Subsea Cables & Internet Infrastructure

1. The Gate to Hell in Dante's Inferno has an inscription stating "Abandon All Hope Ye Who Enter Here". I think he was actually referring to the Senegal telecommunications market. My conversations with African carriers and Tier 2 ISPs suggest that entering the market is extremely difficult. Much more than Ghana, Ivory Coast, Nigeria or South Africa. Only the Sonatel/Orange joint venture (cartel?) has managed to secure a gateway license for 2Africa capacity. This is the consequence of the pernicious and incestuous relationship between Senegal's government and the de facto Sonatel monopoly.  Developing a country is best accomplished via competition with low barriers to entry, not government sanctioned monopolies with a good dose of under-the-table brown bags stuffed full of Euros and US dollars.  The Senegal government is spending a huge amount on the Numerique data center and other digital projects, but they will not live up to their promise unless transport prices mat...

AAE1; $21.5K MRC; Two Year Term. Peace; $17.5K MRC; One Year Term. A points: MRS2. Z points: SG1/SG3. Customer responsible for cross connects.

Google's Topaz cable offers exceptional physical diversity due to its unique Vancouver, BC and Japanese landings. It is also one of the shortest paths from the Tokyo financial exchanges to Seattle, Chicago, and the CME data centre. A point: CC1, TY4. Z points: Cologix VAN4, Vancouver. Also Equinix at 350 East Cermak, Chicago, and Digital Realty, Westin Building, Seatle. Service: Standard Layer 1 10G waves. OTN. Terms: 1, 2 & 3 Years.

In the late 90s Tyco Electronics (TE) purchased AT&T's subsea cable laying division. Stock prices of new fibre optic networks were soaring and priced at multiples similar to American Tech Giants today. So TE built a global subsea network, Tyco Global Network (TGN), to sell wholesale capacity. By the time it was completed in 2003, bandwidth pricing had collapsed. It was clear that the billions spent on TGN would never be recouped. By 2005, TATA, then known as VSNL, scooped it up for $130 million in one of the great contrarian investments in the telecom industry (Hibernia Atlantic's purchase in 2001 is another example). TATA got a lot. It included two Atlantic cables structured as a ring, dual cables linking India to Marseille and to Singapore plus a number of Pacific cables. TGN Pacific was one of those distressed assets. TGN Pacific was designed like most cables of that era to be self healing. Today most customers provide their own route protection via routers or switches....

Great Marseille/Singapore 100G AAE1 Pricing: $21.5K MRC A point: MRS2. Z point: SG3. Term: 1 Year. Service: Layer 1 100G Wave. NRC: $0. Estimated Latency: 135 ms RTD. Customer responsible for cross connects. Delivery: Four weeks from customer signature.

A point: Most Tokyo Equinix sites. Also CC1 & Osaka Equinix. Z point: San Jose Equinix, LA Coresite 2. Service: 10G Wave. Layer 1. Term: 2 years. MRC: $5250. NRC: $2500. Customer responsible for cross connects. Cable: Juno. Remarks: Diverse back haul to both LA & San Jose. A point: TY5. Z point: Secaucus Equinix. Service: 10G wave. Layer 1. Term: 3 Years Latency: 163 ms RTD. MRC: $8500. Customer responsible for cross connects. Cable: No outages in the last two years. Note: Ideal for financial trading.

Singtel is leading a consortium of Asian carriers and American tech giants that will finance a new intra-asian cable connecting Japan, Singapore, Brunei, Indonesia, Malaysia, the Philippines, Taiwan, and South Korea. Details are sparse. Press releases say it will be a high fibre count network. My guess is at least 12 pairs and as much as 24. Two core fibre is also a possibility given NEC's participation. What is striking is that SJC2 and ADC just went in service with Apricot also expected to be RFS this year, yet here is another cable with a similar Southeast Asian footprint under development. This suggests to me that traffic is growing faster than anticipated with the carriers under pressure to catch up to the rapidly growing market. The other notable feature is the absence of any landings in Hong Kong or mainland China. Since China has emerged as the regional bully, the cable's name is probably not a coincidence. Consortium members include Singtel, Amazon, Micro...

100G Equiano Waves: Most Lagos DCs/European Cities: $17K MRC. A Points: Equinix, Medallion LOS1/2, Medallion LKK2, and Rack Centre. Z points: Key Marseille, Paris, Amsterdam, Frankfurt, and London DCs. Service: 100G Wavelength. Term: 3 Years. MRC: $17K. Lagos Metro Route Protection Included. Promotion Ends September 15th.

