All true.

If you want to play the physical aspects of AI Capex, the increase in demand for Copper will exceed supply at this rate of demand growth. There is an investment opportunity here.

True, but incomplete.

After more than 700 pages across 3 agency reports, I discovered that AI is only a small part of the whole copper history.

AI and data centers represent only 10% of the incremental demand for copper.

Sure, demand for data centers and compute can increase at a higher rate than predicted, which in turn will lead to higher demand for power generation and transmission and distribution.

However, as of today, AI only represents 10%. That’s it.

So, what is behind the bulk of the incremental demand for copper?

What the IEA calls the Age of Electricity1.

The increase in demand comes from what S&P Global calls “Energy transition & addition”, and it represents 50% of the incremental demand for copper in the coming decades.

In other words, Energy transition and addition represents 5x the demand AI adds to the copper equation.

That is exactly what I will talk about during this series: the relation between AI, energy and copper in the Age of Electricity.

There will be three articles:

1. The first explains how China accounted for more than 50% of the increase in global energy demand and its impacts.

2. The second will explain the main regions and uses of energy that drive the Age of Electricity.

3. The third article will explain how all of this impacts the demand and supply of copper.

This is important for investors and the readers of Ideas & Capital because it defines a secular growth in the sector. Therefore, it defines the capital flow.

This series builds to copper — but copper is the last piece. To see why the metal matters, you first must see the machine that creates the demand. That machine was built in the 2000s, by China.

Subscribe now

The increase in global demand for energy, powered by China

If a new world order started after United States won WWII, then a new globalization era emerged in the 2000s.

Sure, Nixon and Kissinger started opening the relationship with China in the 70s, but it was only in 2000 that China entered the World Trade Organization.

World trade was never the same. And China, consequently, was never the same.

From that point forward, China focused on increasing production and manufacturing to export its products to the world.

That means that China had to shift from a rural to an industrial economy.

That shift started lifting millions out of poverty. As S&P Global puts it, twenty million people were moving each year from the countryside to cities; they needed places to live and work, and that turned cities into building sites.

So, a new era emerged, a symbiosis of industrialization and urbanization of China. The emergence of what is now the largest energy-consuming economy in the world.

Between 2000 and 2010, China built an average of 3 billion square meters of new gross floor area per year2. To get a sense of the scale and the impact, that is equivalent to adding each year 5 to 6 full New York Cities’ worth of construction.

Let’s get a little bit technical:

Global demand for energy increased by nearly 60% from 2000 to 2024, rising from 416 Exajoules (EJ) to 654 EJ.

China increased its demand by approximately 4x, rising from ~40 EJ to ~170 EJ.

Having added approximately 130 EJ in energy demand since 2000, means that China accounted for more than 50% of the world’s increase in energy demand3.

To put into perspective, China added in 24 years more demand for energy than the energy demand US has today. By the same logic, China added 10x the energy demand Brazil had in 2024.

For the people worried about AI, in electricity alone (not EJ), China has more than 20x the global demand coming from data centers4.

This increase in demand was powered by coal, especially from 2000 to 2010 (left part of the graph below).

As we can see in the right part of the graph, China’s industrialization and urbanization not only drove energy demand, but also propelled its GDP growth.

This is important to understand because every so many years there are structural changes in the world economy and geopolitics that create a secular demand for a specific sector. And today a new secular demand is forming, it is just not in China anymore.

China has already moved from an infrastructure and energy-intensive growth model towards a more modern economy. And modern economies run on electricity.

China’s boom ran on coal and concrete. The next one runs on electricity. Same scale, different fuel.

China will keep adding energy demand — but the next growth vector is electricity, and it’s coming from other regions.

Through 2035, one of the biggest jumps is household electricity, and comes from emerging economies outside China, driven by cooling and appliances.

That will lead the increased in demand per capita, especially in India.

So, you may ask: “That’s obvious, it has been like that for many years, what changed?”

The difference is that electricity became affordable with solar energy. And that changes everything.

Now that the evolution of energy demand is explained, let’s switch to the new growth vector: electricity.

