Announcing Project Pangea: Helping Underserved Communities Expand Access to the Internet For Free

Community networks: connectivity that communities build and run

Project Pangea is Cloudflare’s effort to support community networks — locally operated infrastructure that provides Internet access to underserved areas and populations. Estimates indicate half the world still has no connection, and many more rely on service that is slow, expensive, or unreliable, despite heavy public and private investment.

An Internet Society definition frames community networks as what happens “when people come together to build and maintain the necessary infrastructure for Internet connection.” These projects usually arise out of necessity, in places where commercial or public options have stalled. Many stay local, but some mature into regional or national operations. Common operating models are not-for-profit and cooperative structures, which keep surplus inside the communities. Some networks, for example, reinvest revenue to replace Wi-Fi with fibre-to-the-home.

Community networks have a record of dependable service over years, not just pilot projects, and draw on knowledge sharing among organizers. Yet several assumptions tend to block support and policy. Four persist:

• Only remote or rural communities are underserved. Remote regions certainly lack access, but so do communities within roughly 10 km (about six miles) of major cities — and even dense urban areas, as several of the program’s launch partners demonstrate.
• Underserved means low-income. Rural and remote populations often can afford service and simply have none available. In cities, the gap may be price-based: access exists but is unaffordable for many, which raises egalitarian incentives to build alternatives.
• Service has to cost more. Claims that incumbents cannot offer affordable access so nobody can are false. Community networks use governance and cost models that let subscribers pay rates comparable to the broader market, as the launch partners do.
• Technical skill is both required and absent. Evidence shows communities can build reliable local networks from commodity hardware and non-specialist equipment with modest training and support.

One truth remains: sustainability is hard. Many networks begin on volunteer effort or grants, resources that cannot carry them indefinitely. The transition point — “willing to charge and willing to pay” — is where Project Pangea intends to help.

Why free connectivity is not enough

Project Pangea provides Cloudflare’s network services — Cloudflare Network Interconnect, Magic Transit, and Magic Firewall — at no cost to nonprofit and community-focused networks. Qualifying organizations are those whose primary aim is connecting underserved or developing locales.

The components work together to lower both cost and operational burden:

• Cloudflare Network Interconnect reaches Cloudflare’s edge across 200+ cities through physical or virtual connections.
• Magic Transit extends connectivity to the wider Internet and deflects DDoS attacks in seconds at the edge.
• Magic Firewall provides a network-layer firewall as a service, filtering malicious traffic before it lands on the community network.

Why bundle security with connectivity? Because plain Internet access is not enough to stay connected. Malicious traffic aimed at a network can saturate its upstream links, prompting providers to throttle or cut off service entirely until an attack subsides. Community networks operate on tighter budgets and thinner margins than incumbent carriers, which makes them more exposed. Cloudflare’s position is that no attacker should be able to disconnect a community from the web just by saturating a link.

Why local networks still can’t reach the Internet

Building a community network is one thing; connecting it to the Internet is another. Communities can and do raise antennas, lay fibre, and build robust internal networks — but the services that link those networks to the wider Internet remain prohibitively expensive, when they are available at all.

Two separate components are involved, each with its own cost:

• Backhaul — the physical or wireless link from the community network to an interconnection point, which may be anything from a local cabinet to a large Internet exchange point (IXP).
• Internet services — provided by a network that connects to the wider Internet and agrees to carry traffic to and from the community network on its behalf.

These are distinct. Backhaul carries packets across a medium; Internet service routes them onto the global network and may be delivered over that same link or at its endpoint. Because Internet service costs dominate and vary with usage, most networks buy both as a bundle from operators — often telecommunications or energy companies — that own their own physical infrastructure.

The irony is that long-distance backhaul is far cheaper and operationally simpler than providing Internet service. If reliable, high-capacity backhaul were affordable, community networks could extend their own governance models to run it themselves. But of everything a community can build and manage, the one cost entirely out of its control is the Internet service itself.

