All Government Is a Tech Company Now (And This Is Not Going Away)

Brazil's Central Bank, the Banco Central do Brasil (BCB), is responsible for Pix. It is the country's instant payment rail, launched in November 2020, that reduced peer-to-peer transaction costs to zero and processed R$ 26.5 trillion in 2024 alone, a 54% increase over the prior year, across 63 billion transactions. Pix now moves more volume annually than Visa and Mastercard combined. It runs 24 hours a day, seven days a week. And it is operated and supervised by an institution that, between 2009 and 2023, lost 31% of its staff, while the rest of the federal executive branch grew by 2%. 

BCB president Gabriel Galípolo has been unusually frank about this in recent Senate hearings, displaying a descending curve of headcount while asking, with increasing urgency, for help. His language escalated from "support" to "help" to, eventually, "rescue." As of April 2026, the BCB has over 3,100 vacant positions. The Auditor role (formerly Analyst), which is the core supervisory function, has a vacancy rate above 50%. And the gap between what the institution needs and what it can hire is not a bureaucratic problem. It's a market problem: a central bank auditor with deep knowledge of payment protocols, APIs, macroeconomics, privacy, and financial regulation can start at R$ 18,000 per month, and has a constitutional threshold of about 45000 BRL (which would be less than U$10000) a month in Brazil. The same profile, in a European financial center, earns multiples of that with better healthcare, better schools for their children, and paid vacations they can actually take.

Galípolo put it simply in a Senate session: sometimes a single BCB supervisor is responsible for monitoring 30 financial institutions. In Europe, teams of up to 30 people audit one.

I'm watching this particular story closely because I think it's the wrong level of abstraction for the question it's actually raising. The BCB's staffing crisis is a symptom of something much larger: the fundamental tension between what modern governments need to operate and what they can structurally afford to build and retain.

Brazil is, for practical purposes, a fully digitized state. There is a constellation of public tech infrastructure that most people outside the country don't fully register. Dataprev handles social benefits and public health data processing for millions of citizens. Dataprev has around 3000 workers. Serpro manages gov.br, tax systems, and a range of other federal services, with around 10,000 employees. Both institutions operate by extending their capacity and combining internal processes, outsourcing arrangements, and partnerships with large cloud providers including Amazon, Google, Microsoft, and Huawei, while charging other public entities, like ministries, for their services, and sometimes selling services for the private market. Almost every ministry also has its own small IT unit for procurement and tendering, managing contracts for SaaS, hardware, data collection tools, and increasingly, generative AI.

What this means in practice is a currency mismatch with no clean solution: the technology these institutions depend on is priced globally in dollars. The taxes that pay for it are collected in reais. And for smaller municipalities, the gap between those two realities is a math problem.

Researchers at the University of São Paulo estimated that the Brazilian public sector spends somewhere between R$ 3.5 and R$ 4 billion per year on cloud computing, software licensing, and digital security, across federal, state, and municipal levels. And that's before cybersecurity, which sits in a separate budget, partly connected to military spending, partly to security forces, and thoroughly underfunded across all of it.

The municipal layer makes this even more concrete. Some Brazilian municipalities are so resource-constrained that their managers can't procure basic computers or fund staff training. The result is a bifurcation: wealthier, larger municipalities can build or buy functional digital services; smaller ones effectively outsource their digital sovereignty to the federal government, which extends centralized software to fill the gap. The federal government becomes, by default, the IT department of hundreds of small localities, which is expensive, creates dependencies, and is not particularly democratic in its design.

This model is already showing its limits. And it's not a Brazilian-specific problem. It is, in fact, the problem that every government in the world that has not yet resolved its digital infrastructure will be working through in the next five to ten years. Brazil is simply farther along the curve, which makes it a more useful case study than it usually gets credit for.

Also, there is a compounding variable here that doesn't get enough attention: the people who need to build, operate, and regulate all of this.

The minimum wage for a public school teacher in Brazil is currently R$ 5,130.63 per month for a 40-hour week which is about $900 USD at current exchange rates. These are the people who are supposed to teach the next generation the mathematics, statistics, logic, Portuguese, history, and philosophy that underpin everything from data literacy to AI readiness. The profession, at the floor, does not pay enough for its practitioners to invest seriously in their own reskilling. And the skills required are about to get more demanding, not less. 

This creates a structural loop with no obvious exit: digital government requires specialized talent, specialized talent commands market rates, market rates exceed what public institutions can offer, the pipeline for producing that talent depends on a teaching workforce that is itself undercompensated and under-resourced. Meanwhile, the software, cloud infrastructure, and AI tools that might bridge some of these gaps are priced by a market that is not interested in pulling costs down for public sector buyers, and that AI-driven moment, if anything, is likely to increase the premium on skilled professionals, not reduce it.

These are the questions I keep returning to, and I don't think there are clean answers:

What is the actual cost of digital government? Not just procurement but the full cost, including the human infrastructure needed to operate it, regulate it, and ensure it doesn't fail in ways that expose millions of people to harm.

Where will countries like Brazil find the professionals to build sustainable public software? If they buy from the market, where does that market find its own people? If those people price in dollars, where do those dollars come from in a public budget denominated in reais?

And at what point does the gap between what digital government requires and what it can pay become a democratic problem, not just a fiscal one?

Brazil built Pix. It works extraordinarily well. Now the institution that built it is asking for financial independence so it can pay its workers. To me, that is a preview.

These are questions I would like to see seriously addressed by institutions like the World Bank and the IDB (the same ones actively pushing governments toward full digitization). The ecosystem needs to be thought of as what it actually is: not a single institution deploying a single solution, but a layered, interdependent system with real costs that compound over time. I want to read rigorous studies on what it actually costs to build government AI infrastructure,  the kind my son will be paying for with his taxes in twenty years. Somehow, we have to make it payable. Not by cutting public health or education or the public services that are already underfunded, but by thinking systematically about the full pipeline: from teacher salaries to public servant recruitment to long-term technology sustainability. The obvious retort is privatization. I don't think that's the answer either. Outsourcing this constellation of services to private operators, funded by government contracts and people's taxes, would create a massive oligopoly with no competitive pressure to improve. Services don't become more efficient or more equitable through privatization and we have enough evidence of that across sectors. And in systems with extractive political institutions, like Brazil's, privatization of critical digital infrastructure doesn't reduce corruption. It just changes who benefits from it.

Efficiency gains from digitization will not automatically lower prices. And if the institutions funding and promoting digital public infrastructure don't account for its true costs, including the human ones, they risk building systems that are impressive to announce and impossible to sustain.

Yasodara Córdova thinks and builds at the intersection of digital identity, fraud ecosystems, privacy, security and why systems break in ways nobody planned for. She has developed work with Harvard University, the World Bank, W3C, the World Economic Forum, and SXSW, among others.