The Squeeze

Yesterday, Jack Dorsey laid off 40% of Block’s workforce . Not because Block was failing — gross profit was up 24% year-over-year — but because, in his words, "something has changed." A smaller team, armed with intelligence tools, could now do the work of 10,000. Wall Street rewarded him with a 24% stock surge in after-hours trading. The market didn't mourn the 4,000. It celebrated the efficiency.

Earlier in the week before Dorsey gutted Block, Perplexity AI released a product called "Computer" (similar to Claude Cowork): a system that orchestrates multiple AI models to execute entire workflows end-to-end. Within hours, someone used it to build a functional market-analysis terminal . The Bloomberg Terminal, a $30k per-year-per-seat institution that has been the backbone of global finance for decades, was functionally replicated for $200 a month. Let that sink in. Bloomberg generated $12.6 billion in revenue last year, largely from terminal subscriptions. Perplexity did this with a fraction of the headcount, a fraction of the history, and a fraction of the cost. The iShares Expanded Tech-Software Sector ETF (IGV) is down roughly 24% year to date, as investors begin to internalise what this means for traditional software businesses

I believe what follows; this transitional period between the old economy and whatever emerges on the other side, will be one of the most painful economic adjustments of our generation. This is what's been troubling my mind. Not a vague anxiety about the future, but a specific, structural intuition about what happens next — the period between now and whatever new equilibrium AI eventually creates. I've been calling it The Squeeze.

What I Mean by The Squeeze

Every major technological shift in history has followed a pattern: disruption, displacement, suffering, adaptation, and eventually, expansion. The printing press displaced scribes but created publishers. The industrial revolution displaced artisans but created factory workers. The internet displaced travel agents but created an entire digital economy.

What's different about AI is the compression of the timeline and the breadth of the displacement. Previous revolutions targeted specific labor categories: manual labor, repetitive tasks, information retrieval. AI targets intelligence itself. And intelligence, unlike manual labor, is what the modern economy was designed to reward.

I studied statistics. I picked up software engineering. I accumulated skills I believed were durable; data analysis, systems thinking, building products from code. These were supposed to be the safe bets, the fields people pointed to when they said, “Get into tech.” I have recently come to a difficult but honest acceptance: everything I have studied and learned along the way, AI will do better and faster. My economic contribution —the specific intelligence I sell to the market—is being automated away. Not in some distant future. Now.

This isn't self-pity. It's arithmetic.

The entire class of knowledge workers (analysts, designers, writers, junior developers, financial modellers, etc.) who trade cognitive output for income is staring at a cliff.

The Anatomy of a Squeeze

Here's how I envision it unfolding. Think of it as a pressure system; economic pressure building from multiple directions simultaneously, with no immediate release valve.

The Individual Squeeze

If you are a knowledge worker, your competitive moat just got shorter. Dorsey said it plainly: "Within the next year, I believe the majority of companies will reach the same conclusion and make similar structural changes." He's not wrong. Pinterest, CrowdStrike, Chegg, Salesforce, Amazon — they've all already started citing AI as the reason for cuts. The pattern is consistent: companies aren't shrinking because they're losing revenue. They're shrinking because they're discovering they don't need as many people to generate that revenue. You will soon be competing against 1000s of displaced professionals for a shrinking number of positions. You will have to apply to a hundred jobs to land what used to take 10 applications. Salaries for middling competence will compress. Mediocre intelligence will no longer be rewarded. The only roles that hold value will be those at the extreme ends—the prodigies, the AI researchers, the physicists, the people whose thinking genuinely pushes boundaries—and even those salaries may face downward pressure as the supply of displaced talent surges. Hiring companies, overwhelmed by volume, will automate their screening too. You will be at the mercy of an AI to even get seen by a human. The cruel recursion of it: the technology that took your job will also gatekeep your next one.

The Corporate Squeeze

Any company whose primary value proposition is white-collar intelligence; analysis, research, synthesis, code, design; is quite literally, one model away from being AI’d away. The moat that once existed in having smart people in a room, thinking hard about complex problems, is eroding in real time. Amazon has said it needs “fewer layers” to operate quickly, calling AI the most transformative technology since the internet. The market has spoken: fewer people, more intelligence tools, higher margins. This is the new equation.

The Government Squeeze

Now zoom out to the macro level, and things get truly uncomfortable, and here, I think, the conversation is dangerously underdeveloped — especially in Africa.

According to the OECD’s Revenue Statistics in Africa 2025, taxes on income and profits account for roughly 40% of total tax revenues across the 38 African countries surveyed, with personal income tax contributing about 16.5% and corporate income tax about 21.4%. The average tax-to-GDP ratio across these countries is just 16.1%—already well below OECD countries at 33.9%. Now consider what happens when AI begins to displace the very workers whose income taxes fund government operations. If a government is earning a meaningful portion of its revenue from economic activities that require white-collar intelligence and those workers suddenly lose their livelihoods before the economy adjusts to new realities, there is going to be a fiscal squeeze of significant proportions. This will be most pronounced for economies that have not managed to build their own AI platforms and are mostly on the receiving end of AI products built elsewhere. These governments will not even generate tax revenue from the AI models that displaced their workers. The value extraction runs one way.

