Vol. III, No. 78
Covering 1 December - 14 December 2025
Monday, December 15, 2025
Late edition · A Relentless publication
The Relentless Dispatch
All the fortnight that mattered, in technology and in the world, read next to what we were building at the time.
VENEZUELA

America squeezes Maduro toward the exit, and the region holds its breath

A vast American military build-up in the Caribbean, strikes on alleged drug boats, and open pressure for regime change brought Venezuela's crisis to a head this fortnight, as the Maduro dictatorship faced the most serious threat to its survival in years and the region braced for how it might end.

The confrontation between Washington and Venezuela's authoritarian ruler, Nicolás Maduro, reached its most dangerous pitch this fortnight, as the American administration massed military force in the Caribbean, continued its campaign of lethal strikes on boats it alleged were carrying drugs, killing dozens without trial, and made increasingly explicit its desire to see Maduro gone. Whether the goal was a negotiated exit, a coup encouraged from within, or outright military action, the pressure was extraordinary, the largest American military deployment in the region in decades, and it forced a regime long thought entrenched to contemplate its own end. Maduro, isolated and squeezed, maneuvered for survival; the region, wary of both the dictator and the prospect of American military intervention in its hemisphere, watched with deep unease; and the fundamental question, whether the pressure would produce a democratic opening, a bloody intervention, or merely a more desperate dictatorship, hung unresolved over a country already broken by years of misrule and the exodus of a quarter of its population.

The Venezuela confrontation captured the paradox of the administration's foreign policy in a single crisis. Here was raw American power deployed against a genuine dictator, an oppressor of his own people whose removal many Venezuelans devoutly wish, and yet deployed in a manner, extrajudicial killings, threats of unilateral military action, the treatment of a sovereign nation as a target, that troubled even those who despised Maduro. The same transactional, force-first approach that this paper has criticised all year was now pointed at a target many would agree deserved pressure, and the discomfort was instructive: that the method could be aimed at a worthy target did not make the method safe, because a power that kills without trial and threatens invasion at will, even against a dictator, is establishing precedents and habits that will not stay confined to the deserving.

THE ECONOMY

The Fed cuts again, and the economy runs, strangely, on AI

The Federal Reserve cut interest rates for a third time this year on December 10, continuing its careful easing as the labour market softened, in an economy whose resilience through a year of tariff chaos and political turmoil had surprised almost everyone. Beneath the surprising resilience, though, economists noted an unsettling fact: a remarkable and growing share of American economic growth was being driven by a single source, the vast investment pouring into artificial-intelligence data centres and the infrastructure to support them. The economy was running, to a striking degree, on the AI build-out itself, a dependence that looked like strength while the investment flowed and would look like grave vulnerability if it ever stopped.

IN BRIEF

A person of the year; a war's uncertain end; a crackdown deepens

The rituals of the year's end proceeded amid the turmoil, the naming of persons of the year, the Nobel ceremonies in Stockholm honouring the patient work of science and letters, small islands of the cooperative, truth-seeking human enterprise the year had so relentlessly assaulted. The negotiations to end the war in Ukraine ground on, the pressure on Kyiv to accept a hard settlement intensifying without resolution. And the administration's immigration crackdown deepened into American cities, the deployments, the raids, the detentions expanding, as the confrontation between federal power and the cities and states that resisted it hardened into a defining domestic struggle.

The Column

The calm that builds the storm

A striking share of American growth now comes from AI investment itself. The worry is usually framed as concentration, one sector too large. The deeper worry is a paradox: that the very stability of the boom is what makes it dangerous.

By Daniel Reeve

A striking and growing share of American economic growth now comes not from what artificial intelligence produces but from the investment in building it, the data centres and chips and power, the spending itself counted as the growth. The standard worry is concentration, an economy leaning too much on a single sector, and that worry is valid. But there is a deeper and more counterintuitive danger, one that the boom's very success conceals, and it was named by an economist who understood, better than anyone, that in finance the seeds of a crisis are sown not in the bad times but in the good.