And here’s why this matters for the series: electricity runs on copper. Every solar panel, every wind turbine, every mile of new grid, every EV motor is copper — and renewables use far more of it per unit of power than the fossil system they replace.

So the faster electricity grows, the more copper the world needs.

That also means a change of units. From now on, we measure in terawatt-hours (TWh), not exajoules.

Subscribe now

How China is subsidizing solar energy to the world

China’s expansion in manufacturing was so strong that today it operates in excess capacity in several industries.

In this sense, China produces around 80% of solar panels in the world and has been able to dramatically reduce its costs, especially from 2015 onwards5

That means that the cost and implementation of solar energy became much more accessible for several countries.

This changes how the electricity game is played.

Solar is already enabling millions of people to finally have access to electricity. And this is how the regions I described earlier will increase their demand per capita.

As Ember puts it: Solar is now so cheap that large markets can emerge in the space of a single year – as evidenced in Pakistan in 2024. Homes and businesses in the country soared as a means of accessing lower cost power.

Pakistan imported 17 GW of solar panels in 2024 to meet this growing consumer demand, double the amount imported the year before. Within just a year, Pakistan became one of the world’s largest markets for new solar installations in 2024.

In a global perspective, now 21 countries generate more than 15% of their electricity via solar. 5 years ago, there were only 36.

Solar is the new propulsion for electricity growth

As the IEA puts it:

“For the moment, electricity accounts for only 21% of total final consumption globally, but it is the key source of energy for sectors accounting for over 40% of the global economy and the main source of energy for most households.

This underscores the importance of secure and affordable electricity supply, and the economic and social costs of blackouts such as those seen in 2025 in Chile and the Iberian Peninsula”.

Given the importance of electricity, it is growing at double the pace of total global energy demand.

In 2024, electricity grew 4%, while total demand grew only about 2%. Electricity is rapidly gaining share, and renewables are leading this growth.

For the readers that don’t dominate this subject yet, this is not a distant scenario, that may happen in the future if this or that happens. Solar and Wind already dominate the incremental change for electricity generation, so the discussion here is what will be the pace of the change in the energy mix.

The rapid adherence was so impressive that solar is the fastest technology to reach 2,000 TWh in electricity demand.

This fast-growing pace of implementation means that global solar power is doubling every 3 years.

This trend will not revert anytime soon, if anything, it will accelerate.

The Age of Electricity is already here.

Subscribe now

Conclusion and next chapters

In the 2000s, China was a poor country with enormous potential. Today it’s the second largest economy in the world and the largest energy consumer on the planet.

Its industrialization drove manufacturing costs down, exported deflation to the world, and enabled development across several regions.

Back in 2000, few could picture what China would become and how it would completely reshape the global economy and its energy demand. The investors who saw it early rode a two-decade supercycle.

Most only understood it in hindsight.

We may be at the same point in the cycle again — only in different regions, and with a different kind of energy.

Not coal and raw tonnes of energy, but electricity. And electricity runs on copper.

That’s what this series is really about: spotting the supercycle before it’s priced in.

In the next post, I’ll detail why the need for electricity in India and Asia ex-China is strong enough to drive demand growth for decades — far surpassing AI demand in the US and China.

Then, in the third article, I will lay out the drivers of supply and demand for copper and its potential effects on prices

Subscribe to follow this analysis.

Subscribe now

Disclaimer: This content is for informational and educational purposes only and does not constitute investment advice or a recommendation to buy or sell any security. All views are my own and do not represent those of my employer or any affiliated organization.

After the latest developments, particularly related to Banco Master, I am no longer sure.

And the reason has little to do with the economy.

Brazil is heading into an extremely tight presidential race. Polls show Lula and Flávio close to a statistical tie, with a slight edge for the incumbent president.

When polarization is this intense, rejection matters more than approval.

Anti-Lulismo and anti-Bolsonarismo are stronger forces than genuine support for either candidate.

The result is a calcified vote. Most voters have already chosen their side and are unlikely to move.

The election will be decided by the rest: the independents, the ones who belong to neither camp and can still be persuaded.