And it gets worse. Consider agricultural economies. Farming and agriculture—long assumed to be insulated from automation because of cheap human labour—are increasingly vulnerable to AI disruption through precision farming, autonomous equipment, and synthetic biology. Economies that previously held a competitive advantage through abundant low-cost labour will find that advantage neutralised when a country with better AI and robotics can produce the same goods locally, cheaply, and at scale. If a country like Uganda or Kenya previously exported products that the recipient country can now produce domestically in abudance because it has access to superior AI and automation, that is less revenue for the exporting nation. Trade balances shift. Export economies contract. The revenue squeeze tightens from multiple directions.

Most African governments are not thinking about this. There is no post-AI economic strategy. There is no conversation about how to tax AI-generated value. There is no plan for what happens when the education system producing graduates whose skills are obsolete before they receive their diplomas. There is no doubt in my mind that this dynamic, left unmanaged, could cause government fiscal crises—and in some fragile states, political instability.

The Education Paradox

This one is personal to me, because I've watched it play out in Uganda. Long before AI entered the conversation, Uganda was already producing a crisis of misaligned human capital. According to the World Bank, around 700,000 people enter Uganda’s labour market every year, but the economy can only absorb just over 200,000; often in low-quality jobs. Only about 13% of Ugandan graduates secure formal employment, according to Mastercard Foundation data. Youth aged 15 to 24 who are not in employment, education, or training stand at a staggering 42.6%.

The education system has been churning out graduates—overwhelmingly in arts and humanities—into a market that has no capacity to absorb them. Families spend massive investments relative to their incomes on degrees that yield diminishing returns. And here is the critical psychological dimension: these graduates have been conditioned to seek office jobs. They have been taught, implicitly and explicitly, that education leads to a desk, a salary, and a title. Many would rather endure prolonged unemployment than pivot to craft work, technical trades, or creative entrepreneurship. The conditioning runs deep.

This is why the Ugandan government eventually moved to reform the curriculum; making it more practical, more skills-oriented. But the damage of decades of misalignment between education output and economic demand has already been done. Today’s education system, across much of Africa, is optimising for skilling that does not meaningfully add to economies. Now layer AI on top of this existing dysfunction. The few formal-sector jobs that did exist for graduates—data entry, basic programming, customer service, administrative analysis—are precisely the jobs most immediately threatened by AI. The squeeze will not arrive into a healthy labour market. It will arrive into one already gasping for air.

What survives the squeeze? Judgment. Taste. The ability to frame problems in ways that machines can't anticipate. Emotional intelligence. The capacity to work with AI, not as a tool user, but as an orchestrator — someone who understands what to ask, how to evaluate outputs, and when the machine is wrong. These are not skills our education systems are designed to teach.

The Paradox of Africa's Position

There is another side to this, and I want to be honest about it even as it complicates the narrative. Africa, precisely because it has a smaller knowledge workforce relative to its total population, may experience slower initial disruption than the West. The first casualties of the AI squeeze are knowledge workers in advanced economies: the software engineers in San Francisco, the analysts on Wall Street, the consultants in London. In much of Africa, the majority of economic activity is still in agriculture, informal trade, and manual services. AI doesn't threaten a boda-boda driver or a market vendor the same way it threatens a data analyst. Not yet.

This could be read as a temporary buffer. But it is only a buffer, not an immunity. I read it as a warning: when the disruption does arrive — and it will, through automated agriculture, AI-enabled logistics, and the erosion of labor-cost advantages — economies that didn't use the buffer period to adapt will be hit harder than those that did. The West will have already restructured. Africa will be restructuring while simultaneously trying to industrialize. That is a uniquely painful position.

The African Development Bank projects that around 100 million young people on the continent won't be able to find jobs by 2030 due to technological advancement. Sub-Saharan Africa needs to create roughly 20 million jobs annually by 2035 just to absorb its young population. These numbers existed before the AI acceleration. They now look almost quaint in their optimism.

And here is where something I have observed on the continent becomes instructive, not as a digression but as a warning.

What Comes After The Squeeze

I want to be clear: I do not believe this ends in permanent collapse. I believe AI will eventually create enormous economic value, new industries, new forms of work, and potentially an era of material abundance that previous generations could not have imagined. I have written extensively about this future. Whereas, history also suggests post-collapse-recovery, it also suggests the transition period can be brutal, prolonged, and unevenly distributed. The industrial revolution created immense wealth — eventually. It also created child labor, urban squalor, and decades of social upheaval before institutions caught up.

Here's what I think comes after the squeeze, if we navigate it even half-competently:

New economic models

The post-AI economy will likely generate value in ways we can barely imagine today, just as the pre-internet economy couldn't have predicted that "social media manager" or "cloud architect" would be real jobs. The challenge isn't whether new value will be created — it almost certainly will — but whether it will be created fast enough and broadly enough to absorb the displaced.

