SEABED TECHNOLOGY REVIEW
Telecommunications & Digital Infrastructure Intelligence
Security Supplement • March 2026
Infrastructure Security & Resilience
Physical Resilience, Geographic Redundancy, Island Breaks, and Cyber Threats in Pacific Submarine Cable Architecture
■ BLUF — Bottom Line Up Front
The 14,215-km Honomoana transpacific trunk faces a layered threat environment encompassing three distinct risk categories:I. The Fault Baseline: How Often Do Long Cables Break?
Before evaluating the specific threat posture of Honomoana, it is necessary to establish the statistical baseline against which any cable system must be measured. The International Cable Protection Committee (ICPC) reports approximately 100 to 200 cable faults per year on the global submarine telecommunications network — a rate that has remained broadly stable for more than a decade, despite the rapid growth in installed cable kilometers. A 2025 analysis by the ICPC cited by multiple sources confirms roughly 200 faults annually in 2025 alone across 597 active and under-construction cable systems.
The cause distribution is well-established and consistent across analytical sources. According to the Congressional Research Service's 2023 report on undersea cable protection, approximately 75% of cable breaks result from human activities — primarily fishing gear and ship anchors — with natural hazards (earthquakes, submarine landslides, turbidity currents) accounting for about 14% and equipment failure for roughly 6%. A 2024 Recorded Future (Insikt Group) analysis of 44 publicly reported cable damage events in 2024–2025 found that unknown causes accounted for 31% (many of which may have involved anchor drag with opaque attribution), anchor dragging for 25%, and seismic or natural phenomena for 16%.
For a 14,215-km cable with approximately 200 repeaters spaced at roughly 70-km intervals, the exposure surface is significant. Each repeater represents a potential equipment failure point, and the cable's geographic extent crosses multiple seabed environments of varying risk — shallow shelf zones near California and Australia where anchoring and fishing activity are concentrated, and deep abyssal plain in the central Pacific where the predominant risk is seismic or turbidity-current related. The design life expectancy is 25 years, during which industry benchmarks suggest a well-engineered transoceanic cable should require fewer than two ship repair operations attributable to submerged equipment failure alone — but external physical events are a separate matter entirely.
- ~100–200 faults/year globally (ICPC); ~200 reported in 2025
- 75%+ of breaks caused by human activity (fishing, anchoring)
- Fishing: ~50% of identified faults (ICPC)
- Anchor drag: ~25% of identified faults (2024–25 Insikt data)
- Seismic/natural: ~14–16%
- Equipment failure: ~6%
- Fewer than 100 cable repair vessels worldwide (global fleet); most are over 20 years old
- Repair staging for a mid-Pacific fault: typically 2–4 weeks from vessel dispatch to completion
II. The Physical Threat Environment Along the Honomoana Route
A. Shore Approaches and the Shallow-Water Vulnerability Window
The most statistically dangerous segments of any cable system are the shore approaches — the relatively shallow zones within the first 200 nautical miles of each coast, where shipping traffic, commercial fishing, and recreational and commercial anchor activity are concentrated. These areas are also, not coincidentally, where cables are most accessible for deliberate interference without the technical complexity required for deep-ocean operations. At Carlsbad, the cable terminates in a horizontal directional bore beneath the surf zone, emerging 3,000 feet offshore — a standard design intended to push the accessible cable end beyond the typical recreational anchor zone, though well within the commercial shipping lane environment offshore Southern California.
The Australian shore approaches at Torquay, Victoria, and Maroubra, New South Wales, face analogous risks. The Bass Strait — through which approaches to Melbourne must pass — is a notoriously active fishing ground. The SUBCO-Google shared landing station arrangement at these two sites is partly motivated by the efficiency of concentrating shore protection and monitoring at a single physical point, reducing the number of exposed cable approach corridors that must be actively managed.
B. Deep-Water Seismic and Turbidity Hazards
The central Pacific route of Honomoana traverses one of the geologically more benign portions of the Pacific seabed in terms of turbidity current risk — unlike the more active tectonic margins of the western Pacific or the earthquake-prone zones of the Southwest Pacific. However, the route does skirt the seismically active region around French Polynesia, and the branch segments to Tahiti Nui and Tahiti Iti are positioned in a zone where underwater volcanic activity has historically damaged regional cables. The 2022 Hunga-Tonga volcanic eruption destroyed Tonga's sole submarine cable, leaving the island's 100,000 residents in digital isolation for more than a month. The cable broke again two years later, underscoring the long-term vulnerability of systems transiting seismically active zones.
C. State-Sponsored Gray-Zone Anchor Drag
The most operationally significant threat evolution in the past 24 months is the demonstrated use of commercial vessels — operating under the cover of routine maritime traffic — to conduct anchor-drag attacks on submarine cables as a gray-zone tactic. The pattern is now well-documented in the Baltic Sea, where a 2025 Insikt Group analysis of 44 cable damage events in 2024–2025 attributed at least four to Chinese- or Russia-linked vessels operating with "opaque ownership structures" or "suspicious maneuvers near damaged cables." The technique is technically simple, requires no specialized equipment, and is designed to maintain plausible deniability because UNCLOS Article 113 assigns jurisdiction for cable damage offenses to the flag state of the vessel — an enforcement gap that has not been closed by any binding international agreement.
