Subsea Cables & Internet Infrastructure

A shunt fault is damage to the insulation around the power conductor that all subsea cables require for undersea optical amplification. It causes an electric short. They are most common in the deep sea where the cable lies exposed on the sea floor as opposed to near shore where burial is routine. Abrasion caused by deep sea currents that rub cables against sharp surfaces like rocks are the usual suspects. Depending on the extent of the abrasion, a shunt fault may or may not cause an outage. It depends on the ability of off-shore power facilities to compensate for a partial loss of voltage. When I was worked at Hibernia Atlantic, our cables experienced shunt faults that did not disrupt service. A cable ship would be dispatched to fix it. Below is a shunt fault on Subco's Oman/Australia cable. You can see the power burns on the eroded insulation. 

An Irish company, MDM, whose founding members have done a lot of design work for hyperscalers, has announced it is building two Trans-Atlantic cables in addition to its Pisces project that will connect Ireland to Europe. The estimated construction cost is $500 million. The network is a classic dual ring design where the Northern cable lands at Galway and on Long Island. Now there are several Long Island cable landing stations and MDM already operates one in Galway. So MDM may use these existing facilities for its project. In contrast, the Southern cable is expected to land at a new CLS on Ireland's South Coast. On the US side it will land in New Jersey where it will probably use the existing Wall Township CLS. I have concerns about this project because there is nothing to indicate it is fully funded. In fact, the last carrier financed and led project across the Atlantic was Aquacomms' AEC-1 cable that went live in 2016. EXA recently purchased Aquacomms and all its Atlantic ass...

Light takes .489 milliseconds or roughly a half millisecond to go 100 kilometers. Get the route miles and you have a great approximation because optical amplifiers or end point electronics have little impact. So Dunant is 13,200 kilometers CLS/CLS. That is 132 times a half millisecond or 66 milliseconds RTD. For full details, click on https://www.m2optics.com/blog/bid/70587/calculating-optical-fiber-latency. 

Google announced today three new cables: Alisios, Canoa, and Olaluz. President Trump will not be happy with the names. Not American enough. Alisios is a Spanish word denoting the trade winds that cross the Carribean and the Equator. This cable will connect the Dominican Republic to Chile via Panamanian landings. So it will include new long haul fiber routes across Panama. Canoa links the Dominican Republic to Bermuda, where both of Google's new Trans-Atlantic cables, SOL and Nuvem, will also land. So similar to the Pacific where Google is landing six cables on Guam to create its own traffic exchange point, Bermuda is becoming another major private switching center for Google traffic. Traffic between Europe and the Caribbean or the West Coast of South America can now bypass the United States, improving resiliency, yet reducing latency. Indeed, traffic between Europe and Australia can now be routed via Chile because Google's Humboldt cable will link Chile and Austra...

Two year term.

"It's not out of the question that, at some point, Starlink will operate most of the world's internet," Musk said." Despite this optimism, shares of SpaceX fell by more than 7% in after hours trading on Tuesday, wiping out gains made during the day.

 

The 2Africa cable has been up and down since Wednesday. Repair was declared completed and all services live Friday Morning Abidjan time. WACS should be fixed by Wednesday next week.

 Oracle's stock price is diving. Why? A lesson for the AI industry. An AI data center customer faces no switching costs as they are simply installing data and software in bare metal servers leased to them by the data center. The result is a buyer's market as opposed to the traditional sellers market in which Equinix and Digital Reality operate. Oracle, the software company, reinvented itself as a cloud provider and then pivoted to AI data centers when the cloud angle proved disappointing and AI became hot.  But in a traditional data center the customer installs their equipment and this makes leaving the facility very expensive and difficult as it means they lose an operating network node. In many cases the exit costs can be in the millions of Euros or dollars and require a network redesign to ensure traffic flows are not disrupted. So the traditional data center is more like the Hotel California in the Eagles song where you can check in, but never leave.  In contrast, mos...

Nvidia guaranteeing OpenAI's 250 billion in data center commitments is a huge gamble given that OpenAI is forecasting it will burn through $665 billion in cash by 2030. Nvidia's entire future depends on OpenAI's success or failure. The lack of prudence and financial discipline is astonishing. Russian roulette, anyone? History is quite clear: never finance money losing customers. Any credit check would have disqualified OpenAI using its own financial projections. For more details, https://thetechcapital.com/nvidia-to-backstop-250-billion-of-openai-data-centre-commitments/.

Just a few months ago Linkedin Prophets were proclaiming SpaceX hegemony. It was in their words 'a space utility' that would dominate space infrastructure due to 'vertical integration', Musk Infallibility, and 'rocket launch cadence'. Now their only excuse is 'short selling' as though $18 billion in annual revenues naturally implies Space is worth two trillion dollars. This inference is built on the flimsy foundation of rocket usability and 'they launch a lot of stuff into space'. In fact, SpaceX is a space trucking company. Logistics. There is no intellectual moat, and in fact, SpaceX has no patents on rocket design. Not a single one. It has an edge on rocket design, but textbook economics tells us abnormally high profits attract competition and similar designs. Just last week a Chinese company completed their first successful reusable rocket launch.