A redefinition of work

If a company of 6,000 can generate $12 billion in gross profit — as Block just demonstrated it intends to — then the old equation of "more workers = more output" is broken. Work will be redefined not as labor exchanged for wages, but as value created per unit of intelligence applied. This is a profound shift, and it demands new social contracts: universal basic income, AI-generated value taxes, new models of ownership and profit-sharing.

A premium on the irreducibly human

Care work. Creative direction. Ethical judgment. Community building. Spiritual guidance. Physical craftsmanship. The things that are valuable precisely because they come from a human being, not despite it. These will become the new high-status work. The irony: the jobs that our education systems currently undervalue — the trades, the arts, the care professions — may end up being the most resilient.

Sovereign AI as a geopolitical imperative

Most governments, especially in Africa, are not even thinking about a post-AI era. They are still operating on economic assumptions from a pre-AI world. This is the most dangerous form of complacency: not the denial that AI will change things, but the quiet assumption that it will change things for other people, in other places, on timelines that do not require urgency. Countries that build their own AI infrastructure; their own models, their own data systems, their own platforms; will retain economic sovereignty. Those that don't will become digital colonies, dependent on foreign AI platforms for their cognitive infrastructure the same way some nations once depended on foreign powers for physical infrastructure. For Africa, this isn't a technology question. It's an independence question.

What I Think You Should Do

I'm not an economist, and I don't pretend to have a complete answer. But I have an instinct, and it comes from watching this unfold both from the inside (as someone whose skills are being automated) and from the outside (as someone who thinks about systems).

If you're an individual: assume your current skills have a shorter shelf life than you think. Learn to work with AI, not as a button-presser but as a thinker who uses AI as cognitive infrastructure. Diversify your economic identity. Don't be one thing. Build something that generates value independent of your labor; a product, a creative body of work, a community, a brand. The squeeze rewards those who are not purely employees.

If you're a company: look at Block's announcement not as an outlier but as a preview. Dorsey said most companies will reach the same conclusion within a year. He's probably right. The question isn't whether to integrate AI — it's whether you're the one doing the restructuring or the one being restructured.

If you're a government; especially in Africa: start the conversation now. What does your tax base look like if 30% of your white-collar workforce is displaced? What does your export economy look like if your trade partners can now produce locally what they used to import? What does your education system need to look like if the jobs it's training people for won't exist in a decade? These are not future questions. They are present questions with a shrinking window for action.

The Uncomfortable Honesty

I want to end with something uncomfortable, because I think honesty matters more than comfort right now. I am a person who studied statistics and taught himself software engineering. I built a career on cognitive skills. And I am watching, in real time, as the economic value of those skills compresses toward zero; not because I got worse at them, but because a machine got better. Everything I've invested in —the late nights, the debugging sessions, the years of accumulating expertise — is being replicated by systems that don't sleep, don't tire, and improve faster than I ever could.

This is not a complaint. It's a data point. And if it's true for me, it's true for millions of others who haven't yet realized it. The squeeze is here. The question isn't whether it will affect you. The question is whether you'll be ready when it does ; or whether you'll be one of the 4,000 who find out on a Thursday afternoon that something has changed.

1. Jack Dorsey (@jack), post on X, February 26, 2026. https://x.com/jack/status/2027129697092731343
2. Aravind Srinivas (@AravSrinivas), post on X demonstrating Perplexity Computer capabilities, February 2026. https://x.com/AravSrinivas/status/2026780915800871331
3. Piero Cingari, “Perplexity AI Just Turned A $30,000/Year Bloomberg Terminal Into A $200/Month Subscription,” Benzinga, February 26, 2026. https://www.benzinga.com/markets/tech/26/02/50893664/perplexity-ai-computer-bloomberg-terminal-software-disruption
4. OECD/AUC/ATAF, Revenue Statistics in Africa 2025: Commonalities and Specificities across African Revenue Classifications, OECD Publishing, Paris, December 2025. https://www.oecd.org/en/publications/2025/11/revenue-statistics-in-africa-2025_d880cbe4.html
5. World Bank, “Uganda’s human capital challenge: Turning development aspirations into reality,” Africa Can End Poverty Blog, November 26, 2025. https://blogs.worldbank.org/en/africacan/uganda-human-capital-challenge-turning-development-aspirations-into-reality
6. Plus News Uganda, “Study Finds Unemployment Remains Ugandans’ Top Concerns,” November 14, 2025, citing Mastercard Foundation and Uganda Bureau of Statistics 2024 Census data. https://plusnews.ug/study-finds-unemployment-remains-ugandans-top-concerns/
7. Ministry of Education and Sports, Uganda. University enrolment grew from 57,114 in 2002 to 345,000 in 2016. See also: National Council for Higher Education (NCHE) records; Employment and Skills Status Report (ESSR) 2022, National Planning Authority, Uganda.
8. OpenAI, “Scaling AI for Everyone,” February 27, 2026. https://openai.com/index/scaling-ai-for-everyone/