Google's New Umoja Cable: Linking Africa To Australia

Google has been a path breaker in the submarine cable world. It financed the first spatial division multiplexing cable across the Atlantic, namely the 12 fibre pair Dunant system. It is also the owner of the new 16 pair Topaz system which directly links Canada to Asia and bypasses the US. Not to mention the billion dollars Google is spending to crisscross the Pacific with a web of new cables connecting Japan, Guam, Hawaii, Chile, Australia, Fiji, French Polynesia, and other island chains. Umoja is no less surprising than its predecessors. It has two key components. A new terrestrial fibre highway in partnership with Liquid Technologies that will head Northwest from Kenya's data centres to Uganda and then South to traverse and connect Rwanda, the Democratic Congo Republic, Zambia, Zimbabwe, and finally, South Africa. 

The bane of sub-Saharan Africa has been the lack of domestic long haul fibre connecting the landlocked countries to each other and to the coasts where the subsea cable landing stations reside. Several obvious factors have stymied efforts to build international digital highways, namely, distrust and conflict between nations, lack of resources, the fact than any project member could sever connectivity, and finally, the well known externality effect. The last factor reflects the divergence between a project's total benefit and the benefit to any individual participant. Each potential participant in a terrestrial project linking several African states generally considers only the benefit their country receives from the project. They do not take into account the benefit that the other consortium members get. This means that many worthwhile projects might not get funded and that appears to be the case in Africa. 

Most international connectivity is via subsea cables encircling the Continent. International terrestrial connectivity either does not exist or has limited capacity and extraordinarily high pricing. Part of the problem is that many landlocked African states have telecom monopolies. So end to end international terrestrial Layer 1 circuits don't exist. Instead, each monopoly provides optical Layer 1transport to the border and service ends there. There are border facilities that provide cross connects between one monopoly's wavelengths and its partner on the other side of the border. But no end-to-end service from one country's telecom facilities to another. For example, one can get a 100G wavelength from Freetown, Sierra Leone to the Northern border with Guinea. Service ends at the border. So a customer wishing to create a digital highway from Freetown to Guinea's capitol, Accra, must purchase wavelengths from the Sierra Leone and Guinea telecom monopolies and get them to cross connect at the border. This severely complicates provisioning, network testing, troubleshooting, and other matters as two independent service providers are involved with a general tendency to blame the other party for any problems that may arise. In contrast, one can get an end-to-end service from Paris to Berlin with one provider and no border cross connects. Seamless with one party responsible and provisioning taking only a few weeks. No politics involved. 

Umoja is as much about creating a well functioning interstate digital highway for Africa as it is about subsea cable resiliency and diversified routing options. Google is way behind in the race for Cloud dominance with an estimated 11% of the global market. In contrast, Amazon leads the pack with probably has a third of the global market and Microsoft's Azur is probably in second place with market share in the low twenties. However, Google could potentially end up in first place in the emerging African Cloud market due its Equiano and Umoja cable investments combined with this new terrestrial Kenya/South Africa route. Amazon's sub-Saharan African investments have been limited to edge facilities. Microsoft has been more aggressive and built two data centers in South Africa. In contrast, Google's strategy is to use its superior low latency network to back haul African cloud traffic to Europe for the time being. This helps avoid the problems associated with data centres in Nigeria and South Africa which include power shortages and outages and very high cooling requirements. Obviously, long term Google must too build hyperscaler data centres on the Continent. 

The subsea part of Umoja is Google's answer to telco consortium dominance. It prefers cables that it controls, which means majority or sole ownership. Whereas Umoja's natural Asian destination might seem to be Singapore, the city state is a poor choice for hyperscaler facilities relative to land rich Australia. And Singapore has become a single point of failure. Combined with Equiano, Umoja offers Google an almost complete, resilient African network without any pressing need to participate in an East African cable consortium. Few details are available today concerning this subsea cable project. A good guess is that it will go live in 2026, be a spatial division multiplexing system with at least 12 fibre pairs, probably 16, and that ASN will be the supplier. I expect a quarter of the cable's capacity to be sold to telecommunication carriers. To learn about Umoja straight from the horse's mouth, click on https://cloud.google.com/blog/products/infrastructure/investing-in-connectivity-and-growth-for-africa.