The Pacific threat environment differs from the Baltic in important structural respects. The Baltic is a semi-enclosed sea with dense shipping traffic and relatively shallow water throughout, making anchor-drag attacks easy to execute and difficult to attribute. The transpacific route traverses vast areas of open ocean at depths of 4,000–6,000 meters — far below the operational range of any ship's anchor under normal conditions. The highest-risk segments remain the shore approaches and shallow shelf zones, which mirrors the Baltic experience. The broader Pacific region has seen multiple cable incidents around Taiwan — five in 2024–2025 alone, according to Insikt data — with Chinese vessels operating in the area during at least some of these events.
For Honomoana specifically, the primary gray-zone exposure points are: (1) the Carlsbad shore approach, in proximity to one of the world's busiest commercial shipping corridors; (2) the approaches to the Australian landings in Victoria and New South Wales, transiting Bass Strait and the Australian continental shelf; and (3) the French Polynesia branch terminations, where the cable enters shallow coastal water around Tahiti. The deep abyssal trunk is not immune to sophisticated state-actor interference, but the technical challenge is substantially greater.
III. The Island Break Question: Resilience Asset or Vulnerability Node?
The question of whether intermediate island landings — in Hawaii, French Polynesia, Fiji, or the Solomon Islands — enhance or complicate the security and resilience of a long transpacific cable requires disaggregating two distinct functions that such breaks serve: traffic restoration and repair staging. The answer differs materially depending on which function is being evaluated, and further depends on the geopolitical character of the island in question.
A. Traffic Restoration: The Case for Island Breaks
An island break converts what would otherwise be a single point of failure on a trunk cable into a set of separate segments with independent failure modes. If the eastern segment of a cable (US to Hawaii) is severed, traffic can continue to flow on the western segment (Hawaii to Australia) — provided the island has sufficient onward connectivity to alternative systems to carry the diverted load. This is the function that Hawaii performs in the Tabua system: Google's April 2024 announcement extended the Tabua cable to include an Oahu segment, with Google's director of Asia Pacific network planning explicitly describing Hawaii as becoming "a key digital hub in the Pacific." Hawaii's lieutenant governor noted that two of the state's three existing transpacific fiber links were approaching end of operational life, making the Tabua Hawaii extension a resilience investment as much as a commercial one.
French Polynesia's role in the Honomoana architecture serves a similar function. The two Tahiti branch segments connect to a seabed branching unit on the main transpacific trunk, and the South Pacific Connect ring — connecting the Honomoana and Tabua trunks via an interlink cable between Fiji and French Polynesia — creates a closed loop. In a fault scenario affecting the US-side portion of Honomoana, traffic can in principle be rerouted eastward through French Polynesia and Fiji on the Tabua path, and vice versa. Google Cloud VP Brian Quigley described this explicitly as "one of the first projects of its kind in the Pacific, providing the ability to bring redundant international connectivity to a region that is susceptible to natural disasters."
The Tabua system provides a further resilience design feature that deserves technical attention: it terminates at two physically separate landing points on the Fijian island of Viti Levu — on both the east and west sides of the island. According to Submarine Networks documentation, this dual-island landing "provides path diversity, redundancy in island connectivity in case of a single branch failure, and greater resiliency by supplying the cable with single end power source capability." This is sophisticated resilience engineering that directly addresses the single-point failure risk that made the Tonga incident so severe.
B. The Solomons: Strategic Asset or Geopolitical Liability?
The Solomon Islands occupy a genuinely strategic geographic position in the Pacific cable network — situated roughly mid-path between Sydney and Guam, they could serve as a natural transit node in a more distributed "mesh" architecture of the kind proposed by University of Auckland researchers writing for APNIC. However, the security calculus has become considerably more complex since 2022.
The Australian government's track record on this issue is instructive. When the Asian Development Bank originally offered to finance a cable connecting the Solomons using Huawei Marine Networks, Australia objected and funded the CS2 itself — a AU$200 million project connecting Sydney to Port Moresby and Honiara, completed in December 2019. The stated rationale was that Australia was "unwilling to have Chinese equipment connected to its infrastructure." The US Trade and Development Agency's (USTDA) proposed Central Pacific Cable — a feasibility study currently underway — would include the Solomons in a mesh connecting American Samoa, Cook Islands, Fiji, Guam, Kiribati, Marshall Islands, Micronesia, Nauru, Papua New Guinea, Samoa, Solomon Islands, Tuvalu, Vanuatu, and Wallis and Futuna. Whether that project can proceed with the Solomons included, given the current security treaty with Beijing, is an open question that USTDA has not publicly resolved.
The general principle that emerges from the Solomons example is that the security value of an island intermediate node is inseparable from the security governance of the landing station infrastructure at that node. A landing station in a jurisdiction with robust rule of law, allied security relationships, and effective exclusion of adversary access to physical plant is a genuine resilience asset. A landing station in a jurisdiction where an adversary power has negotiated security access rights is potentially the opposite.
C. Repair Staging: The Practical Value of Intermediate Nodes
The global cable repair fleet consists of fewer than 100 vessels worldwide, most of them aging — the majority over 20 years old, and only one major Pacific repair vessel (KDDI Cable Infinity) built after 2010. A mid-ocean fault on the Honomoana trunk — at, say, the midpoint between California and French Polynesia — would require a repair ship to travel approximately 3,500 km from the nearest port with repair capabilities. The Japan-based Yokohama Zone maintenance arrangement, established under a 1997 multinational agreement, provides two standby vessels equipped with ROVs and spare parts for rapid response across the Asia-Pacific region. However, "rapid" in this context means days to weeks, not hours.