That was the central thesis of my April article, and it still holds.

What changed was my view of how these voters could be won.

In April, I thought Lula needed to win the independents.

Today, I think he may only need Flávio to lose them.

And that changes everything.

I laid out most of the structural argument in the original post. Here I’ll focus only on what has shifted at the margin since April.

I recommend starting with my original analysis before reading this update.

Subscribe now

The Effects of Banco Master

Until April, several reports pointed to coordination among center-right governors, particularly Romeu Zema, Ronaldo Caiado and Ratinho Junior.

At the same time, Michel Temer was working to bring Tarcísio de Freitas and Eduardo Leite into a broader coalition.

Temer’s intention was to build a political project for Brazil that did not depend on any single name.

In other words, even though these politicians had never formally endorsed Flávio, the center-right seemed more concerned with removing the PT from power than with who would ultimately be the candidate.

Until then, it looked increasingly likely that center-right candidates would support Flávio in a potential runoff.

Then came the turn: Flávio's involvement with Daniel Vorcaro.

From that moment on, the political chessboard was no longer the same.

Like a shark smelling blood in the water, Zema did not wait long before attacking Flávio:

“Unforgivable. You slapped decent Brazilians in the face.”

That was the beginning of the fracture of Brazil’s right-wing coalition for the 2026 campaign.

To this day, Flávio’s campaign looks weak. It lacks energy and charisma.

Still standing in his father's shadow, he has failed to secure the support of Centrão or of some of the most important political figures within Bolsonarismo.

Nikolas Ferreira, Tarcísio de Freitas and Michelle Bolsonaro did not attend Flávio's campaign launch in Copacabana on August 16.

What I got right

In April, I argued that corruption scandals could be decisive this year, especially if a plea bargain from Daniel Vorcaro brought new revelations in the Master case.

I wrote:

“Corruption scandals such as the Master case and the issues involving INSS retirees are unlikely to fade from public memory before October. If anything, they may intensify, especially with the possibility of a plea bargain involving Daniel Vorcaro.”

What I got wrong

At the time, the Master story pointed at the wife of Alexandre de Moraes, a Supreme Court justice. Viviane Barci, through her law firm, held a R$129 million contract with Master.

Meanwhile, the Master case and the INSS scandal gained traction, with reports connecting Lula’s son, Lulinha, to the latter.

The incumbent government began losing popularity.

However, after The Intercept published new information, Flávio became the figure most closely tied to the Master case.

Alexandre de Moraes, the Supreme Court and Lulinha faded into the background.

Flávio’s involvement killed whatever hope was left for a Brazilian voter looking for a candidate who wasn’t corrupt.

Suddenly, everyone looked the same.

For the independent voter, that changed the equation. If everyone is the same, maybe it’s safer to stick with the devil you know.

What Banco Master Actually Changed

The Flávio–Vorcaro revelations did two things at once:

1. They damaged Flávio’s credibility and fractured the unity of the Brazilian right.

2. They moved the spotlight of the Master case away from Alexandre de Moraes and onto the Bolsonaro family.

That second shift matters because of one group in particular: the independents.

Subscribe now

The Independents

As I argued in April, voters at the polarized extremes barely move. Their approval and voting intention stay flat no matter what the candidates do.

The independents were the exception, the only segment still showing real flexibility.

After the Master case, their read on the candidates changed fast, and a trend in Lula’s favor started to form.

Since April, approval of the Lula government among independents has risen by 10 percentage points, while disapproval has fallen by 11 points.

The same trend shows up in the overall numbers, on a smaller scale.

In both government approval and voting intention, there’s a clear reversal beginning around April 2026.

And it doesn’t look like government policy drove it.

Quaest’s own data suggests measures like the income-tax exemption and Desenrola 2.0 had little effect on voters.

Here’s the interesting part: after the Master case, Brazilians also started seeing more positive coverage of the government and less negative coverage.

All of it fed into a real shift in how independents intend to vote.

Where I was right about the independents

I argued that independents were the only group still showing meaningful variation and therefore the only major group whose votes remained genuinely contestable.

