I’ve been thinking about a conversation I had in a hotel lobby in Singapore last month — the kind of conversation that doesn’t make it into pitch decks or keynote slides. A friend of mine who works in procurement for a Southeast Asian government was describing the process of selecting a facial recognition vendor for a national ID program. The shortlist had three companies: one Chinese, one Israeli, one French. No American firms. Not because they weren’t technically capable, but because — and I’m quoting here — “the Americans ask too many questions about use cases, and the Chinese just ship.”
That sentence has been rattling around my head ever since. Not because it’s surprising — anyone paying attention to the global surveillance technology market already knows this — but because of how cleanly it captures a dynamic that almost nobody in the Western tech ecosystem seems willing to name directly: the most consequential infrastructure buildout of the 2020s isn’t happening in San Francisco or Berlin. It’s happening in Nairobi, Shenzhen, São Paulo, Dhaka, and Lagos. And it’s not a buildout of cloud computing or fintech rails or AI copilots. It’s a buildout of surveillance — comprehensive, layered, and accelerating — in precisely the places where civil society has the least institutional capacity to push back.
When I say “surveillance stack,” I don’t mean a single camera on a pole. I mean the full vertical integration: biometric ID systems at the base layer, facial recognition and behavioral analytics in the middle, predictive policing and social scoring at the application layer, and cloud infrastructure tying it all together. Each piece is sold separately — often by different vendors, in different procurement cycles, under different justifications — but the result is cumulative and architectural. It becomes a system, even when no one planned it as one.
The numbers are staggering. The global video surveillance market is projected to exceed $80 billion by 2026, according to MarketsandMarkets. But that headline figure obscures the geography. The fastest growth isn’t in North America or Western Europe — it’s in sub-Saharan Africa, South Asia, and Latin America. These are the regions where governments are leapfrogging from minimal digital infrastructure to comprehensive biometric surveillance in a single generation, often funded by Chinese development loans or enabled by European export credit agencies that face almost no democratic scrutiny.
Huawei’s “Safe City” initiative — now rebranded and fragmented into dozens of subsidiary projects — has been deployed in over 100 cities across more than 40 countries, with the densest concentration in Africa and Southeast Asia. ZTE, Hikvision, Dahua — these aren’t fringe actors. They’re publicly traded companies with market capitalizations in the tens of billions, and their fastest-growing market segments are in the Global South. The Israeli firm Cellebrite, which sells phone-cracking technology to law enforcement, counts agencies in dozens of developing nations among its clients. The French company Idemia — formerly Morpho — supplies biometric systems to governments from Senegal to Bangladesh.
And here’s the part that really unsettles me: almost none of this is secret. It’s all in public filings, procurement databases, and trade publications. The information exists. What doesn’t exist — at least not in any meaningful, sustained way — is a conversation about it in the places where tech power is concentrated.
I have a few theories about the silence, and none of them are flattering.
The first is simple market logic. The major American tech companies — Google, Microsoft, Amazon — are competing fiercely for cloud contracts in these same regions. AWS and Azure are expanding their data center footprints across Africa, South Asia, and Latin America at an unprecedented pace. The cloud market in these regions is projected to grow at compound rates exceeding 20% annually through 2026. When your growth strategy depends on winning government contracts in Kenya or Brazil or Indonesia, you don’t lead with public criticism of those governments’ surveillance ambitions. You lead with partnership language. You lead with “digital transformation.”
The second theory is ideological. Silicon Valley’s self-narrative is still fundamentally libertarian — technology as liberation, connectivity as democracy, data as the new oil that lifts all boats. This narrative has no room for the possibility that the tools being built and exported are, in practice, instruments of authoritarian consolidation. Acknowledging that would require a level of moral inventory that the industry has shown very little appetite for.
The third — and this is the one I find most damning — is geographic parochialism. The AI ethics discourse that dominates Western tech media is overwhelmingly focused on domestic concerns: algorithmic bias in hiring, deepfakes in elections, the future of creative work. These are real issues. But they exist within a regulatory and civil society framework — however imperfect — that includes courts, journalists, advocacy organizations, and functioning legislatures. The surveillance buildout in the Global South is happening in contexts where those countervailing forces are weak, captured, or nonexistent. And because the people affected don’t attend TED talks or write for Wired, their experiences don’t register in the discourse.
I’m not writing from a position of purity here. I run a media company — Silicon Canals — that covers European and global tech. We write about cloud infrastructure, market expansion, and venture capital. The same economic forces I’m describing shape our editorial landscape. When a European surveillance tech firm raises a Series B, we have to decide how to cover it — and the honest truth is that the commercial framing is always easier, always more legible to our audience, than the civil liberties framing. The incentive structure of tech journalism itself is part of the problem.
What makes this moment different from previous waves of authoritarian technology transfer — Cold War arms sales, for instance — is the nature of the infrastructure being built. Physical weapons degrade, become obsolete, require maintenance and ammunition. Surveillance infrastructure, once installed, becomes self-reinforcing. The data it collects justifies its expansion. The behavioral patterns it identifies create demand for more granular monitoring. The bureaucracies built around it develop institutional interests in its perpetuation.
Consider Kenya’s Huduma Namba program — a national biometric identity system that, when it was proposed, was framed as a tool for financial inclusion and efficient service delivery. Reasonable goals. But as Access Now and other digital rights organizations documented, the system was designed with minimal data protection safeguards, no meaningful consent framework, and expansive government access provisions. The Kenyan High Court actually struck down parts of the program in 2020, but the underlying infrastructure — the biometric databases, the facial recognition capabilities, the integration points with law enforcement systems — didn’t disappear. It was repackaged, renamed, and redeployed under new legal authority.