An island intermediate node reduces the maximum repair-mobilization distance by providing a potential forward staging base for cable repair equipment and personnel. Hawaii is the most valuable such node for the US-side of the transpacific route: it is a major US military logistics base with existing port infrastructure capable of supporting cable ship operations, and Google's decision to extend the Tabua cable to Oahu makes Hawaii an active node in the South Pacific Connect architecture rather than a simple geographic waypoint.
French Polynesia provides analogous staging value for the central Pacific segment of Honomoana. Papeete, the capital of French Polynesia, has deep-water port facilities and is already an established supply point for Pacific maritime operations. The existing OPT cable infrastructure at French Polynesia — including the Honotua (2010) and Manatua (2020) systems — means that local technical expertise in cable operations already exists in-country.
IV. The Cybersecurity Dimension: From Tempora to Today
A. Physical Tapping at Landing Stations
The cybersecurity threat to submarine cables operates at a fundamentally different layer from physical sabotage. While physical attacks seek to deny service, signals intelligence (SIGINT) exploitation seeks to access the content and metadata of traffic flowing through the cable — ideally without the cable owner's knowledge and without causing any service disruption that would alert operators to the presence of the tap.
The most thoroughly documented mechanism for cable SIGINT exploitation is the installation of optical splitters or "intercept probes" at cable landing stations, with the cooperation (voluntary or legally compelled) of the cable operator or landing station host. This is the technique described in detail in the Snowden disclosures of 2013, which revealed that GCHQ's TEMPORA program was tapping 18 or more international submarine cables landing in the United Kingdom, sharing the collected data with the NSA under the INCENSER program. The method involved inserting a device that extracted a small percentage of the optical signal — typically via a fiber splitter — from each fiber pair in the cable, feeding the extracted signal to a processing center for bulk collection and filtering. GCHQ worked with Verizon Business, BT, Vodafone, and other carriers as "intercept partners" for this purpose.
The Snowden disclosures further confirmed that the NSA operated analogous facilities at landing stations in the United States, and that the USS Jimmy Carter submarine had been modified to access cables at repeater regeneration points in locations "where stations that receive and transmit the communications are on foreign soil or otherwise inaccessible." Cold War precedent for this approach dates to Operation Ivy Bells, a CIA/NSA/Navy program running from 1971 that successfully tapped Soviet communications cables in the Sea of Okhotsk — until NSA analyst Ronald Pelton sold the program's details to Soviet intelligence in 1980.
The security implication for Honomoana is direct: as a privately owned system with its landing station in US territory at Carlsbad, traffic transiting the cable is subject to US lawful intercept requirements under CALEA, Section 702 of the FISA Amendments Act, and the conditions of the National Security Agreement between Starfish, Google, and DHS/DOJ/DOD. The NSA's requirement is for access to be technically enabled, not necessarily that it be continuously exercised. Google's NSA compliance obligations are separate from the question of whether a foreign adversary power might attempt to tap the cable at a non-US landing station or at a mid-ocean repeater node.
B. The Adversary Tapping Threat: State Actors and the Deep-Ocean Challenge
The threat of adversary tapping — by a foreign intelligence service without the cable owner's knowledge — differs technically from the landing-station interception model. At a deep-ocean repeater node on a cable like Honomoana, the technical challenge of passive optical tapping without disturbing the fiber is significant but not insurmountable for a well-resourced state actor. Reports, denied by the US government, have consistently attributed to the USS Jimmy Carter (and its predecessor systems) a capability for exactly this kind of deep-ocean cable access. Russia is independently assessed to possess this capability through its Autonomous Uninhabited Underwater Vehicle (AGS) program — described in open-source analysis as small, nuclear-powered submersibles capable of tapping fiber cables in deep water. These vehicles have been observed operating in proximity to transatlantic cables, including in the North Atlantic near Iceland and the UK–Faeroes gap.
The Princeton Journal of Public and International Affairs notes that "espionage against submarine cables accesses transmitted data, usually without damage or notable disruption" and that "since espionage using submarine cables is internationally legal in the high seas and within a coastal state's own waters, spying operations against cables in these areas are limited only by a state's morals and technology." This legal vacuum is a structural feature of the international law framework governing submarine cables — UNCLOS Article 113 criminalizes intentional cable damage but says nothing about surveillance tapping that leaves the cable physically intact.
C. Remote Network Management: The Cyber Attack Surface
A third, distinct cybersecurity threat vector has emerged from the increasing use of internet-connected remote management systems for submarine cable network monitoring and control. The Congressional Research Service's 2023 analysis of undersea cable protection notes that "more companies are using remote management systems for submarine cable networks — tools to remotely monitor and control cable systems over the Internet — which are cost-compelling because they virtualize and possibly automate the monitoring of cable functionality. However, they may also create new risks and opportunities for cyberattack."
This is a generic IT/OT convergence problem with specific consequences for cable systems: a successful intrusion into the supervisory and control network of a cable system could, in principle, enable an adversary to disrupt service, degrade specific traffic streams, or surveil operational parameters that would reveal traffic volumes and routing patterns. The Insikt Group 2025 threat assessment states that "geopolitical, physical, and cyber threats" to submarine cables "have converged," and that the threat environment for the overall cable ecosystem "has very likely escalated" relative to the prior assessment period. The FCC's 2024 Cable NPRM — the first comprehensive rule review since 2001 — was prompted in part by the Salt Typhoon intrusion into at least eight US communications companies, attributed to Chinese state-sponsored actors, which demonstrated the degree to which adversary cyber operations against US telecommunications infrastructure had matured.
