Government in the Cloud
Authority is migrating from territory to infrastructure stacks—and the question of who controls those stacks is becoming the defining political question of our time.
The Urban Bargain and Its Breaking Point
Start with a simple observation: governance, for most of human history, has been a territorial business. Authority attaches to borders. Obligation flows from jurisdiction. You belong somewhere because you are from somewhere, and that somewhere has a flag, a court system, and a tax office.
But underneath the political theory is a more prosaic story—an infrastructure story.
Cities became the dominant unit of political and economic life not through ideology but through physics. Roads, power grids, hospitals, schools: all of these scale more efficiently where people already are. And people went where the infrastructure was. The logic fed itself. Move to the city for opportunity; find the networks already humming; watch them improve because you arrived. The UN found 55% of the world's population was already urban by 2018 and projects 68% by 2050. Cities, the World Bank has observed, generate roughly 80% of global GDP. Urbanisation was never really a choice. It was the path of least resistance, built into the ground.
Telecommunications continued the pattern. Fibre networks, landlines, cellular towers—all built where density made the economics work. The International Telecommunication Union has documented persistent urban-rural gaps in internet access, especially across lower-income regions. India is an instructive case: despite the BharatNet programme, which has connected over 214,000 local councils with nearly 700,000 kilometres of optical fibre, a recent study found only 3.8% of rural households have high-speed fibre access against 15.3% in cities. The gap between coverage and capability is the tell. Coverage—the signal reaching you—can be dispersed. Capability—the speed, latency, and reliability needed for modern economic life—tends to pool where people already are. The installed base doesn't liberate the periphery. It reinforces the centre.
For a long time, this was a tolerable arrangement. Cities were chaotic, expensive, and often badly run—but they delivered. The deal was simple: accept the congestion, the cost, and the compromises in exchange for access to employment, services, culture, and connections that simply didn't exist anywhere else. Leaving meant stepping off the map. Most people didn't.
That deal is now under real strain. The same forces that made cities powerful are generating costs that governance cannot keep pace with. Housing tells the story most bluntly: the OECD's 2024 Regions and Cities assessment found that housing costs in large urban areas rose 68% over the preceding decade—and that by 2023, a home in a major city cost 86% more than the equivalent in a small one. COVID-19 pushed prices further; the IMF documented record house price increases across dozens of countries during the pandemic years. UN-Habitat, meeting in 2025, reported that more than 2.8 billion people are inadequately housed globally, with over 1.1 billion living in informal settlements—not on the rural fringe, but inside the cities that were supposed to be working. Add congestion, climate exposure, and the slow erosion of institutional credibility, and the picture emerges: the city as governing technology is straining under its own success.
The real trap is structural. Leaving the city was never just about geography—it meant severing the professional, digital, and social networks that modern life runs on. The gravitational pull wasn't only about jobs or culture; it was about the infrastructure that bundled everything together. As long as connectivity rewarded density, decentralisation was mostly aspiration. But what if that dependency could be broken?
Infrastructure as Permission
Every generation or so, a technology arrives that doesn't just improve an existing system—it changes the incentive structure that built the system in the first place. Satellite internet is shaping up to be one of those technologies.
Starlink—SpaceX's low-Earth-orbit broadband service—began commercial rollout around 2020 and has scaled faster than most forecasts anticipated. Six million subscribers by mid-2025. Nine million by December. Ten million by February 2026, across 155 countries, adding more than 20,000 new users every single day. Amazon's Kuiper constellation is expanding in parallel. The FCC has approved further large-scale deployments. Against the backdrop of 5.5 billion people online globally, these are still modest numbers—but the comparison misses the point. Satellite's significance isn't in its market share; it's in where it operates. It reaches the geographies terrestrial build-out could never justify economically. South Africa's Vodacom partnered with Starlink in 2025 precisely because rural infrastructure was too expensive to build. That's not a commercial footnote—it's the core mechanism.
On performance: Ookla measurements put Starlink's median U.S. download speed at around 118 Mbps in 2025, clearing the FCC's 100 Mbps benchmark. Upload speeds and edge-case reliability remain uneven, but the headline capability is real. The more fundamental shift, though, isn't about speed—it's about decoupling. Broadband access no longer requires proximity to a wired grid. The geography of opportunity becomes more porous.
Now, a fair challenge. Financial analyst Patrick Boyle has made a pointed case that satellite internet's growth story is somewhat overstated. His argument runs like this: roughly one billion people globally earn more than $32 a day—a realistic floor for a service priced above $100 a month. Most of those people already have wired broadband that's faster and cheaper. Subtract the overlap and the genuine addressable market shrinks to around 32 million households, most of which are already covered by mobile networks at speeds they consider adequate. Ten million subscribers against a 32-million ceiling isn't a conventional growth story. Compound this with structural cost headwinds—satellites have a roughly five-year lifespan requiring constant replacement, and the economics of space-based infrastructure run about three times worse than equivalent terrestrial data centre capacity—and satellite starts to look like a durable niche rather than a mass-market revolution. Genuinely valuable for remote areas, but structurally ill-suited for cities and priced beyond most of the world.