This pattern repeats across the Global South with eerie consistency. In Brazil, São Paulo’s metro system has deployed facial recognition cameras that scan millions of commuters daily — a system built by Chinese and Israeli vendors and operated with virtually no public oversight. In Bangladesh, Rohingya refugees in Cox’s Bazar were subjected to iris scans and fingerprinting as a condition of receiving humanitarian aid — biometric data that was subsequently shared with the Myanmar government that had persecuted them. In Uganda, Huawei-supplied surveillance cameras installed in Kampala were used to identify and track opposition supporters during the 2021 elections.
These aren’t edge cases. They’re the central use case. And the market size projections that analysts cite so enthusiastically — all those billions in growth by 2026 — represent, in large part, the monetization of this dynamic. Every percentage point of market growth in the Global South surveillance space corresponds to some increment of civic capacity being foreclosed.
I want to be precise about something: this isn’t just a China story. The Western narrative that frames surveillance technology transfer as primarily a Chinese geopolitical strategy — “digital authoritarianism” exported via the Belt and Road — is both partially accurate and deeply self-serving. Yes, Huawei and Hikvision are major players. Yes, the Chinese government has strategic interests in building digital dependencies. But European companies are neck-deep in this market too.
Idemia, headquartered in France, is one of the world’s largest biometric technology providers. Atos, also French, builds integrated security platforms for governments across Africa and the Middle East. BAE Systems’ intelligence division sells mass surveillance tools. The Swedish-Finnish firm Ericsson competes with Huawei for 5G contracts that include surveillance-adjacent capabilities. German firms supply the encryption-breaking tools. Dutch companies build the interception platforms.
And the American role, while more indirect, is no less significant. The cloud infrastructure that stores and processes surveillance data increasingly runs on AWS, Azure, or Google Cloud — including in markets where the data is being used to monitor dissidents, journalists, and minority populations. The AI models that power facial recognition and behavioral analytics are often trained on datasets assembled by Western researchers. The venture capital that funds surveillance startups — even those based in Israel or Singapore — flows overwhelmingly from American limited partners.
The tech ecosystem isn’t divided into “good Western companies” and “bad Chinese companies.” It’s a single, interconnected system in which every layer of the stack — from hardware to cloud to AI — involves actors from multiple geographies, all of whom have financial incentives to look away from downstream use cases.
I don’t have a clean policy prescription here, and I’m suspicious of anyone who does. The dynamics I’m describing are deeply structural — rooted in global capital flows, geopolitical competition, and the fundamental economics of software (zero marginal cost, infinite scalability, no natural geographic boundary). You can’t regulate your way out of a system that spans 50 jurisdictions and involves actors who are specifically seeking out the least-regulated ones.
But I do think there are things that could change — not through policy alone, but through a shift in how the people who build and fund technology understand their relationship to its consequences.
First, export controls on surveillance technology need to be taken as seriously as export controls on weapons. The EU’s dual-use regulation, updated in 2021, was a step — but enforcement has been anemic, and the categories are too narrow. Human Rights Watch has called for a global moratorium on the sale of surveillance tools to governments with documented records of human rights abuse. That’s a starting point, not a ceiling.
Second, the cloud providers need to be held accountable for what runs on their infrastructure. If AWS or Azure is hosting the databases that power a facial recognition system used to track political dissidents in East Africa, that’s not a neutral infrastructure provision — it’s a choice. The “we just provide the platform” defense is the tech equivalent of “I just drive the getaway car.”
Third — and this is the one that hits closest to home — the tech media needs to stop treating the Global South as a market and start treating it as a place where people live. Every article that frames Africa’s “growing digital economy” as an investment opportunity, without simultaneously interrogating the surveillance infrastructure being built alongside it, is participating in a kind of epistemological violence. We are rendering invisible the very people who bear the greatest cost of the systems we celebrate.
I include myself in this critique. I’ve written market analysis pieces that optimize for the same readability and commercial relevance I’m criticizing here. The incentives of the attention economy don’t exempt anyone — they just make it easier to look at aggregate numbers than at individual faces.
There’s a concept in moral philosophy called “the identifiable victim effect” — the idea that we respond with urgency to a named, visible individual, but with indifference to statistical suffering. The global surveillance stack exploits a version of this at industrial scale. The people being surveilled — street vendors in Lagos, factory workers in Dhaka, favela residents in São Paulo — are so far from the conference rooms where procurement decisions are made, so far from the venture capital offices where funding decisions are signed, so far from the editorial desks where coverage decisions are prioritized, that their experience simply doesn’t register as real.
And that distance isn’t accidental. It’s structural. It’s built into the same economic geography that makes the Global South attractive as a market in the first place: lower regulatory friction, less organized civil society, fewer journalists covering the beat, weaker courts, more pliant legislatures. The very conditions that make these markets profitable for surveillance vendors are the conditions that make resistance hardest for the people being surveilled.
I keep coming back to that conversation in the Singapore hotel lobby. “The Americans ask too many questions. The Chinese just ship.” What strikes me now, weeks later, isn’t the contrast between American caution and Chinese pragmatism. It’s the shared assumption underneath — that the people on the receiving end of these systems don’t get to ask questions at all. That their role in this transaction is to be the market, not the citizen. Not the person. Just the addressable space on someone else’s growth chart.
Almost nobody in Silicon Valley is talking about this. And the silence — comfortable, profitable, geographically convenient — is itself a form of architecture. It’s the outermost layer of the stack.