— Center for Strategic and International Studies, 2024
V. The Repeater Architecture of a 14,215-km Cable: Security Implications
Honomoana's transpacific trunk at 14,215 km will require approximately 200 optical amplifier repeaters, based on the industry-standard spacing of 50–80 km (centered on ~70 km) for high-capacity transoceanic cables. Each repeater is a pressure-sealed unit containing erbium-doped fiber amplifiers (EDFAs), which amplify the optical signal without optical-to-electrical conversion — a significant improvement over older regenerator-based designs. Power is supplied from both shore ends of the cable via a high-voltage DC conductor running within the cable, typically ±7.5 kV per shore end (±15 kV total potential difference), sufficient to power 100 repeaters from each end with the midpoint at virtual earth.
From a security standpoint, the repeater architecture has several implications. First, each repeater housing represents a potential physical access point for a tapping device; the sealed pressure vessel is designed to prevent ocean ingress, not to prevent tampered access by a technically sophisticated actor deploying a deep-submersible. Second, the power conductor that feeds the repeaters is itself a source of intelligence: monitoring the current draw on the power feed reveals information about repeater operational status, and anomalies in power consumption patterns can indicate faults or tampering. Third, the supervisory channel embedded in the optical overhead of the cable — used for order-wire communication and equipment status monitoring between shore stations — is a potential attack vector if the cable management network is compromised.
On a 14,215-km system, the "midpoint earth" zone — where the voltage of the power conductor passes through zero — is approximately 7,100 km from each shore terminal, roughly in the central Pacific between French Polynesia and Southern California. This zone is of particular interest for physical tapping operations because it is the point of minimum voltage on the power conductor, reducing the electrical hazard to any device being physically attached to the cable housing.
VI. The South Pacific Connect Ring: Engineering Response to a Layered Threat
Assessed against the threat taxonomy developed in the preceding sections, the South Pacific Connect ring topology — Honomoana plus Tabua plus the Fiji–French Polynesia interlink — represents a reasonably well-engineered resilience response to the physical and geographic threat categories. The ring creates two independent transpacific paths from California to Australia (one via French Polynesia, one via Fiji), with a southern interlink that allows traffic to transit around a failure on either trunk. Multiple Australian landings (two for Honomoana, two for Tabua, with the SUBCO SMAP system providing additional domestic diversity) mean that no single cable cut at an Australian shore approach can isolate the continent.
What the ring topology does not address is the cybersecurity threat. Two cables, both owned by the same entity (Starfish/Google), both subject to the same NSA and FCC oversight framework, and both landing at facilities managed by the same operator, do not provide diversity against a compromise of the cable management network or the landing station intercept architecture. True cyber resilience requires ownership diversity — traffic on Honomoana is most safely protected when the alternative path (Southern Cross, Hawaiki, or a future system) is operated under a different ownership and management structure, with independent security controls.
The APNIC research community has for several years advocated a more ambitious architectural vision — a distributed Pacific mesh in which island nations serve as transit and peering nodes rather than dead-end spur recipients, creating multiple short hops with many independent operators rather than a small number of long transoceanic trunks dominated by hyperscalers. University of Auckland researcher Dr. Ulrich Speidel has argued that "islands serve as natural transit and peering venues" and that a mesh architecture would make "monolithic 12,000+ km transpacific cables less crucial." That vision would take an estimated $1 billion in coordinated multilateral investment to achieve and would require a level of Pacific island institutional capacity that does not yet exist. The South Pacific Connect initiative, with its pre-positioned branching units and US-Australia joint funding of $65 million for Pacific Island connections, is a partial step in this direction — but the mesh remains aspirational rather than operational.
| Threat Vector | Probability | Impact | Applicable Segments | Mitigation Status |
|---|---|---|---|---|
| Fishing gear / accidental anchor (unintentional) | HIGH | MODERATE (outage, recoverable) | Shore approaches CA, AU; shelf zones | Cable burial to 1m depth; monitoring; repair fleet |
| State-sponsored anchor drag (gray-zone) | MODERATE | HIGH (targeted outage + attribution problem) | Shore approaches; shelf zones; potentially near French Polynesia | Ring topology for rerouting; monitoring; no enforceable international legal remedy |
| Seismic event / turbidity current | LOW–MOD | HIGH if seabed route through active zone | French Polynesia branches; seismically active shallow zones | Route selection avoids highest-risk zones; ring rerouting |
| Landing station intercept (SIGINT) | HIGH (US lawful; adversary: MODERATE) | Data access without outage | Carlsbad, AU landing stations; potential adversary interest at French Polynesia | NSA/FCC National Security Agreement (US); physical security at landing stations |
| Deep-ocean repeater tapping (state actor) | LOW (technically demanding) | HIGH (no outage, deniable, sustained access) | Central Pacific trunk; midpoint earth zone (~7,100km from CA) | No publicly disclosed technical countermeasure; encryption of traffic provides partial protection |
| Remote management network cyberattack | MODERATE | MODERATE–HIGH | Network-wide (landing station management systems) | FCC CALEA compliance; post-Salt Typhoon FCC review (2024 Cable NPRM) |
| Adversary landing at island node (Solomon Islands scenario) | LOW for Honomoana (no Solomons landing planned) | HIGH if materialized | N/A for current Honomoana design; relevant to future mesh expansion | US/Australia policy of denying Chinese-built systems at allied landing stations |
| Repeater power feed attack | LOW | HIGH (could disable hundreds of km of cable) | Shore power terminals at Carlsbad and Australian landings | Physical security at cable stations; redundant power feed from both ends |
VII. The Encryption Question and the SMART Cable Opportunity
A frequently misunderstood aspect of cable security is the role of encryption. Fiber-optic cables transmit photons, not encryption — the security of traffic flowing through a cable depends entirely on whether the applications and protocols generating that traffic apply end-to-end encryption. A successful physical tap at a landing station or a deep-ocean repeater node will capture whatever optical signals are present in the fiber; if those signals carry encrypted traffic (TLS 1.3, IPSec, or application-layer encryption), the captured data is of limited intelligence value for content analysis, though metadata — source, destination, volume, timing — remains accessible at the physical layer regardless of encryption. Google encrypts traffic traversing its own network infrastructure as a matter of policy, which provides meaningful protection against content-level exploitation; it does not protect against traffic analysis.