This critique is accurate—as a snapshot of today. Its limitation is that it treats the thing being changed as fixed. Infrastructure doesn't find demand; it creates it. Electrification wasn't adopted by people who had already decided they wanted refrigerators. Mobile phones didn't scale because a pre-existing population had identified "portable telephony" as an affordable priority. What happens with transformative infrastructure is that it changes what people do, which changes who needs it, which changes the economics of providing it. Satellite is beginning to make connected life viable outside the urban grid. As that normalises—as settlement patterns shift, as work arrangements adjust, as service expectations evolve—the demand signal changes.
The economics follow the behavior, not the other way around. Imagine satellite connectivity becoming the primary channel through which a dispersed household accesses public services, healthcare, financial infrastructure, and education. At that point, the $100 monthly subscription is the wrong unit of analysis. The real comparison isn't satellite versus cable—it's satellite-plus-bundled-services versus the full cost of living in a city. That's a very different calculation. Platform providers and governments who want to be the preferred interface in people's lives have compelling reasons to subsidise connectivity as the entry point. Competitive pressure from Kuiper, emerging EU satellite programmes, and state-backed alternatives will push prices down further. Boyle's addressable market is the market for satellite as a standalone product at today's margins. The market actually forming is something structurally different: connectivity as the precondition for participation in a digitally administered world.
When the Network Becomes the Neighbourhood
Here's where the argument shifts register—from infrastructure to behavior.
For the first time at genuine scale, where you live and with whom are becoming separable from where the installed base is. This sounds technical, but its implications are social and political. Preferences that previously had to yield to geography—values, norms, lifestyle, worldview—can now, at least partially, drive it. The data is already visible. Remote work more than doubled its share of the American workforce between 2019 and 2023, from 5.7% to 13.8%, moving roughly 13 million additional workers off the daily commute. The OECD tracked a corresponding shift in housing demand—a "doughnut effect" where purchasing activity hollowed out from city centres and redistributed into surrounding areas. Governments noticed quickly: by mid-2024, more than 40 jurisdictions had created digital nomad visa schemes, many structured to convert temporary remote stays into longer-term residency. The mobility permit is becoming institutionalised.
The second-order consequence of this sorting is less obvious but more consequential. As value-aligned clusters form and satellite connectivity threads them together, the provider of that connectivity stops being a neutral utility. It becomes the primary interface through which attention flows, services are accessed, and—this is the part worth sitting with—social reality gets calibrated. Infrastructure and governance start to blur.
Look at the scale of existing interface concentration: Google commands roughly 90% of worldwide search as of early 2026. Meta serves 3.58 billion people daily across its platforms. The EU's Digital Services Act, which designates platforms above roughly 45 million monthly European users as systemically significant and subjects them to heightened obligations, is making a structural argument dressed in regulatory language: at sufficient scale, a product becomes an environment. And environments govern the behavior of people inside them, whether that's their stated purpose or not.
Something complementary is emerging at the opposite end of the scale. While satellite provides connectivity from above—global reach, cluster-to-cluster coordination—local mesh networks are rebuilding it from below. Jack Dorsey's Bitchat, launched in July 2025, is an instructive example: a Bluetooth mesh messaging app that works with no Wi-Fi, no cellular service, no accounts, no central servers. Each device is a node; messages hop encrypted from phone to phone. It includes a "panic mode" that wipes all stored data on demand. Within weeks of launch, it saw 70,000 downloads in a single week from Madagascar during political protests—against roughly 360,000 total worldwide downloads at that point. Nepal recorded nearly 50,000 downloads in a single day the same month. This is a well-established pattern: Bridgefy, a predecessor app, was downloaded more than a million times in a single day following Myanmar's 2021 coup. When centralised infrastructure is contested or blocked, people rapidly self-assemble alternatives. The dual stack—satellite for reach, mesh for resilience—isn't a future scenario. It's being assembled right now, mostly outside the attention of policymakers.
Starlink Ad in my X Feed
Satellite internet connections are falling, and fast!
The Stack Is the State
Pull back and look at what's accumulating.
New connectivity infrastructure is enabling communities to form and function outside the traditional urban bargain. Some of this takes dramatic form—Honduras's ZEDE zones were fully autonomous governance experiments with long concession terms, significant enough that their 2024 Supreme Court ruling of unconstitutionality produced genuine political crisis. The broader "network state" movement, documented extensively by WIRED, represents a cultural and financial push toward governance-lite enclaves operating outside conventional political contracts. But the more significant version of this isn't dramatic at all. It's mobile professionals, slowly sorting into communities by shared values, organised through digital platforms, powered by satellite. The individual choices are unremarkable. The aggregate effect is not: when enough households exercise exit, the centre has to adapt—in policy, in service models, in what it's willing to offer. The relationship between centre and periphery becomes genuinely reciprocal. Cities don't lose their dominance, but they lose their monopoly on setting the terms.