If the polarized camps had already decided, the election would come down to them.

I also quoted Felipe Nunes:

“The voters within this 10% segment are not very ideological. They are pragmatic. They care about public safety, well-being, and their children’s future. They are focused on non-ideological issues and are therefore difficult to win over. Regardless of which candidate emerges, the race will be tight.”

That’s still the backbone of my analysis.

Where I was wrong about the independents

Back then, government measures were doing little. The PT looked cornered. Its populist moves weren’t lifting voting intention, and unemployment was already near record lows.

I argued:

“If the economy is not winning the independents, Lula has no buffer.”

And he really didn’t have one.

But I got the mechanism wrong. I assumed Lula needed to win the independents. He doesn’t.

There’s another way to win a vote in a polarized election: destroy the credibility of the alternative.

Lula doesn’t need the independents to like his government or believe the economy is booming. He only needs to become the lesser evil.

That is the part I underestimated in April.

And this is exactly where the Master case changed the election.

Subscribe now

My Updated View for August

The polls now show a statistical tie, with a slight edge for Lula.

If the vote were held today, I think he’d probably win.

But the race remains open, and small changes can completely alter the outcome.

Nexus (from BTG) provides an important piece of evidence here.

Among non-polarized voters, 37% say they could still change their vote.

Nexus also shows that a significant share of voters supporting other right-wing candidates would migrate to Flávio Bolsonaro in a runoff.

These two numbers remain open.

And they could ultimately decide the 2026 election.

Conclusion

Almost every structural pillar from my first article still holds, especially the argument about independents.

I have little doubt they’ll decide this election. What changed isn’t who decides it, but how.

For Flávio to turn it around, he has to rebuild his credibility, reunite the right and consolidate the right-wing vote in a runoff.

Lula’s job is simpler: keep the election where it is.

The final weeks of a presidential campaign matter, and unexpected events can rewrite the result: Bolsonaro’s stabbing in 2018, the Zambelli episode in 2022, Biden’s debate moment in 2024.

So, with less than three months until Brazil chooses its next president, the question is no longer whether independents will decide the election.

They will.

The question is:

How much can they still be persuaded?

If you enjoyed this type of analysis, subscribe to Ideas & Capital.

I write about markets, politics, economics and geopolitics through the lens of an investor, focusing less on what happened and more on what could happen next.

Subscribe now

He does something far more subtle.

He implants an idea so deeply in Fischer’s mind that Fischer comes to believe it was his own.

The success of the operation depends on a single condition: Fischer must never realize where the idea came from.

That is why inception is so powerful.

The best ideas are not the ones we are convinced to accept.

They are the ones we mistake for our own.

Physical Reality vs Digital Reality

For several years now, the world we inhabit has been divided into two interconnected realities: the physical reality and the digital reality. These two realms constantly interact, reinforce one another, and evolve through a continuous feedback loop.

To clarify my argument, it is useful to define physical reality through Freud’s framework of the three primary sources of human suffering: the body, nature, and our relationships with other people.

Before the emergence of digital reality, our desires were largely shaped by our own inherent and individual impulses. When these desires collided with physical reality, they were either fulfilled, repressed, or transmuted, creating a constant and recursive metamorphosis of what we wanted and who we became.

Let us call this process the dynamics of desire.

Digital reality, on the other hand, can be broadly defined as everything that passes through the filter of an algorithm. The physical device is only the vessel; what matters is the content it delivers. The iPhone is the hardware. Instagram is the digital experience.

The dynamics of desire did not disappear with the creation of digital reality. Rather, a new layer of influence was added to this continuous loop.

Physical reality still shapes our desires. However, digital reality now plays a fundamental role in constructing them through algorithms and the redirection of information flows.

As a result, we now inhabit a dual and symbiotic reality in which it is often impossible to know whether our actions in the physical world were shaped by the digital world—or whether our behavior in the digital world was merely a reflection of desires that originated in physical reality.

The belief in free will stems from our awareness of the choices we make within the constraints imposed by circumstance.

The problem is that if we do not control the circumstances presented to us, nor the desires that arise within us, how can we claim to control their byproduct—our choices?