A separate technological dimension with both scientific and security implications is the emerging SMART cable concept — Science Monitoring And Reliable Telecommunications — which integrates temperature, pressure, and seismic sensors into repeater housings. As noted by Scripps Institution geophysicist Mark Zumberge in the context of the Carlsbad landing, submarine cables are increasingly recognized as distributed sensor platforms capable of detecting earthquakes, ocean temperature changes, and acoustic signatures including marine mammal activity. The SMART cable concept, actively promoted by the UN Joint Task Force on SMART cables and championed by University of Hawaii researcher Bruce Howe, would convert every repeater node into a seismic and oceanographic monitoring station. From a security standpoint, SMART sensors that detect seismic anomalies can also detect the acoustic signatures of unusual underwater vehicle activity in the vicinity of a cable — providing an early-warning capability against physical tapping operations that has no equivalent in current passive cable monitoring systems.
— Mark Zumberge, Scripps Institution of Oceanography
VIII. Assessment: Is the Current Architecture Adequate?
The South Pacific Connect architecture as designed — Honomoana, Tabua, the Fiji–French Polynesia interlink, and their integration with the Australia Connect program — represents a substantial improvement over the pre-2023 US-Australia cable environment, which consisted of the aging Southern Cross/NEXT and Hawaiki systems with limited redundancy and a single dominant US landing cluster in Los Angeles and the Pacific Northwest. The addition of a San Diego County landing at Carlsbad, physically diverse from existing Southern California systems, is a meaningful geographic diversification of the US shore approach.
The ring topology that Google has engineered into the South Pacific Connect initiative directly addresses the leading cause of significant outages: segment-level physical failures that previously had no alternate path. The pre-positioned branching units for future Pacific Island connectivity add resilience for the regional network even where the branching units are not yet activated. The dual Fiji landings on Tabua represent best-practice resilience design for island intermediate nodes.
What the current architecture does not fully address is the repair capacity deficit. With fewer than 100 repair vessels globally, most aging, and with the China-linked SBSS company dominating the regional maintenance market while being assessed as willing to delay operations in strategically sensitive areas, the time-to-repair following a major trunk fault remains a strategic vulnerability. The US and Japan have been pushing to diversify repair fleet ownership and strengthen domestic cable-laying capabilities — but this is a multi-year industrial base investment, not a near-term solution.
The cyber threat posture is adequate for the current threat environment against commercial-grade adversaries, but the NSA/GCHQ-level SIGINT capability demonstrated in the Snowden disclosures — and the Russian deep-ocean tapping capability assessed to exist — are not addressed by ring topology, redundant landings, or FCC national security conditions. The ultimate protection against content-level exploitation is end-to-end encryption by the applications and users generating the traffic, combined with robust traffic analysis countermeasures at the network layer. Google's own infrastructure encryption practice is a meaningful partial control; it does not protect third-party traffic transiting the cable under IRU or capacity lease arrangements.
The Solomon Islands remain an unresolved variable. They are not in the current Honomoana or Tabua routing, and the Coral Sea Cable System (CS2) connecting them to Sydney is Australian government-funded and managed — ensuring that the most sensitive link in their connectivity chain remains under allied control. Any future expansion of the Pacific mesh to include the Solomons as an active transit node would require either a resolution of the current China security treaty or a technically isolated connectivity arrangement that prevents Chinese access to the landing station infrastructure. Neither condition is currently satisfied.
Sidebar: Cable Laying Contracting
Google in bed with Chinese contractors/partners
The most significant case is not a Chinese contractor per se, but a Chinese co-owner — the Pacific Light Cable Network (PLCN), announced 2016. The PLCN cable system was jointly built and owned by Google, Meta, and PLDC (Pacific Light Data Communication), a Hong Kong company. PLDC owned four of the six fiber pairs in the system. Submarine Networks PLDC was acquired by Dr. Peng Telecom & Media Group in late 2017 Submarine Networks — a Beijing-based broadband provider with close Huawei ties — which triggered US alarm bells. Team Telecom urged the FCC to block the Hong Kong connection, citing the risk that the cable's Hong Kong landing station could "expose US communications traffic to collection by the PRC." Quartz
The resolution was telling: Google and Meta withdrew the original application and refiled, reconfiguring PLCN to connect the US only to Taiwan and the Philippines, excluding Hong Kong entirely. The two NSAs prohibit the applicants from allowing PLDC access to the cable and from using the disconnected Hong Kong segment. Submarine Networks
The contractor intelligence problem — survey data
Access to geographical data is the more subtle and underappreciated risk, and it goes well beyond just the laying operation itself. A submarine cable project generates multiple categories of intelligence-sensitive data:
The pre-lay route survey is particularly consequential. Prior to installation, the survey contractor provides the installer with integrated geophysical and geotechnical data — including bathymetric charts, seabed feature charts, and geological charts — to finalize the installation plan and procedures. Hydro International This data covers a 500m–1,000m corridor along the full cable route at multibeam sonar resolution — essentially a detailed classified-quality bathymetric strip chart across the entire Pacific floor along that specific track. For a 14,215-km cable like Honomoana, that's a comprehensive seabed intelligence product of obvious military value — submarine transit corridors, seamount positions, sediment type, and existing cable locations are all captured.
