Over the past three years, Google has deployed two metro fiber networks in Africa, under an umbrella of initiatives dubbed "Project Link." The first network was rolled out in Kampala, Uganda in 2013, with the second going live earlier this year in Accra, Ghana's capital-city. As is often the case with projects initiated in Africa by one of the major web-scale players, the Google Project Link initiatives have attracted somewhat hyperbolic headlines; Google has been lauded for "bringing real broadband to Africa" or offering the "solution to [Africa's] last mile Internet connectivity."
That is a little excessive -- but research for our "Future of African Bandwidth" series suggests that the Google (Nasdaq: GOOG) metro projects do stand to have an enduring impact on the structure of African metro fiber economic models.
Domestic fiber as the last hurdle to African connectivity
The availability of plentiful metro and interurban fiber priced on a competitive, open-access basis is arguably the last most significant bottleneck to increased connectivity in African markets. Long the main obstacle to Internet adoption, international capacity is now pervasive. Around 3 Tbit/s of international capacity has been added to African markets since 2009 -- so much that supply is now substantially higher than current-state demand and less than 10% of international cable design capacity is effectively being used. Likewise, last-mile access is improving, primarily thanks to the accelerated rollout of 3G and 4G mobile networks across the continent, along with pockets of FTTH expansion in markets such as Kenya or South Africa.
Then there's the "middle-mile" link, which, but for a few exceptions, has remained doggedly problematic. In many countries, metro and inter-urban capacity markets are competitive on paper. In practice, they are something of an African telecom anomaly, monuments to market failure in a sector that has otherwise contributed billions of dollar in economic development thanks to market competition.
The paradox of African metro markets is that there actually is a lot of fiber in the ground, connecting the continent's main economic urban centers. But, depending on the country, the metro segment is also characterized by ineffectual monopolies; dozens of companies duplicating fiber trenches on a small number of key routes; and fiber wholesale players keeping prices high to meet short investment payback periods or protect their retail units by making life more difficult for Internet retail competitors. All this has constrained capacity demand, complicated the fiber business case, and, at a broad level, created a peculiar mix of woefully inefficient use of capital and fiber capacity usage that remains desperately below potential. In many countries, frustrated governments are taking matters in their own hands to fill the middle-mile gap, building national fiber backbones (with Chinese vendor support primarily) with mixed results.
Google, disrupting again
The Google Project Link rollouts are now disrupting this longstanding (and largely unproductive) equilibrium. In Uganda, Google blanketed the capital city of Kampala with around 800 kilometers of fiber. The capacity is sold on a wholesale basis only to ISPs and institutions that need (and can afford) capacity in relatively high volumes. Google complemented the fiber network with a Wi-Fi network of around 120 hotspots, also made available on a wholesale basis only to ISPs. In Ghana, Google is already selling metro capacity as part of a project that should see the rollout of around 1200 kilometers of metro fiber in the country's economic heart of Accra/Tema, and Ghana's second largest city, Kumasi.
The impact of Project Link in these two countries has been noteworthy. ISPs and other capacity purchasers interviewed over the course of our research into African fiber markets commonly used expressions like "game changer" or "disruptive" to describe the Google Fiber approach.
Project Link seems different in a number of critical areas. For one thing, it is upending traditional pricing models. African wholesale capacity has long been sold on a capacity/distance basis. Users pay per link, per unit of capacity (E1 or STM-1) per kilometer, a multi-variable equation that leads to byzantine pricing books that almost seem intended to discourage capacity purchases. While there are volume discounts, the entry price point is so high that few can get to those volumes in the first place.
By contrast, Google charges per bundle of sites connected (e.g., up to 25, 25-50, 50+), regardless of the distance within the coverage area, and regardless of the amount of capacity within capacity brackets. In effect, the Google African pricing model puts the emphasis squarely on volume usage -- the more one uses, the less one pays, with no other factor (e.g., distance) coming into play. Indeed, there actually is a penalty of sorts for low-capacity usage; below 10 Mbit/s, the price differentials versus those of other providers (within the metro area) are fairly low.
Above 10 Mbit/s of capacity, the savings are a factor of 1x to 2x or more depending on the number of sites connected. It's important to note that this type of flexible open access pricing isn't a Google invention in Africa -- it was pioneered in Africa by the erstwhile Kenya Data Network in Kenya. The South African metro market is also largely built on similar open access foundations. The rest of Africa, however, has largely been impervious to it.
Project Link also promises much better quality of service than rival offerings. While SLAs do exist, they are often meaningless. As one capacity customer put it: "The SLA is nice and I get a discount on purchases, but it's entirely useless to me if my users can't connect because of the persistent downtime." Along the same lines, responsiveness has been a common refrain (some smaller customers have complained of being treated dismissively by fibercos, presumably because their capacity purchases are too small).
Finally, there is the availability of dark fiber through Project Link. African telcos are loath to sell dark fiber, generally for fear of cannibalizing recurring revenue from their leased capacity sales. Project Link has no such compunction, making dark fiber available in return for a support and maintenance fee.
Next page: Changing the conversation