Hyman Minsky built his career on a paradox he called the financial-instability hypothesis, and its central claim is one the current boom illustrates almost too perfectly: that stability is destabilising, that a prolonged period of calm and success does not make a financial system safer but progressively more fragile, because the very stability changes how people behave. In good times, Minsky observed, when returns are reliable and defaults are rare, everyone, lenders and borrowers and investors alike, gradually revises their sense of what is prudent; the caution that felt necessary in uncertain times comes to seem excessive when nothing has gone wrong for years, and so leverage creeps up, risk is taken more freely, the margins of safety are quietly eroded, all of it feeling entirely reasonable because the good times appear to justify it. Stability breeds confidence, confidence breeds risk-taking, risk-taking erodes the buffers, and the system moves, boom by boom, from a robust state where most commitments are safely covered to a fragile one where they depend on everything continuing to go well, and it does all of this precisely because things have been going well, so that the calm is not the absence of the storm but its incubator.

The counterargument, which the boom's defenders press, is that this time the investment rests on real technology and real demand, not on the pure speculative leverage Minsky studied, and that the AI build-out is funded substantially by the genuine profits of genuinely profitable companies rather than by the tottering debt pyramids of a classic Minsky cycle. There is truth in this, and it is a real difference from the crises Minsky dissected. But the difference is smaller than it comforts, because Minsky's mechanism is not fundamentally about debt specifically; it is about the psychology of prolonged success, the way a long boom erodes everyone's sense of prudent risk, and that mechanism operates on an economy leaning its growth on a single confident bet just as surely as on a leveraged housing market. When a large share of growth comes from investment in one thing, and that investment is justified by the confidence that the thing will pay, the whole structure rests on the confidence continuing, and Minsky's warning is precisely that the longer the confidence holds, the more the economy builds itself around the assumption that it will keep holding, until the assumption is load-bearing and its failure is systemic. The profitability funding the boom does not exempt it from the paradox; it is the very success that, in Minsky's account, breeds the fragility.

And that is the danger the calm conceals, and why an economy that has come to run on a single confident bet is more fragile precisely because the bet has been paying. Each quarter the AI investment delivers, each quarter the confidence deepens and more of the economy reorganises itself around the expectation that the delivery will continue, more capital commits, more growth comes to depend on the spending, more of the whole structure assumes the boom, so that the very reliability of the returns so far is what is quietly converting a bet into a dependency and a dependency into a systemic exposure. The stability is not reassuring; it is the mechanism of the fragility, the long calm in which everyone learns to treat the bet as safe and builds accordingly, until the economy is arranged so that the bet must keep paying or a large part of the whole comes down. Minsky's insight, delivered to an audience that always wanted to believe the good times were simply good, was that the most dangerous moment in any boom is the one when it has gone on long enough that no one can remember why they were ever cautious. By that measure the AI boom is not made safer by its run of success. It is made more dangerous, one confident quarter at a time.

Field Notes
A Relentless build, told plainly

Building the system that outlives you, and the documentation nobody wants to write

This fortnight's column is about an economy dangerously dependent on one thing. This is about the opposite virtue at the level of a single system: building something that does not depend on you, that a stranger could inherit and run, and the unglamorous discipline that makes it possible. No client is named.

By Alosh Paul

The truest test of a system I have built is not whether it works while I am there tending it. It is whether it keeps working after I am gone, when someone I have never met inherits it and has to understand it, run it, change it, without me to ask. And I have come to believe that building systems that can outlive their builder, that a stranger can inherit and operate, is one of the highest and least-practised disciplines in my trade, and that it rests almost entirely on the most unglamorous, most-skipped activity in all of software: writing down how the thing works, so that the knowledge lives in the system and its documentation rather than only in the head of the person who built it.