According to the DHS/ODNI Analytic Exchange Program, cables that lie on the seabed are "somewhat protected because their exact location is not publicly disclosed." Congress.gov A contractor who laid the cable knows the exact location — every waypoint, every burial depth, every course alteration. That Route Position List (RPL) is among the most sensitive technical documents the project produces.
The industry structure problem
The global cable market is dominated by four manufacturers and installers: SubCom (US), Alcatel Submarine Networks (France), NEC (Japan), and HMN Technologies (China, formerly Huawei Marine). In 2021, the three Western firms collectively held 87% market share, with HMN holding about 11%. Center for Strategic and International Studies But HMN's actual footprint on already-installed cable is larger than that current share suggests — between Huawei Marine and HMN Tech, the two entities have participated in at least 40 international projects, helping lay 94,000 km of cable Nikkei Asia, much of it before the US crackdown intensified.
For sensitive US projects, Washington now works only with SubCom, according to five industry sources. SubCom now works almost exclusively for the US military and large US tech firms. Marine Technology News The practical implication: HMN's cost advantage — reportedly 20–30% cheaper than allied competitors University of Washington — made it attractive to commercial operators globally for years, meaning a large portion of the existing Pacific cable network was built by what is now a Chinese state-linked entity with access to the corresponding survey data.
The South China Sea access-denial dimension
There's a reciprocal dimension worth flagging. The last submarine cable laid in the South China Sea by a non-Chinese vendor was the Asia Direct Cable, which applied for a permit in 2019. Since then, no publicly available information indicates that any project accepting non-Chinese EPC contractors has been granted approval to lay cables in the South China Sea. Springer China has effectively imposed a Chinese-contractor-only requirement in its claimed waters — which means Western cable operators transiting the South China Sea must use HMN or accept the route denial. Beijing is acquiring the survey data for all cables through its maritime territory while denying Western contractors equivalent access to generate survey data in Chinese-claimed waters. This is an asymmetric intelligence arrangement.
Where Honomoana stands
Honomoana routes through open Pacific — not the South China Sea — and SubCom is the contractor for sensitive US-connected projects Marine Technology News, so the current cable should use a Western EPC. The Honomoana FCC filing identifies Starfish Infrastructure (Google) as owner and the national security agreement conditions require compliance with DHS/DOJ/DOD requirements. However, the Carlsbad uncrewed survey vessel deployed in early February 2026 was described in the San Diego Union-Tribune reporting as part of a marine survey operation — worth noting that survey vessel ownership and contracting chain on the Australian end of the project has not been publicly detailed to the same degree as the US-side permits.
The residual problem
The deepest unresolved issue is the legacy. HMN/Huawei Marine built or upgraded cables that are still in service and carrying traffic. Those operators have the survey data, the RPLs, the repeater placement coordinates, and in some cases participated in the cable management network design. China held up at least one cable project for several months in its territorial waters, with a former US submarine officer noting that "China is attempting to exert more control over undersea activities in its region, in part to prevent US surveillance systems from being installed as part of undersea cable deployment." Data Center Dynamics The surveillance concern runs both ways — and the years when cost optimization drove Google, Meta, and others toward Chinese-adjacent partners left a bathymetric and infrastructure intelligence legacy that forward-looking contractor restrictions cannot retroactively address.
Sidebar: Local Security in Carlsbad
The physical sabotage narrative gets all the press — Baltic anchor drags, Red Sea cuts — but for a US-terminus cable in the current intelligence environment, the more consequential threat is probably passive exploitation at the shore end, not destruction of it. The Carlsbad configuration is particularly interesting from that angle.
The geometry of the intercept opportunity
The directional bore brings the cable from ~3,000 feet offshore into a 100 sq ft vault on State Parks land, then 4.3 miles inland to the Cosmos Court SLTE building. That vault-to-building conduit segment is where the cable transitions from a sealed, pressurized, high-voltage ocean system into an accessible terrestrial fiber run. At that transition point the cable is no longer armored ocean cable — it's a fiber conduit in a standard telecommunications trench. The optical splitter technique used in the GCHQ TEMPORA program (inserting a device that extracts a small percentage of the optical signal from each fiber pair) is much more practical to execute on a terrestrial conduit than on an armored repeater housing at 4,000-meter depth. The physics are identical — a fiber bend or a fused coupler tapping a few percent of the photon flux — but the access difficulty is orders of magnitude lower.
What the Snowden architecture tells us about the Carlsbad design
The NSA/GCHQ intercept model documented in the Snowden disclosures operated at landing stations with the cooperation of the cable operator or its host facility owner. At Carlsbad, there are actually three entities with physical access to different segments of the cable path: the State Parks Department (vault site), Vero Networks (conduit operator along Palomar Airport Road), and Elkhorn Enterprises/Google (Cosmos Court building). The national security agreement between Starfish/Google and DHS/DOJ/DOD will almost certainly include a lawful intercept architecture requirement — meaning the technical capability for NSA access is built into the design at Cosmos Court, per established CALEA and FISA Section 702 practice. That's the authorized side of the tapping equation.
