Skip to Main Navigation
Download the report

Connecting the world

Data infrastructure

The digital character of modern data makes infrastructure a prerequisite for collecting, exchanging, storing, processing, and distributing data. Harnessing the full economic and social value of modern data services calls for digital infrastructure that is universally accessible, while also offering adequate internet speed at affordable cost. Yet the developing world is lagging, with major gaps evident between rich and poor people on broadband connectivity, and a substantial divide emerging between rich and poor countries when it comes to the availability of data infrastructure. Well-designed infrastructure policies are needed to redress these adverse trends.

Data infrastructure forms a supply chain that originates in global data storage centers and data processing facilities (known as cloud computing platforms). From here, data passes through Internet Exchange Points (IXPs) for transfer between different users. Then, data flows in and out of countries through an intercontinental network of submarine cables. Once on domestic soil, data are distributed through national fiber optic and microwave backbone networks, until they reach a proximate location for distribution to local communities. Finally, data are exchanged with individuals, businesses, and public institutions through fixed lines or mobile signals, and increasingly with inanimate machines, cameras, and sensors connected to the internet in the Internet of Things (IoT).

Both poor people and poor countries face fundamental inequities in their ability to access data infrastructure. To participate in the data-driven economy, people require internet connectivity. Large swathes of the population remain excluded from the internet: particularly the poor, the uneducated, the elderly, those living in rural areas, and—in some parts of the world—women. This complex situation reflects both supply-side challenges entailed by rolling out coverage of the latest mobile technologies, as well as demand-side barriers preventing potential users from taking up the service even when it becomes available. Moreover, given the growing volumes of data underpinning economic and social activity, connectivity is only meaningful if it can be provided at affordable cost and adequate speed.



The shortfall of digital connectivity in the developing world can be understood in terms of three different types of gaps. The coverage gap refers to the fact that last-mile digital infrastructure has yet to reach all inhabited locales. The usage gap refers to the fact that, even when coverage becomes available, uptake of the service by the affected population will typically not be universal. The consumption gap refers to the fact that, even when people do take up the service, data consumption typically remains relatively low.

Low incomecountries}Coverage gap% of people with coverage66%34%gapHigh incomecountriesCoverage gap% of people with coverage99%1%gap

Source: World Bank. 2019. Innovative Business Models for Expanding Fiber-Optic Networks and Closing the Access Gaps. Washington, DC: Digital Development Partnership, World Bank.

Closing the coverage gap

In 2018 more than 600 million people lived without access to the internet, a far cry from the United Nations’ Sustainable Development Goal target of universal and affordable access to the internet by 2020. Most of those unconnected live in lower-income nations. Estimates suggest that achieving universal broadband internet access by 2030 will require an investment of approximately US$100 billion in Africa alone. Closing the gap entails reducing retail costs, introducing backbone competition, enabling sharing of infrastructure, improving availability and affordability of spectrum, exploiting new technologies and reforming universal service funds.

Tackling the usage gap

Of the 3.8 billion people not using the internet in 2018, 3.1 billion lived within range of a mobile broadband signal. Government efforts to provide universal service access have traditionally focused on eliminating the coverage gap through rolling out coverage on the supply side, but such policies should increasingly be reoriented toward addressing the demand-side barriers that limit service uptake, thereby creating such a sizable usage gap. This entails targeting the most critical underserved segments, addressing the widespread problem of digital literacy and making digital devices and data services more affordable.

Narrowing the consumption gap

Even among people who connect to the internet and subscribe to data services, a wide consumption gap remains in mobile data usage across country income levels and regions. The consumption gap is even wider if fixed broadband is considered. The number of fixed broadband subscriptions is much higher in high-income economies, and because of more favorable data plans, these subscriptions support much higher levels of consumption than mobile subscriptions—potentially as much as 100 times more. Closing the consumption gap entails tackling affordability constraints and addressing technical constraints such as slow internet speeds.

Note: the indicators are defined as follows: coverage gap is the percentage of the population that live beyond the range of a broadband signal; usage gap is the percentage of the population that do not make use of the internet despite living within range of a broadband signal; consumption gap is the shortfall between the average data consumption and the proposed norm of 5GB per month in percentage terms.


Many internet users in low- and middle-income nations limit their mobile data usage because of affordability constraints. A survey carried out in 11 emerging countries found that a median of 48 percent of respondents had difficulty paying for their mobile data usage, and 42 percent frequently or occasionally restrict the amount of data they use. Instead of purchasing large amounts of data on a monthly basis, users buy it in small amounts when they have the money.

Low incomecountries}Purchasing a smartphoneCost as % of average monthly incomeMonthly average income80%High incomecountriesPurchasing a smartphoneCost as % of average monthly income ** upper-middle-incomeMonthly average income13%

Handset affordability

Poor people wishing to avail themselves of internet access must first be able to afford a mobile device.. The Alliance for Affordable Internet study found that the cost of a low-end US$42 smartphone is more than 80 percent of the monthly income in low-income countries. Taxes, import duties, and other fees also affect device affordability. Despite the low purchasing power of their populations, low-income countries on average impose the highest customs duties on mobile phones, adding 7 percent to prices on average.