The AFR-IX cable project has selected Nokia as their primary coherent optics vendor. Medusa will deploy the Nokia 1830 platform that supports coherent optics. Nokia's Infinera acquisition provides the ICE7 coherent optics. This equipment announcement illustrates how slowly the interesting Medusa project is moving. This project was announced in 2020. While it is smart to delay terminal gear purchases towards the end of the project, subsea cable projects have not historically taken upwards of six or seven years to finish. I can only speculate on the cause of these delays. The system has many landings in dysfunctional North African countries where governments are authoritarian, corrupt, incompetent or compromised by their ownership or control of PTTs. Cable ships are scarce. So booking them for the deployment may have been a big obstacle. I am disappointed that Medusa made the 'safe choice'. Big projects with lots of cash usually end up picking either Ciena or Infinera for the...

The two fibre pair Greenland Connect system links Greenland to Iceland and Canada's Newfoundland province. Photos of the cable landing stations below. The first photo is the Icelandic facility with the one below it being the Greenland counterpart. Greenland Connects lands South of Reykjavik and uses the same landing station as the DANICE cable. In Greenland it lands in Nuuk, the country's capital, and also in Qaqortoq. Although the cable began with 2x 10G waves in operation, it has been upgraded to 12.8 Tbps by Alcatel and Hauwei Marine.  Greenland Connect went live on March 23, 2009. It brought the first terrestrial Internet connectivity to Greenland and slashed RTDs to content delivery points by over 500 milliseconds. Up to that point Greenland had relied solely on geostationary satellites. I strongly believe that the Far North, which includes Siberia, Arctic Canada, Greenland, Iceland, and the Far Northern parts of the Nordic countries, will experience massive migration and ...

1. The UK, South Africa, and Portugal POPs are passing traffic today. A. UK POPs include Slough Equinix for most 2Africa consortium members. Many are also at Telehouse London.  B. Lisbon includes both LS1 & the new Altice facility. C. South Africa: CT1, CT2. 2. Raxio is the main 2Africa POP in Abidjan, Ivory Coast. Hand offs also include the CLS itself. Originally the planned PAIX data center was intended to be the main 2African POP. However, it is either cancelled or delayed. Not clear at this moment as I have received conflicting opinions.  3. 2Africa Lagos POPs include MDXI and also new Digital Realty site. The former Medallion DC where most 2Africa SLTEs will be housed only opens its doors in late September. So Lagos may not be live til late 2025. 4. There are plans to extend 2Africa from the Pointe Noir and Muanda cable landing stations to Kinshasa OADC. It will take the form of a fibre ring. The Kinshasa back haul fibre was part of the original 2Africa network plan. ...

1. 2Africa is vastly superior to the older African cables in cost, performance, and footprint. It has 9x the capacity of WACS or ACE. Cross connects are capped at $150 per month at the CLS and most hand offs are at carrier neutral facilities. The cable is buried 2 meters in deep in shallow coastal waters (1000 meters or less in depth) versus 1 meter or less for older systems. It bypasses all known danger spots such as subsea canyons with their debris slides. 2. Sénégal A. ONIX is the main 2Africa POP in Dakar. Note that consortium members are free to place their SLTEs where they desire in Dakar. So other sites may also be on-net. The Dakar CLS just houses the power feed equipment. B. ONIX just opened its doors to customers so expect 2Africa service to begin near year's end. C Sénégal is a significant challenge for capacity buyers because at least one major consortium player elected to skip it. D. I have great 100G pricing for the LS1 to ONIX route at $17.5K a month on a 1 year term...

A. Three cables have lost their Dubai connectivity: EIG, TATA Gulf, and IMEWE. EIG is depicted in the top panel. It goes as far West as London. IMEWE lands at Marseille. Finally, TATA Gulf is a branch of TGN-EA. B. EIG uses a branching unit to split into a subsidiary trunk that goes up the Persian gulf and a main trunk to Europe. Hence the main trunk is probably ok. The same comment applies to TGN-EA and IMEWE.