The unauthorized side is the more interesting security question. The conduit along Palomar Airport Road runs through multiple utility corridor junctions where a sophisticated actor could access it without touching the vault or the Cosmos Court building at all. Unlike those two endpoints — which will have physical security controls imposed by the NSA agreement — the intermediate conduit is protected at whatever standard Vero Networks applies to a commercial fiber installation in a public right-of-way. That is typically a locked vault every few hundred meters and a buried conduit with no real-time tamper detection between vaults.
The traffic analysis value even without content decryption
Content decryption is not actually necessary for the most valuable intelligence product at this point in the network. At the Carlsbad landing, before traffic is disaggregated across Google's internal network, the cable carries the full aggregate load of all traffic on the fiber pairs — source and destination metadata, traffic volume patterns, protocol signatures, timing correlations. Even against encrypted traffic, a passive optical tap at this point yields:
- Which data centers at each end are actively communicating and at what volume
- Traffic burst patterns that reveal operational tempo for Google's Australian facilities
- Protocol signatures distinguishing bulk data transfer from interactive communications
- Potential correlation with signals collected at the Australian end to perform traffic analysis across the full path
For a state actor that has already placed intercept capability at the Maroubra or Torquay landing stations on the Australian end — or that can access traffic traversing Australian networks under Five Eyes arrangements — the Carlsbad tap completes a bilateral intercept picture. Espionage and sabotage operations may also be used in concert — the classic example being Britain's WWI operation that severed Germany's cables and then tapped the one remaining route to collect the Zimmermann telegram.
The specific vulnerability of the Cosmos Court facility
The SLTE building at Cosmos Court is where the most sensitive intercept opportunity exists, because that is where the optical signals are terminated and converted to electrical — the point at which traffic is most accessible without specialized optical equipment. Landing stations house network management equipment and power feeds, making them more accessible to threat actors. The use of remote network management systems creates another vulnerability that state-sponsored adversaries, ransomware groups, and other threat actors are likely to exploit.
A cyber intrusion into the network management system at Cosmos Court — rather than a physical tap on the conduit — could achieve several objectives simultaneously: passive monitoring of cable health telemetry (which reveals traffic loading patterns), potential manipulation of routing decisions, and access to the supervisory channel that runs through the cable's optical overhead between the US and Australian shore ends. Nokia's introduction of submarine cable terminal equipment had failed to clearly show the systems were not vulnerable to the attacks used in the Stuxnet operation against Iran — a reminder that the SLTE itself is an embedded system that may carry legacy vulnerabilities no different in character from the industrial control systems that Stuxnet targeted.
The irony of the NSA compliance architecture
The national security agreement conditions on Honomoana require Google to build in lawful intercept capability for US intelligence agencies. This is standard practice and entirely expected. The irony is that the technical implementation of that capability — the optical splitter or equivalent device installed at Cosmos Court to give NSA access — is itself an attack surface. If a foreign adversary can access the intercept facility rather than the cable itself, they obtain both the collected traffic and potentially the collection architecture. The Room 641A facility that AT&T operated in San Francisco for NSA upstream collection is the canonical public example: a fiber-splitting room that, if physically accessed or cyber-compromised, yields both the traffic and knowledge of what is being collected and how.
What adequate mitigation would look like
The NSA-conditioned licensing requirements will address the Cosmos Court facility directly — SCIF-equivalent physical security, access logging, cyber hardening of management systems. The gap that is harder to close is the intermediate conduit. Best practice for high-value cable backhaul in this threat environment would include distributed fiber sensing along the conduit (optical time-domain reflectometry monitoring that detects bending, tapping, or physical intrusion at any point), encrypted optical transport between the vault and the SLTE building so that any tap on the conduit captures ciphertext rather than plaintext optical signals, and regular physical inspection of all conduit access points. Whether Vero Networks — a commercial fiber operator — is required to implement those standards as a condition of its right-of-way permit is not clear from the public record. That gap in the chain of custody between ocean and SLTE building is the most plausible exploitation vector for a sophisticated actor who wants access without triggering the security controls at either endpoint.