Mobile data plan affordability

What is an affordable level of expenditure on data services has been the subject of some debate. The Alliance for Affordable Internet established a normative affordability threshold of 2 percent of monthly income linked to a normative consumption threshold of 1 gigabyte per month. This threshold was subsequently adopted by the UN Broadband Commission. Research conducted for this report suggests the normative consumption threshold could be 660 megabytes per month for basic services, but almost ten times as high if social media and video streaming are included.




Infrastructure is a prerequisite for collecting, exchanging, storing, processing, and distributing modern data because of its digital character. Yet the developing world is lagging behind, with a substantial divide emerging between rich and poor countries in the availability of data infrastructure. Well-designed infrastructure policies are needed to redress these adverse trends.


Creating internet exchange points (IXPs)

By keeping domestic data traffic within the country, IXPs can reduce reliance on international bandwidth, lowering costs and raising the speed of data exchanges. One study covering Latin America noted that "local bits" are cheaper than "exported bits," finding that the region spent around US$2 billion a year for international bandwidth—a sum that could be reduced by one-third through greater use of IXPs. IXPs reduce the time it takes to retrieve data, enhancing user engagement. In Rwanda, it is 40 times faster to access a locally hosted website (<5 milliseconds) than one hosted in the United States or Europe (>200 milliseconds).

Developing colocation data centers

Data centers have emerged as a vital component of the digital infrastructure ecosystem. In a data center, networked computers provide remote storage, processing, and distribution of data. The centers are mainly operated by global information technology (IT) companies, governments, and enterprises that host other companies’ data (colocation data centers). Data centers range from small rooms in organizations where data are kept on storage devices connected to computer servers to giant warehouse-like facilities where thousands of servers are arranged on racks. Colocation data centers offer companies multiple advantages, including the physical space to store a growing amount of data, the security associated with high industry reliability standards (as they almost never shut down), and easy internet access due to their growing association with IXPs.

Creation of cloud “on-ramps”

Processing of data is being handled increasingly by cloud platforms, which provide cutting-edge computing capabilities, combined with significant improvements in the capacity and speed of processing data. Cloud on-ramps are private connections between data centers and cloud providers managed directly by the latter. They allow clients to interact directly with overseas cloud providers through domestic IXPs located in colocation data centers without needing to use the internet to access cloud services, offering important gains in security, reliability, speed and cost.

Source: Comini, Niccolo, Sharada Srinivasan, and Michael Minges. Forthcoming. “The Role of a National Data Infrastructure: Internet Exchange Points, Content Delivery Networks, and Data Centers.” Background paper, World Development Report 2021: Data for Better Lives, World Bank, Washington, DC


Internet speed is a key metric to understand the quality of service provided. While in developed economies fixed download speed is usually higher than mobile, the scenario in emerging countries is the opposite. This is due to the older technologies supporting the fixed network in many low- and middle-income countries. Even so, mobile download speed in lower-income countries remains far behind that in higher-income countries.

Download speed (Mbps)00303060609090120120150150180180210210Low incomecountriesMadagascarTajikistanNepalTogoLiberiaMaliSomaliaUgandaTanzaniaRwandaEthiopiaHaitiBeninSierra LeoneGuineaBurundiGambia, TheAfghanistanMozambiqueSyrian Arab RepublicBurkina FasoMalawiCongo, Dem. Rep.Yemen, Rep.High incomecountriesSingaporeHong Kong SAR, ChinaMonacoSwitzerlandFranceSwedenUnited StatesDenmarkLiechtensteinHungarySpainMacao SAR, ChinaCanadaKorea, Rep.NorwayLuxembourgNetherlandsNew ZealandPortugalAndorraLithuaniaLatviaJapanMaltaUnited Arab EmiratesPolandChileIsraelBelgiumFinlandSan MarinoPanamaBarbadosGermanyQatarIrelandSlovak RepublicKuwaitSloveniaUnited KingdomEstoniaItalySaudi ArabiaTrinidad and TobagoCzech RepublicAustriaAustraliaUruguayCroatiaOmanBahamas, TheCyprusBahrainGreeceBrunei DarussalamSt. Kitts and NevisSeychellesAntigua and Barbuda

Speed of mobile data plan

Of all 24 low income countries considered, only Madagascar (4%) has an average mobile download speed of 30 Mbps or more. In contrast, most high-income countries - that is 88% of the 58 countries considered - have an average fixed download speed of 30 Mbps of more.

Source: Speedtest. 2021. “Speedtest Global Index”, Speedtest.


The amount of data transferred on the network is increasing steadily and the current pandemic increased the already growing trend. Depending on where you are located, transit prices for data may differ. As a result of connecting almost 90% of internet capacity, cities in the European Union and United States are the cheapest hubs for data transit.

Cost of exchanging 10Gb of data$0.5$1$2$4$8

Source: TeleGeography

Price of data movements

Significant price differences persist for the exchange of data traffic. The cost is relatively modest in high income countries where traffic is highest. For example, the cost of exchanging data is around US$0.45 per megabyte per second in North America and US$0.62 per megabyte per second in London.

IP transit prices in developing countries

Transit prices in the Global South are much higher compared to the EU and USA. They are the result of an unbalanced data flow due to less developed digital ecosystems (e.g. little domestic data infrastructure and a limited amount of competition). For example, the cost of exchanging data is around US$2.38 in Sao Paulo and US$5.00 in Johannesburg.