Why the problem persists

The author of this project description, Marwan, draws on his experience as a computer science professor and co-founder of HUBS c.i.c., a not-for-profit backhaul and Internet provider in Scotland. HUBS is a cooperative of more than a dozen community networks — some serving areas with no road access — across thousands of square kilometres of Scotland’s West Coast and Borders. Like many community networks, including several of Pangea’s launch partners, HUBS has won awards and is active in policy advocacy.

Despite engagement with funders, development agencies, three levels of government, and countless communities, the answer to why the problem persists remains unclear. But some observations stand out:

• Animals chew cables on the ground — even small ones, and even when buried 15 centimetres deep, apparently drawn by the electrical current.
• Property owners may need convincing that 8–10 square metres for a small tower in exchange for free Internet and community benefit is worthwhile.
• Legislation sometimes blocks small towers on the assumption that private non-residential structures can only be a shed, or no taller than one.
• Privately run fibre backbones installed with public funds are often inaccessible, or priced by distance even though the cost to light 100 metres of fibre is identical to lighting 1 km.
• Civil servants can be enthusiastic but cautious; patience and persistence are required.
• Where national telecoms have plans to serve an area, it is best to avoid trying to deliver service there.
• Tidal fading is real: twice a day, wireless signals over water alternate between excellent and nonexistent.

Thanks to prior efforts from organisations like ISOC, the APC, and A4AI, the challenges and solutions are far better understood now than ever before.

Three ways community networks reach the Internet

There are no purely technical barriers to success. Constraints come from geography and regulation — hills and ordinances shape wireless line-of-sight, property rights and digging rules guide cable routes, and grazing animals gnaw at wiring. But once a community has its own area network, the options for reaching the Internet are more conventional, determined largely by proximity to exchanges, points of presence (PoPs), or regional fibre cabinets. Three broad connection models exist.

Colocation

A fortunate community network may cover an area that overlaps with, or is near, an IXP. In that case, the natural choice is to colocate a router inside the exchange, next to an Internet service provider’s router. Launch partner NYC Mesh connects this way. But exchanges are almost always urban, which makes colocation unavailable to most community networks.

Conventional point-to-point backhaul

Remote community networks need a physical long-distance link to the exchange, with a dedicated router at each end. This is familiar territory, but hinges on a big “if”: the infrastructure must exist. When it does, backhaul is most often sold by a utility company that may bundle Internet service to reduce total cost. Even bundled, the cost is variable, grows with distance, and is unaffordable to any single community network. Some networks have secured backhaul through universities, research networks, or publicly funded infrastructure obliged to serve the public interest. On Scotland’s west coast, Tegola launched with service from the University of the Highlands and Islands and the University of Edinburgh.

Backhaul cooperatives

The final model, used by several launch partners, overcomes prohibitive costs by forming a cooperative in which subscriber community networks are also members. Here, a collection of community networks may optionally interconnect with each other for reliability, and each has a connection to a remote router at the near end of the point-to-point link. Crucially, the cooperative owns and operates both the backhaul service and its colocated endpoints, turning an unaffordable expense into a shared cost.

Launch partners Guifi.net and HUBS c.i.c. are organised this way; their decade or more of operation demonstrates both success and sustainability. Because the backhaul provider is a cooperative, members have a say in how revenue is saved, spent, and reinvested in the shared infrastructure.

Project Pangea and how to apply

Cloudflare’s mission is to help build a better Internet for everyone — not just those whose geography grants them privileged access. Project Pangea extends the faster, more reliable, more secure Internet the company works to build to underserved communities, offering free service of up to 5 Gbps at p95 levels for eligible networks.¹

Community networks interested in participating should visit the Project Pangea landing page to learn more and apply.

The project is the culmination of many hours of effort across multiple teams at Cloudflare, and the company acknowledges the many contributors who brought it together in the spirit of community that the project itself seeks to support.