Verified Sources and Formal Citations
- [1] Insikt Group / Recorded Future. "Submarine Cable Security at Risk Amid Geopolitical Tensions & Limited Repair Capabilities." 2025. https://www.recordedfuture.com/research/submarine-cables-face-increasing-threats
- [2] Bulletin of the Atomic Scientists. "To keep the world's data flowing, countries need to quickly fix broken undersea cables." July 30, 2025. https://thebulletin.org/2025/07/to-keep-the-worlds-data-flowing-countries-need-to-quickly-fix-broken-undersea-cables/
- [3] CSIS. "Safeguarding Subsea Cables: Protecting Cyber Infrastructure amid Great Power Competition." July 16, 2025. https://www.csis.org/analysis/safeguarding-subsea-cables-protecting-cyber-infrastructure-amid-great-power-competition
- [4] Congressional Research Service. "Protection of Undersea Telecommunication Cables: Issues for Congress." R47648. https://www.congress.gov/crs-product/R47648
- [5] Jackson School of International Studies, University of Washington. "Undersea Alliances: Japan, the U.S., and the Geopolitics of Submarine Cable Security." October 2025. https://jsis.washington.edu/news/undersea-alliances-japan-the-u-s-and-the-geopolitics-of-submarine-cable-security/
- [6] Melbourne Asia Review. "Securing the backbone: Security challenges to and governance of submarine cables in the Indo-Pacific." June 24, 2024. https://www.melbourneasiareview.edu.au/securing-the-backbone-security-challenges-to-and-governance-of-submarine-cables-in-the-indo-pacific/
- [7] LightReading. "2024 in review: Submarine cables become a battleground." December 30, 2024. https://www.lightreading.com/cable-technology/2024-in-review-submarine-cables-become-a-battleground
- [8] SIPRI. "A legislative route to combat sabotage of undersea cables: A Q&A with Pierre Thévenin." 2025. https://www.sipri.org/commentary/topical-backgrounder/2025/legislative-route-combat-sabotage-undersea-cables
- [9] Air Power Asia. "Protecting Under Sea Cables – New Strategic Challenge and Imperative." September 21, 2025. https://airpowerasia.com/2025/09/21/protecting-under-sea-cables-new-strategic-challenge-and-imperative/
- [10] Eurasia Review. "Digital Lifelines: Undersea Cables, Chokepoints, And The Evolving Sea Lines Of Communication." December 26, 2025. https://www.eurasiareview.com/26122025-digital-lifelines-undersea-cables-chokepoints-and-the-evolving-sea-lines-of-communication-analysis/
- [11] SubmarineNetworks.com. "Tabua — system specifications." https://www.submarinenetworks.com/en/systems/trans-pacific/tabua
- [12] Breaking Defense. "US, Aussies funding Google undersea internet cables for Pacific Islands, with geostrategic implications." https://breakingdefense.com/2023/10/us-aussies-funding-google-undersea-internet-cables-for-pacific-islands-with-geostrategic-implications/
- [13] Google Cloud Blog. "Honomoana and Tabua subsea cables connect South Pacific." October 25, 2023. https://cloud.google.com/blog/products/infrastructure/honomoana-and-tabua-subsea-cables-connect-south-pacific
- [14] APNIC Blog. "Rethinking submarine fibre cable projects in the South Pacific." April 20, 2022. https://blog.apnic.net/2022/04/20/rethinking-submarine-fibre-cable-projects-south-pacific/
- [15] Global Infrastructure Hub. "Coral Sea Cable System (CS2) Case Study." 2020. https://www.gihub.org/connectivity-across-borders/case-studies/coral-sea-cable-system/
- [16] Pacific Island Times. "The silent battle beneath the waves." November 5, 2023. https://www.pacificislandtimes.com/post/the-silent-battle-beneath-the-waves
- [17] US Trade and Development Agency. "USTDA Advances Secure Internet Connectivity in the Pacific Islands." https://www.ustda.gov/ustda-advances-secure-internet-connectivity-in-the-pacific-islands/
- [18] Honolulu Civil Beat. "Details Emerge On New Billion-Dollar Undersea Cable Project." April 2024. https://www.civilbeat.org/2024/04/details-emerge-on-new-billion-dollar-undersea-cable-project/
- [19] Princeton Journal of Public and International Affairs. "Leveraging Submarine Cables for Political Gain: U.S. Responses to Chinese Strategy." https://jpia.princeton.edu/news/leveraging-submarine-cables-political-gain-us-responses-chinese-strategy
- [20] Electrospaces.net. "INCENSER, or how NSA and GCHQ are tapping internet cables." 2014. https://www.electrospaces.net/2014/11/incenser-or-how-nsa-and-gchq-are.html
- [21] VICE/ProPublica. "What We Still Don't Know About the NSA's Internet Surveillance." https://www.propublica.org/article/what-we-still-dont-know-about-the-nsa-secret-internet-tapping
- [22] Boundless Discovery Newsletter. "Subsea Secrets: Spies, Sabotage, and the Global Race for Internet Cables." March 28, 2025. https://newsletter.boundlessdiscovery.com/p/subsea-secrets-spies-sabotage-and-the-global-race-for-internet-cables
- [23] Guice Offshore / FCC. "Submarine Cable Buildout Modernization — 2024 Cable NPRM." November 7, 2025. https://www.guiceoffshore.com/submarine-cable-buildout-modernization-fcc-welcomes-comments-on-proposed-rules-by-november-26-2025/
- [24] MapYourTech. "Undersea Repeater: Everything About It!" November 2025. https://mapyourtech.com/undersea-repeatereverything-about-it/
- [25] Ciena. "To repeat, or not repeat, that is the question." https://www.ciena.com/insights/articles/to-repeat-or-not-repeat-that-is-the-question.html
- [26] MDPI Journal of Marine Science and Engineering. "A Cost-Effective Approach to the Risk Reduction of Cable Fault Triggered by Laying Repeaters of Fiber-Optic Submarine Cable Systems in Deep-Sea." 2021. https://www.mdpi.com/2077-1312/9/9/939
- [27] KIS-ORCA. "Submarine Cable Design." https://kis-orca.org/subsea-cables/design/
- [28] LightReading. "Google to build two more subsea cables in Pacific Islands." January 2024. https://www.lightreading.com/cable-technology/google-to-build-two-more-subsea-cables-in-pacific-islands
- [29] San Diego Union-Tribune. "Carlsbad hub proposed for undersea data cable between U.S. and Australia." March 9, 2026. https://www.sandiegouniontribune.com/2026/03/09/carlsbad-hub-proposed-for-undersea-data-cable-between-u-s-and-australia/
No comments:
Post a Comment