corporate-registry research; Kaspersky’s FinSpy mobile analysis.

Internet controls, shutdowns and cyberlaw. Freedom on the Net scores run from 0 to 100 and ranks cover assessed countries; 2026 uses the 2025 report. DSP v8 provides the filtering baseline, using its 2025 row for 2026. STOP supplies 2,102 documented records for 2016–2025, revised through September 16, 2026, with new and continuing events distinguished. Its 2026 view combines ongoing 2025 records with separately verified 2026 events. OONI aggregates begin in 2012, with partial 2026 totals and live named target checks. UNCTAD's 2026 snapshot covers 195 countries and five legal areas. Historical 2011–2025 legal shading counts current linked instruments with recoverable title years at or before the selection; undated categories remain unknown. The 2026 view uses current tracker status. Taiwan, Cook Islands, Greenland, Hong Kong and Kosovo have no separate tracker row. Kosovo's additional legal context uses its Official Gazette. Sources: Freedom House Freedom on the Net reports and methodology; QoG Freedom House time series; Digital Society Project Dataset v8 data and codebook; #KeepItOn campaign; STOP data dashboard; official 2016–2025 workbook; methodology; OONI Explorer; public measurement API; UNCTAD Global Cyberlaw Tracker; Law No. 06/L-082 on Protection of Personal Data; Law No. 08/L-173 on Cyber Security.

Current social-media restrictions and child-account rules. The current status review, checked September 18, 2026, distinguishes general platform blocks, feature restrictions, enacted age limits, transitions and proposals. Australia uses eSafety's named services; India uses official app orders; UK child safety context uses Ofcom. The EU Kids Act proposal dated September 17, 2026 remains labeled proposed. Temporary and historical events appear separately in the selected year chronology. Sources: Freedom House’s 2025 Key Internet Controls; OONI measurements and findings; Access Now #KeepItOn; eSafety Commissioner’s official age-restricted-platform list; Ofcom’s child-safety rules.

Shutdown and service-block links. The STOP snapshot contains 2,102 records across 100 countries for 2016–2025, with reporting links on 2,069 records. Of these, 1,105 contain multiple reporting links and 311 include an Access Now link. Selected country and year references appear in the reference register.

Curated policy-event references. OONI findings, Iran report, Russia report, China finding, and OHCHR shutdown submission; Government of India notices from June 2020, September 2020, and November 2020, plus the December 2022 Lok Sabha answer on blocked Chinese applications; Access Now reporting on Sudan, Sudan’s WhatsApp call block, and South Sudan; and the US State Department DPRK report. Record-specific links are collected in the selected-country reference register below.

Who restricts whom. Who restricts whom covers specific digital and closely related measures supported by official law, current guidance or registers, with firsthand network evidence identified separately. Each entry has sources and an effective period. Historical years show measures active at any time during that year; 2026 includes only measures verified in force on October 14, 2026. Ended or superseded phases stop at their operative end dates. Unconfirmed current measures are limited to supported historical periods. EU incoming measures count once; member records follow membership, including the UK transition ending December 31, 2020. Platform and company attribution follows the documented operator or jurisdiction. Map shading continues to show the separate internet filtering indicator.

Basemap and general limitations. Country geometry uses World Atlas and Natural Earth. Source dates and geographic definitions are shown where relevant. Gray or ‘Not published’ identifies missing source observations. Sources: World Atlas / Natural Earth.

Expanded cyberlaw evidence. UNCTAD's linked instrument register supplies named laws under five legal areas. UNODC and the UN Treaty Collection supply convention signature and ratification, acceptance or accession dates, checked September 18, 2026. The convention was adopted in December 2024 and opened for signature in October 2025. Sources: official linked-document register; UNODC Convention against Cybercrime overview; UN Treaty Collection participant table.

Russian-linked infrastructure expansion checked 19 September 2026. The September 19, 2026 expansion uses operator maintained PeeringDB registers for Rostelecom, MTS, MegaFon, TransTeleCom and VimpelCom / Beeline, with one record per foreign operator country pair. These interconnection snapshots appear only in 2026. Additional dated evidence covers Kaspersky transparency facilities, NtechLab customer presence, KazSat-3, Yandex Cloud Kazakhstan and Russian manufacture or launch services involving Iran's satellites. Each entry retains its evidence type and status. Sources: Rostelecom AS12389; MTS AS8359; MegaFon AS31133; TransTeleCom AS20485; VimpelCom / Beeline AS3216; PeeringDB documentation; Global Transparency Initiative status history; Kigali facility release; published NtechLab country-footprint list; KazSat-3; Yandex Cloud Kazakhstan region; Pars-1 and earlier Khayyam cooperation; Kowsar and Hodhod launch; Nahid-2 launch.

Russian operator global context not converted into country records. Rostelecom reports access to over 150 operator networks in 70 countries and cooperation with 600 fixed and mobile operators, including international cables, TEA and EPEG. These company wide figures remain context; country records use separately identified locations and dates. Sources: international-cooperation page.

EU-linked digital infrastructure layer checked 21 September 2026. The EU layer covers physical and operational digital infrastructure tied to EU institutions, Team Europe, joint undertakings or companies headquartered in member states. Cross border and joint public projects can appear within EU countries; ordinary company home country footprints are excluded. Sources include Global Gateway, EuroHPC, OVHcloud, Hetzner, Orange and named owner cable records. Current inventories enter in their observation year. Proposals, studies and construction stages retain their status. Shading counts country source evidence relationships. Sources: Global Gateway digital-project catalog; EuroHPC JU supercomputer inventory; OVHcloud location inventory; Hetzner data-center register; 2025 Universal Registration Document; TeleGeography Submarine Cable Map.

21 September 2026 correction audit. The September 21, 2026 cable review covers 697 systems and 1,562 country system relationships across 157 mapped locations. Planned systems appear as 2026 evidence with future service dates and are excluded from active shading. Systems operational before 2010 appear from the timeline minimum with their actual dates. Device vendor, Android version and iOS version traffic remain separate measures. Dated ASPI and operating system records retain their original source years.

21 September 2026 follow-up audit. The September 21, 2026 review added Iran's UAE and Kuwait cables, Pakistan terrestrial fiber, Mir Business Bank, the initial Shetab–Mir stage and labeled Oman and Azerbaijan corridor agreements. Pre 2010 infrastructure retains its actual evidence date from the timeline minimum. App origin lists reflect verified publisher locations. Oxford's US Table 3 count is 865,369.

21 September 2026 table review. ASPI dates recovered from record text or supporting sources place 132 records, representing 140 country relationships, in their supported evidence years. Remaining undated records enter from the 2021 source release. MyMCI and EWANO remain excluded from Oman and Iran origin lists while publisher address and product origin evidence conflict. The September 20 supplier audit remains a separately dated corpus.

Greenland territory expansion · 22 September 2026. Greenland uses separate territory observations: Statistics Greenland population, World Bank and ITU access indicators, DataReportal adoption reports from 2020 and recent Tusass subscriptions and network milestones. Provider and observation dates remain visible. Operator coverage footprints retain their reported form. Greenland's 2016 personal data law is separate territory evidence outside UNCTAD's table; four other legal categories remain unassessed. Infrastructure, apps and network observations use Greenland records.