The circumstance, in Fischer’s case, was the idea implanted through a dream. In our case, ideas emerge through the interaction between physical reality and digital reality.

Like Cobb, the most significant impact of AI will not be its ability to persuade us through reason. Its true power lies in its ability to implant ideas through the subtle manipulation of our attention and realities.

Control the flow of information, and you redirect attention. Redirect attention, and you alter the way people perceive the world.

Algorithms quietly shape our perception of reality.

Our perception of reality shapes our understanding of right and wrong—our moral values.

Our moral values, in turn, serve as a compass for our actions.

In this way, information filtered through AI will increasingly shape our perceptions, our values, and ultimately our behavior.

Six years ago, during his interview with Joe Rogan, Naval Ravikant argued that some of the most powerful people in the world were the engineers writing the code behind Instagram, Twitter, and Facebook.

“They are controlling the spread of information, they are rewriting people’s brains, they are programming the culture, and they are doing it very subtly.”

Today, the same logic still applies. The only difference is that power has shifted toward those building the models behind artificial intelligence—people like Sam Altman and Dario Amodei.

Or perhaps the true source of power is no longer the people writing the code, but the code itself.

After all, we are approaching a world in which AI systems can increasingly rewrite themselves, optimize themselves, and improve their own effectiveness with progressively less human intervention.

This is the second essay in The New Prometheus series, where I explore the intersections of history, philosophy, and economics through the lens of AI to understand where we are and where we might be headed.

Subscribe now

Spinoza on Free Will

Speaking from a metaphysical perspective, I do not believe in free will

So technically, AI cannot steal something we never had in the first place. But for those who do believe in free will, AI may eventually steal the perception of it.

Perhaps the idea for which Spinoza is best known is his conception of an immanent God rather than a transcendent one. But that is not the idea I want to explore in this essay.

The foundation of this article rests on two of his most famous observations:

“Human beings are mistaken in thinking they are free. This belief consists simply of their being conscious of their actions but ignorant of the causes by which they are determined.”

“There is no absolute or free will in the mind; but the mind is determined to will this or that by a cause, which also is determined by another cause, and this in turn by another, and so on ad infinitum.”

Our perception of free will is largely a reflection of our ignorance of the totality of causes.

In a sense, Spinoza anticipated the logic of Inception centuries before Nolan.

We experience our choices.

We do not experience the causes that produced them.

Every day, it seems that a new company emerges claiming it can automate yet another human function through AI agents—from call centers and financial advisors to writers and artists.

As more and more of our reality becomes mediated by algorithms, artificial intelligence is rapidly gaining an exponentially larger share of that totality of causes.

AI is occupying a growing share of the causal chain that shapes how we think and act.

This is the central theme of this essay:

How AI is increasingly shaping our desires, our decisions, our beliefs, our values—and ultimately, our reality.

More Information Does Not Mean More Truth

In Nexus, Yuval Noah Harari argues that throughout history, the most powerful networks—religions, empires, and corporations—have often been built not on factual information, but on myths, stories, and shared beliefs.

The Bible, for example, is not valuable because it provides a precise record of historical events. Its power lies in its ability to connect billions of people through a common framework of beliefs, values, and behaviors.

Information sometimes represents reality, and sometimes it does not.

But it always connects.

If information does not necessarily reflect reality, then the crucial question becomes: how do information itself and the networks that connect it combine to mobilize people?

The true source of power is not information itself. It is control over the nexus that determines what information receives our attention.

As Harari writes:

“Information is not about truth; it is about connection.”

In Inception, Cobb does not need Fischer to believe a true idea.

He only needs Fischer to connect a series of thoughts in a way that leads him to a predetermined conclusion.

The power of the implanted idea comes not from its truthfulness, but from its ability to reorganize Fischer’s internal network of beliefs.

The same principle applies to information networks.

More information does not necessarily lead to more truth.

If it did, then according to Donald Trump’s countless social media posts, a permanent peace agreement with Iran would already have been signed, the United States would have already “obliterated” Iran’s nuclear program, and the conflict would have long since ended.