The FCC is introducing a new regulatory framework for subsea cables. The previous approach involved a Federal interagency group known as Team Telecom reviewing each cable landing and the associated owners of the subsea network. Team Telecom includes representatives of the FCC, Department of Defense, Department of Justice, and Homeland Security. Moreover, advisory agencies include Department of State, Treasury, Commerce, Director of National Intelligence, Office of the US Trade Representative, and General Services Administration. Consequently, approval delays were common as well as regulatory overkill because each applicant must go through the same steps regardless of previous successful applications. The new regulatory regime places applicants in different risk buckets. Low risk applicants can bypass Team Telecom and quickly get a landing license approved. An applicant shows it is low risk by meeting the following requirements. 1. Applicant certifies compliance with ten F...

It turns out that the CS Sovereign, a Global Marine ship whose home port is Portland, UK, is sailing down the West African cost to Abidjan to fix WACS. Why was the Sovereign chosen when it is so far away? Every cable system has one or two storage depots where fibre is kept for repairs. WACS spare fiber and other repair supplies are kept at Brest, France. Now Orange Marine has a cable ship, the well known Léon Thévenin, at anchor in Capetown, South Africa. But if the supply depot is in Europe, a nearby cable ship becomes useless. It all makes eminent sense because WACS connects both the UK and Lisbon to West Africa and the cable was laid from the UK to Lisbon and down the West African coast. So the starting point for deployment was Europe and hence that is where the spare supplies are located. However, this is a severe logistical mistake since most WACS outages occur between Senegal and Congo, not in European waters.

Fear is gripping the SpaceX Faithful as they watch their Dreams of Dominating the Galaxy evaporate in the harsh glare of market financial scrutiny. Short term the market is a voting machine. But long term, it's a calculating machine. August 5th is the quarterly earning release. It will show more massive losses just like 1Q2026. Moreover, the investors that financed SpaceX as a private company will soon be able to sell their shares and given the downward price trajectory, they will not hesitate to sell. 😉 I think the stock's fair value is in the 30s. I recommend a buy when it gets there. 😄

The cable ship repair permit has been granted. It is expected to arrive off Abidjan on July 22nd. The 2Africa cable will be the first repaired despite its outage being limited to the Cote d'Ivoire branch. WACS repair is expected to begin at the month's end. The ASN Ile D'Ouessant cable ship left Capo Verde last week and will arrive in the Gulf of Guinea this Wednesday.

Goldman Sachs, the lead for the public offering, justified the $150 IPO price based on the assumption that SpaceX 2025 revenues of $18.67 billion would grow to $474 billion in 2030, including the AI division, which generated $3 billion in 2025, growing to $322 billion in 2030. Given that most of of the AI division's data center space has been leased to third parties in the last several months, this may be a bit generous. 🙃  The entire Goldman Sachs forecast was grounded in the premise that SpaceX will dominate the AI space. Yet its own employees don't use its trademark Grok product and the AI division's subscription growth is 24% per annum with total 2025 revenues of $3 billion. In contrast, OpenAI's revenues for the same year were $25 billion with a 233% annual growth rate. 

Keppel is a member of the Bifrost Consortium. It is essentially arbitraging the difference between per pair consortium cost and what third parties are willing to pay for them. The arbitrage business model comes back to the beginning of telecommunications competition when voice carriers engaged in various kinds of arbitrage to lure customers away from the incumbents. Note Keppel sells no lit services. You must go to Telin today for lit Bifrost services, at least until some of these fibre pair purchases turn into lit wholesale capacity.

Stocks of publicly traded companies including Coreweave, META, Microsoft, and Oracle have lost 4% to 39% of their stock market value over the last year. Coreweave's core business is building data centers stocked with GPUs to rent AI service providers for estimation and inference. Obviously, this makes the standard data center look downright capital light. Racks must be populated with servers and GPUs and capable of handling 120Kw power loads. So power infrastructure and backup alone cost many multiples more on a per square meter basis than the standard telecom hotel. Moreover, these data centers must be bigger because a large language model might have trillions of parameters to be estimated. Coreweave's massive debt loads have led to a 37% decline in its market value over the last twelve months.  Secondly, customer switching costs are very low in the AI data center market. AI service providers are software companies. They download their data into the bare metal ...

NTT and Mitsui Leasing have teamed up together with fibre optic cable manufacturer Sumitomo Corporation to fund a 16 fibre pair subsea cable costing $500 million. Design capacity is 320 Tbps. This is the second project on which NTT Data and Mitsui Leasing have cooperated. The high capacity 20 fibre pair Juno cable was the first project. It is unusual to see a large incumbent player like NTT undertaking a subsea cable project with a non-telecom company as co-owner. But the advantages are clear. Mitsui will contribute cash, but not play a major role in design, vendor selection or wholesale commercials. So it not only reduces NTT's risk by sharing funding requirements, but it gives NTT a free hand in decision making. One lesson that has become perfectly clear is that large carrier consortiums increase the likelihood of deployment delays and lengthen the entire planning cycle because decisions require consensus. Moreover, the consensus requirement makes it difficult for new approaches...

MRC: $32.5K. A point: Most Mumbai DCs. Z point: SG3.  Cable: IAX.