 The Coherent Optics Revolution: Transcending 10G Wavelengths - Part 1 The first stage of optical communication was dominated by what Mark Tinka called the simple optical detection scheme. A pulse of laser light represented a '1' and no light meant a '0' or vice versa. Cisco white papers call this 'on-off signalling'. So this approach is based on the optical power or intensity of light. The stronger a light pulse, the higher its amplitude. See the top diagram.  The Achilles of this approach is chromatic dispersion, namely that fact different frequencies of light traverse a solid medium such as fibre glass at different speeds. Now any laser pulse is a band of frequencies. It may be narrow, but it always has non-zero width. So chromatic dispersion is inevitable (like Donald Trump continuously changing tariff rates). As the fibre path distance grows, the probable outcome is that a laser might transmit a '1 0' but the light will spread over time and the opt...

Japan suffers from a lack of physical diversity in cable landings. For example, the Maruyama CLS serves 8 cables, most of them major Internet arteries including ASE, APG, Jupiter, TPE, and ADC. According to the Submarine Networks website the country has more than 20 facilities. However, they are densely concentrated as the map below shows. Moreover, it would not be surprising if many of the are sharing back haul fibre. Consequently, the Japanese government is giving money to Softbank for the construction of two new cable landing stations in the Hokkaido and Fukuoka projects because they make nations's telecommunications more robust The project's anchor tenant is the ETA (East to America) subsea cable.  In general I am skeptical of subsidies for a variety of very good reasons, namely they usually distort the allocation of resources in pursuit  of political gain. However, aid for new cable landings that are diverse to the existing landing infrastructure may be exception. It woul...

 A point: Lisbon Equinix (LS1). Z point: MDXI, Server House or Medaillion facilities. Term: 1 Year. Cross connects not included.

***Hong Kong/Tokyo, Tokyo/Taipei, and Hong Kong/Taipei. ***Outages started 10:14 GMT, July 6th, 2025. ***Until a cable ship can investigate, no idea of the cause. ***I caution the paranoids among you to refrain from speculation. Hong Kong is part of China so the sabotage theory is most likely bogus. *** The graph shows that the common factor among these segments is Taipei.

Google's SOL cable is the first Trans-Atlantic network to connect Florida to Europe with a landing at the Telxius Santander CLS on the Northern Spanish coast. The other cable, Nuvem, announced some time ago, will link the Myrtle Beach, South Carolina CLS owned and operated by DC BLOX to a landing near Lisbon. Details are sparse regarding SOL, but it will probably be similar in performance and design to the 16 fibre pair Nuvem cable that clocks 384 Tbps. The Florida landing is in the Palm Coast area between Jacksonville and Orlando. An interesting feature is that both cables do island hopping. Both cables land in Bermuda and Azores (where a US Air Force base is located). Island hopping serves three goals. The first is power to offset voltage drop. Intermediate power feeding en route improves throughput. The more often a cable can feed, the higher the bandwidth. The other factor is optical amplifier noise. Amplification introduces noise which accumulates from amplifier to amplifier. ...

According to FCC filings Starlink shut down almost 500 Starlink satellites during the first half of 2025. The company had them reenter the atmosphere where they burned up. What is striking is that these satellites were all less than 5 years old. The general consensus is that LEOs have a life expectancy ranging from 5 to 8 years. Shorter than expected life spans for the satellites will hit Starlink's income statement hard by increasing network depreciation and replacement needs. However, Starlink has managed to lower its LEO's manufacturing costs down to $500K versus initial figures around $1 million. So these production economies of scale might offset some of the higher than expected depreciation. However, there are also rocket launch costs as well. It costs Starlink about $3 million to put a satellite into orbit. The Falcon 9 costs $67 million per flight and delivers 23 LEOs into low Earth orbit. As a private company Starlink financials are a bit of mystery. The company press ...

1. The West coast network from Lisbon and London to South Africa should be all activated by year's end. 2. Note that 2Africa is an open cable system which means each fibre pair and spectrum owner is responsible for their SLTEs. So it quite possible that consortium member X is ready today whereas member Y might be RFS only in December. 3. RFS Guidelines A. London, Ghana, Nigeria, and South Africa will be RFS at the beginning of September. B. Côte d'Ivoire should be live a month later. C. Senegal is at least 2 to 3 months from launch and could be as late Christmas. 4. Buying Guidelines A. I expect the combined impact of 2Africa and Equiano to drive Lisbon/Lagos 100G market pricing below $20K. On this route I recommend 1 year contracts. B. In Senegal, Ghana, Ivory Coast, and DRC you should do long term contracts because there is no guarantee that 2Africa will permanently lower pricing. Short term the cable will do so. But it is least 2 to 5 years before another modern cable lan...