Skip to content
INTERNET INFRASTRUCTURE & NETWORKS

Global Internet Infrastructure Faces Massive Shifts as Hyperscalers and LEO Satellites Reshape Capacity and Economics

The evolving landscape of global Internet infrastructure took center stage in mid-October 2026 as network operators, engineers, and industry leaders gathered in Miami, Florida, for the North American Network Operators Group (NANOG) 98 meeting. Delivering the opening keynote presentation, Paul Brodsky of TeleGeography offered a comprehensive look at the historical trajectory, current pressures, and future economic realities driving global Internet infrastructure.

The overarching narrative of the Internet’s physical infrastructure since its inception has been defined by unrelenting demand growth. During the early 1990s, growth rates surged to extraordinary levels as the core National Science Foundation Network (NSFNET) in the United States transitioned into a mixed hybrid infrastructure. This period saw commercial providers establishing parallel networks alongside existing academic and research institutions, resulting in common annual growth rates of 100% across nearly all parts of the network.

Sustaining such massive demand growth over extended periods presents exceptional operational and logistical challenges. It requires a constant, rolling refresh of the underlying physical infrastructure and, periodically, a comprehensive overhaul of the internal architecture of the global platform. While overall growth rates have naturally moderated from those unsustainable early peaks, positive growth has persisted uninterrupted to the present day, matching the ever-increasing demands placed on network infrastructure worldwide.

To manage these sustained pressures, the industry recently completed a major architectural shift marked by the large-scale adoption of replicated service points and the strategic overlay of content distribution networks (CDNs). This redesign allows content and services to be pre-provisioned much closer to end users, aligning neatly with modern consumption profiles that demand instant, on-demand access. The primary motivation behind this architectural evolution was to push network growth pressures—and their associated infrastructure investment costs—out toward the edge of the network. By doing so, operators successfully relieved intense bottlenecks that were previously choking the inner core of the network’s infrastructure.

Global Internet infrastructure at NANOG 98 | APNIC Blog

This decentralization and redesign of the network platform has proven effective in smoothing out growth curves. When examining international bandwidth demand over the past 15 years, annual growth levels have steadily trended downward year-by-year. However, today’s 30% annual growth rate still represents a massive quantitative jump in absolute terms, driving continuous multi-billion-dollar expansions in total international bandwidth capacity. Each annual increment now demands an additional two Petabits per second (Pbps) of in-service capacity globally, a figure projected to climb to more than ten Pbps of additional annual capacity by the year 2035.

Network planners must also factor in the finite service life of subsea cable infrastructure, which typically spans approximately 25 years. A significant portion of the international cable capacity installed around the turn of the century has now reached the end of its operational lifecycle, requiring complete replacement. Current estimates indicate that approximately a quarter of the world’s active subsea cable kilometers will be decommissioned and retired by 2030.

Fortunately, submarine cable technology has advanced dramatically. Modern systems feature an increased number of fibre pairs bundled within a single physical cable, expanded optical transmission capacities driven by the adoption of L and C band optics, and the deployment of two-core fibre strands instead of traditional single-core alternatives. These engineering breakthroughs have steadily improved the unit cost of capacity, translating directly into falling market prices for leased capacity across major transatlantic and transpacific routes. In turn, falling unit costs unlock latent demand, prompting further investments in capacity. These larger, more efficient systems then operate at even greater economies of scale, driving unit costs down further in a continuous economic cycle.

The combination of an aging international submarine cable inventory, the superior efficiency of newly engineered cable systems, expanding global traffic demand, and a strategic industry desire to improve resilience through route diversity is fueling an unprecedented wave of new cable projects. Industry announcements point to consistent annual investments in brand-new cable systems ranging between USD 5 billion and USD 6 billion over the 2026 to 2029 period.

A profound transformation is also underway regarding who is footing the bill for these multi-billion-dollar maritime projects. Major technology giants—specifically Google, Meta, Microsoft, and Amazon—have significantly ramped up their direct financial involvement in subsea cable developments. Today, approximately 38% of all new cable investments are funded entirely by these four hyperscalers, built as private services for the exclusive use of the funding entity. Furthermore, these companies contribute capital to consortium-based arrangements for an additional 25% of projects, leaving just 37% of funding originating from traditional sources with no hyperscaler involvement.

Global Internet infrastructure at NANOG 98 | APNIC Blog

This investment dynamic varies considerably by geographic region. The transatlantic route has become the most extreme illustration of the rise of privately owned and operated subsea infrastructure, with roughly 85% of its capacity now funded directly by private hyperscalers. Google and Meta stand out as the most active players in this arena, maintaining extensive portfolios of proprietary subsea cables.