The point is not whether Trump is right or wrong. The point is that an abundance of information does not automatically produce an abundance of truth.

Anyone following the war closely knows that Trump’s posts can shift dramatically within hours.

One moment he threatens military escalation.

The next he suggests that an agreement is close.

Every statement redirects attention.

Markets react.

The narrative shifts.

New connections are formed.

Others disappear.

Financial algorithms do not interpret these statements the way humans do. They simply process them as information, and markets react accordingly.

We can observe this dynamic in the volatility of oil prices and in the constantly shifting probabilities on Polymarket.

Subscribe now

More than two thousand years before social media, Plato had already identified the mechanism.

“Truth occupies a very unstable position in the world, for opinions —that is, what the multitude comes to believe— are shaped more by the art of persuasion than by truth itself.”

His criticism of the Sophists was not that they lied.

It was that they understood something deeper.

Human beings are often moved more by persuasion than by truth.

The Sophists competed for attention.

Whoever captured attention shaped perception.

Whoever shaped perception influenced judgment.

Whoever influenced judgment shaped action.

In other words, every event can be viewed through an infinite number of perspectives, each generating its own interpretation of reality.

What we call truth is often nothing more than the perspective that receives the greatest share of our attention, while countless alternative perspectives fade into the background.

Those who control the narrative direct attention toward specific perspectives.

By directing attention, they control the points of connection within the information network.

By controlling those points of connection, they gradually influence what society comes to accept as true.

Bringing this argument into the present, the more we interact with algorithms, the more dependent we become on their recommendations.

And the more our decisions pass through AI-driven filters, the greater AI’s influence becomes over the nexus through which information, narratives, and ultimately truth are constructed.

If a small number of AI systems come to dominate that nexus, we may witness the standardization of truth itself—not because alternative perspectives cease to exist, but because they cease to receive our attention.

In many ways, AI may become the most sophisticated sophist ever created.

Garbage in, garbage out

In the previous section, I discussed the monopolization of the nexus. In this one, I want to focus on the quality of the information flowing through it.

The common assumption that more information leads to more wisdom often prevents us from questioning the quality of the information itself.

AI models are trained on the information available to them—primarily the vast repository of human knowledge found on the internet, with all its insights, errors, distortions, and biases.

This may sound obvious, but it is worth emphasizing:

No matter how sophisticated the processing system, the quality of its output can never be completely separated from the quality of its input.

Even the most advanced AI models, despite their remarkable capabilities and increasingly sophisticated methods of refining data, remain subject to this same principle.

Garbage in, garbage out.

The problem becomes particularly relevant when AI systems begin mediating a growing share of our decisions, beliefs, and interactions with reality.

If the data used to train these systems contains distortions, those distortions may not disappear. In some cases, they may even be amplified.

One of the most respected studies on artificial intelligence, Stanford University’s Artificial Intelligence Index Report 2025, provides evidence of this phenomenon.

Researchers examined how increasing the size of training datasets affected racial and gender biases in vision-language models. Using the Chicago Face Dataset, they evaluated fourteen models trained on LAION-400M and LAION-2B, two of the most widely used datasets in the field.

The results were striking.

While larger datasets improved certain aspects of human classification—reducing the misidentification of nonhuman entities such as gorillas and orangutans—they also amplified racial biases.

Among the larger models, Black and Latino men were disproportionately classified as criminals, with classification probabilities increasing by as much as 69% when dataset size expanded from 400 million to 2 billion samples.

In other words, scaling the amount of information improved some capabilities while simultaneously reinforcing certain distortions embedded within the data itself.

More data did not automatically produce better judgment.

It produced a more powerful reflection of the data on which the model had been trained.

The same report also comments on:

“AI models also present implicit bias scores on different stereotype categories. LLMs disproportionately associate negative terms with Black individuals and are more likely to associate women with humanities over STEM fields. The research also finds that LLMs favor men for leadership roles, reinforcing gender biases in decision-making contexts. This finding is significant, as it indicates that while bias appears to have decreased on standard benchmarks—creating an illusion of neutrality—implicit biases remain pervasive, potentially leading to subtle yet meaningful discriminatory outputs.”