Here is the failure I have inherited more times than I can count, and the reason this matters. Someone builds a system, and it works, and it works because they understand it, because all the knowledge of how it fits together, why it was built the way it was, what the hidden assumptions are, what will break if you touch the wrong thing, lives in their head. And then they leave, or move on, or are let go, and the knowledge leaves with them, and the system becomes a black box that the organisation depends on utterly and no one understands, a load-bearing mystery that everyone is afraid to touch because the person who knew how it worked is gone and wrote nothing down. This is one of the most common and most dangerous situations in enterprise software: the critical system that only one departed person understood, now a fragile dependency that cannot be safely changed because the knowledge of how to change it walked out the door. The system became a single point of failure not in its technology but in its comprehension, dependent for its continued life on a person who is no longer there.

The discipline that prevents this is boring, thankless, and constantly deferred, and it is the discipline of getting the knowledge out of your head and into a form that survives you: documentation of why things are the way they are, not just what they do; clarity in the system itself, built so that its logic is legible to a stranger rather than clever in a way only you understand; the deliberate, patient work of making yourself replaceable, of ensuring that the system's continued operation does not depend on your continued presence. Almost no one does this well, because it has no immediate reward, because it is always less urgent than the next feature, because, if we are honest, there is a quiet incentive not to do it, since the person who alone understands the critical system has a certain job security in being irreplaceable. But irreplaceable is just another word for single point of failure, and the engineer who builds systems only they understand has not built an asset; they have built a dependency on themselves, which is a liability the organisation will discover the day they leave. The best systems I have built are the ones I could walk away from tomorrow and a competent stranger could pick up, understand from the documentation, and run, because the knowledge lives in the system and not in me. That is the mark of a thing built to last: not that it works while you tend it, but that it keeps working when you are gone, which is the only real test, because everyone, eventually, is gone.

The Ledger
AI
The revelation that AI investment is now a load-bearing pillar of American economic growth is the year's most consequential and least-examined fact: the whole economy is developing a dependence on a single unproven bet, a concentration risk of the largest possible scale, celebrated as strength.
Data centers & power
The build-out is now not just a technology story but a macroeconomic one: the data centres and their power infrastructure are propping up national growth, which means the power question, and the profitability question beneath it, now carries the weight of the whole economy.
Rates
The Fed's third cut continues the easing, but the deeper question is what happens to an economy leaning on AI investment if that investment slows; the central bank's tools are poorly suited to a recession caused by a single sector's correction.
Real estate
US 30-year mortgage 6.17 percent (December 11), the year's low, as easing continues. The one reliable comfort of a turbulent year holds into its final fortnight.
India tech
The concentration of the American economy on the AI bet is a warning for India's services sector, similarly exposed to the AI boom; diversification, of clients, of markets, of bets, is the resilience strategy the year's every lesson recommends.
What we called wrong
Nothing to retract. The economy-on-a-single-bet column extends this paper's oldest theme, concentration as fragility, to its largest possible subject: a national economy growing dependent on one unproven wager, precarious in exactly the way that looks like strength until the bet is called.
The Back Page

The exodus continues

Behind the geopolitics of the Venezuela confrontation this fortnight are the people, and their story is one of the great and under-told human catastrophes of the age. Some eight million Venezuelans, roughly a quarter of the entire population, have fled their country over the past decade, driven out by the economic collapse and political repression of the Maduro years, in one of the largest displacements of people anywhere in the world, comparable in scale to the flight from Syria's war, but caused not by bombs but by the slow-motion destruction of a country's economy and the strangling of its freedoms. They have scattered across South America and beyond, walking across borders, riding buses to the ends of the continent, working whatever jobs they can find in Colombia and Peru and Chile and the United States, sending money home to the family members who stayed, building new lives in the interstices of other people's countries, often unwelcome, often exploited, always missing the home that misrule made unliveable. As the great powers now maneuver over Venezuela's future, deploying warships and threatening intervention and calculating the fall of a dictator, it is worth remembering that the real story of Venezuela is not a geopolitical chess match but a human diaspora of staggering scale, eight million individual decisions to leave everything and walk, and that whatever the powerful decide about Maduro's fate, the deepest measure of the tragedy is those eight million, scattered across a continent, waiting for a country to become a place they could go home to. The warships will maneuver and the dictator's fate will be decided. The exodus, and the longing of eight million people for a home that no longer exists, will outlast every headline.