Over that connectivity base, something else is happening to governance itself: it's being turned into software. Identity verification, benefits access, health interactions, dispute resolution—functions once siloed across incompatible agencies with incompatible incentives—are being abstracted into interoperable digital services. The World Bank has argued that AI can fundamentally transform public sector delivery through improved coordination and efficiency. The OECD frames it as enabling the re-imagination of how services get designed and provided. Concretely: the UK government's "Humphrey" AI tool already processes public consultation responses at scale. GOV.UK is building toward AI-mediated interaction as the default channel between citizens and the state. Estonia's KrattAI strategy envisions a single assistant-mediated interface to all public services—one interaction point, every function.
Bundle AI mediation with connectivity infrastructure and the system starts to resemble something new: not a government, not a platform, but a governing platform. Access plus identity plus services plus behavioral coordination, delivered through a single stack. This is the point where the analysis has to stop being purely descriptive, because the question of who owns and controls that stack is not a technical question. It is a political one.
Defaults are governance. Prioritisation is governance. Friction is governance. When a private provider designs the AI layer that mediates between a citizen and their services—deciding what information surfaces first, which options appear prominently, which paths are smooth and which are arduous—it is making governing decisions, at scale, without democratic accountability. Not necessarily through malice. Just through product design. The governing environment is being constructed, largely by default, outside any democratic architecture.
States are beginning to recognise this, and some are responding. The concept being operationalised is Sovereign AI: public investment in models, compute, and data governance sufficient to preserve a state's capacity to function without full dependence on private platforms. India's IndiaAI Mission has committed ₹10,371 crore over five years, with a publicly deployed GPU infrastructure of 10,000 units—one of the most explicit examples of sovereign compute being treated as public infrastructure. The EU's EuroHPC AI Factories programme expanded to six additional sites in October 2025; the EU Council's January 2026 framework for AI gigafactories explicitly treats compute capacity on par with strategic energy assets. France has invested over €2.5 billion in AI research and infrastructure since 2018. The EU's "Apply AI" strategy, as reported by the Financial Times in 2025, is explicitly designed to reduce dependence on U.S. and Chinese AI platforms within public administrations. This isn't nationalism dressed up as technology policy. It's institutional recognition that if the stack mediating governance isn't subject to public accountability, then the accountability question doesn't go away—it just moves somewhere harder to see.
Money is part of this picture too. Decentralised finance—DeFi—introduces an economic layer that can operate across borders and outside conventional financial intermediaries. The Bank for International Settlements defines DeFi as financial services delivered on permissionless blockchains; the European Central Bank notes that it eliminates centralised intermediaries through automated protocols. Standard Chartered, as reported by Reuters in October 2025, estimated that stablecoin adoption could pull one trillion dollars out of emerging-market banking systems within three years, with total stablecoin holdings potentially rising from $173 billion to $1.22 trillion by 2028. Chainalysis data from 2024 shows Central and Southern Asia and Oceania leading global crypto adoption—not because those populations are ideologically committed to decentralisation, but because conventional financial systems in those regions are expensive, constrained, or unreliable. DeFi isn't, in those contexts, an alternative for sophisticates. It's a workaround for a system that isn't working.
Put all four layers together—mesh networks providing local resilience, satellite providing global reach, AI orchestrating governance functions, DeFi enabling portable capital—and you have a multi-scale stack that can sustain communities outside the urban grid and outside the conventional state apparatus. This is the genuinely new thing. It's not that any single technology is revolutionary. It's that the combination, taken together, represents the first credible infrastructure for political and economic life that doesn't require territorial governance as its foundation.
Governance has always worked, in part, through the design of default environments—the architecture of a city, the layout of a government office, the sequencing of a benefits application. These are choice environments that shape behavior before individuals consciously engage with them. What's changed is that those default environments are now increasingly digital, increasingly private, increasingly global, and increasingly AI-mediated. The designer of the interface is, in a very meaningful sense, the designer of the governing environment. When that designer isn't accountable to the people being governed, democratic consent doesn't disappear—it just hollows out quietly from the inside.
The decisive question of political life used to be "who governs this land?" It is becoming "who controls the stack through which life is coordinated?" Right now, that question is being answered largely by default—by a small set of infrastructure providers, platform companies, and AI developers operating well outside the architecture of democratic accountability. The path forward isn't to reject the stack; it can't be rejected, and many of its capabilities are genuinely liberating. The argument is simpler than that: the construction of this stack is a political choice, not a technical inevitability, and it should be subject to the same democratic contestation we apply to everything else that governs how people live. We're still deciding whether government in the cloud is public infrastructure or private. That decision has a closing window.
The author is a behavior design expert and analyst working at the intersection of digital infrastructure, governance, and collective decision-making.