The driving force behind this decisive pivot toward private cable infrastructure lies in the high degree of market concentration within the modern digital economy. This concentration has allowed hyperscalers to accumulate massive capital reserves, empowering them to finance, own, and operate private cable systems that grant them complete, unmitigated control over service delivery without third-party dependencies. Simultaneously, the traditional consortium model has lost its appeal for these dominant players. When a single lessee controls an overwhelming proportion of a cable’s capacity, traditional consortia break down; if that dominant lessee were to withdraw, the cable entity could face sudden insolvency, leaving remaining funders with stranded, highly expensive assets. Consequently, the rise of the private cable model emerged as a natural market adaptation.

Despite their efficiency, these private cable systems have attracted scrutiny and concern from various governments worldwide. Policymakers have expressed unease regarding the heavy centralization of critical international communications conduits in the hands of foreign private entities. This trend also raises difficult questions regarding sovereign control over the specialized submarine repair vessel fleet and existing international maintenance pooling agreements organized through established submarine cable maintenance zones. Given that these international links constitute vital critical infrastructure capable of crippling national economies if disrupted, relying on communication platforms owned and operated by foreign private entities introduces significant strategic vulnerabilities.

While subsea cables dominate the bulk data movement of the Internet core, international satellite communications have undergone a revolutionary transformation of their own. For decades following the 1960s, geostationary Earth orbit (GEO) satellite systems handled various forms of international communication. However, the geostationary orbital plane represents a strictly finite resource, currently housing 569 active spacecraft. Bouncing data packets off a satellite positioned 36,000 kilometers away introduces unavoidable, extended propagation delays. Consequently, as soon as undersea fibre-optic networks reached various parts of the globe, major network users eagerly migrated their international traffic away from satellite links.

Historically, the communication services that remained on geostationary spacecraft were restricted to broadcasting and serving otherwise unreachable, remote destinations. However, continuous refinements in signal processing capabilities, paired with the arrival of reusable rocketry, enabled a new generation of physically small spacecraft deployed into Low Earth Orbit (LEO). Because these constellations fly much closer to the planet, their individual signal footprints are relatively small, necessitating thousands of operational satellites. A user’s active connection seamlessly switches from one satellite to another every few seconds, resulting in a network capable of sustaining data throughput speeds in the hundreds of megabits per second.

Global Internet infrastructure at NANOG 98 | APNIC Blog

Advanced laser links deployed between spacecraft now allow data packets to be relayed across the entire satellite constellation, effectively bridging oceans while relying on a sparse network of terrestrial ground stations. For an end user equipped with a specialized antenna, the available capacity and latency of an LEO broadband service are remarkably comparable to traditional terrestrial broadband infrastructure.

The financial stakes in the LEO space are staggering. SpaceX has invested over USD 20 billion to build out its Starlink platform, and Amazon’s competing LEO initiative is projected to reach a similar USD 20 billion price tag. By comparison, the combined capital expenditure for all new submarine cable systems globally across the three-year window from 2026 to 2029 is estimated at USD 22 billion.

The financial performance of these space-based networks underscores their commercial viability. In the third quarter of 2026, Starlink reported quarterly revenue of USD 5.25 billion alongside a net profit of USD 2.4 billion. This translates to an annualized profit of approximately USD 9.6 billion on an initial capital outlay of USD 20 billion, yielding an extraordinary return on capital of 48%. By contrast, institutional investors backing consortium submarine cables typically target a much more modest return of 12% to 14%, while traditional terrestrial broadband infrastructure generally delivers returns ranging between 5.5% and 9%.

The fundamental financial divergence between subsea cables and LEO satellite platforms stems from their distinct market positioning. Starlink operates primarily as a last-mile access provider, utilizing a direct retail model to service users with specialized connectivity requirements in remote or underserved geographic locations. Functioning essentially as a retail provider of last resort with a near-monopoly in its specific sector, Starlink can command a strong premium for its services. Because it operates direct-to-consumer, it avoids sharing revenues with intermediary retail partners. Furthermore, the steep capital barrier to entry—hovering around USD 20 billion—naturally eliminates many potential competitors. Submarine cable systems, conversely, feature a much lower financial barrier to entry, where individual cables typically cost between USD 100 million and USD 400 million, leading to more competitive market dynamics and correspondingly lower profit margins.

Meanwhile, Amazon’s emerging LEO service holds the potential to disrupt Starlink’s effective monopoly, though the company’s precise long-term market positioning remains to be fully defined. Ultimately, LEO satellite constellations and submarine cables serve fundamentally distinct and non-overlapping markets, making direct competition unlikely. LEO networks successfully fill a niche by providing last-mile service in demanding, hard-to-reach locations, whereas submarine cables function as high-volume, low-risk conduits for bulk data transport across the core of distributed global networks. Rather than competing, these two vital technologies operate as highly complementary pillars of the modern Internet.