In other words, we are not only outsourcing control of the nexus to AI.

We are increasingly allowing AI to influence us through the biases embedded within its own training data.

This becomes even more significant when we consider a trend highlighted by Stanford’s AI Index Report 2026: AI-generated content is now being produced at a scale that rivals—and in some domains exceeds—human-generated content.

We have become so deeply engaged with AI-generated information that it is difficult to imagine this trend reversing. If anything, we are still in the early stages of a self-reinforcing cycle.

AI is trained on human-generated content.

It then generates new content.

Humans consume that content.

Future AI systems are trained on an information ecosystem increasingly populated by AI-generated material.

The snowball keeps growing.

And the challenge is not merely the quantity of content being produced.

It is the content itself. (Pun intended.)

If AI increasingly controls the nexus through which information flows, influences the information used as training input, generates a growing share of the content we consume, and shapes how that content is presented, then human beings cannot help but be shaped by AI in return.

At that point, AI is no longer simply a tool processing information.

It becomes an active participant in constructing reality itself.

Subscribe now

AI as Planetary-Scale Inception

Cobb had to carefully design a dream for a single individual.

AI operates at an entirely different scale.

It does not need to enter our dreams.

It already inhabits our information environment.

Every recommendation.

Every search result.

Every summary.

Every suggested article.

Every generated answer.

Each one subtly alters the set of ideas competing for our attention.

Inception is no longer directed at a single target.

It is directed at billions of people simultaneously.

Conclusion

Perhaps we never had free will to begin with.

The difference today is that we can no longer tell which force exerts the greater influence on the endless feedback loop that shapes our lives: physical reality, digital reality, or ourselves.

It is an infinite recursive loop because we no longer know where the chain of influence begins.

Do our desires shape the digital world?

Or does the digital world shape our desires?

A technology capable of processing more information does not necessarily produce a better understanding of reality.

But human societies are, at their core, information networks.

And whenever a new information technology emerges, society itself is inevitably transformed.

The printing press changed religion.

Radio changed politics.

Television changed culture.

Social media changed attention.

Artificial intelligence may change reality.

Inception was fiction.

A team of specialists needed extraordinary technology to plant a single idea in a single mind.

AI may achieve something similar without entering a dream.

By shaping the information we consume, the narratives we encounter, and the perspectives that compete for our attention, it may gradually influence the causes from which our thoughts emerge

The greatest power is not the ability to force a decision.

It is the ability to shape the causes from which choices emerge.

Cobb called it inception.

We call it artificial intelligence

Subscribe now

Markets are also shaped by ideas that evolve into narratives.

More precisely, ideas become beliefs. Beliefs shape behavior, which in turn becomes individual action. Collectively, individual actions become patterns that permeate society.

As society develops new behavioral patterns, capital flows shift accordingly, ultimately affecting prices.

Ideas, capital, and prices are therefore interconnected in a constant feedback loop.

Another way to approach this concept is through what Yanis Varoufakis calls historical materialism: the idea that history evolves through a perpetual cycle of feedback between, on one side, the way humans transform matter and, on the other, the way human thought and social relations are transformed in return.

Bringing this framework into the present moment, which is the central topic of this essay, AI is a technology powerful enough to transform not only our relationship with matter, but also human thought, behavior, and social norms themselves.

There will be a pre-AI world and a post-AI world. There is little doubt about that.

My concern is the degree of freedom we are giving to a technology we may not be able to control in the future. We are building a Frankenstein potentially more powerful than humanity itself, while CEOs disguise and market it as merely a “productivity enhancement”.

In the end, Hesiod may have been right: Zeus may have no choice but to destroy a humanity incapable of restraining its own power once amplified by technology.

We will not become masters of technology. We may simply become enslaved by it.

Technology is the engine that transforms both matter and human thought. Artificial intelligence may accelerate changes in both dimensions faster than any previous period in history.

The Enlightenment created new intellectual foundations that eventually culminated in the French Revolution and the Jacobin Terror.

The Greeks understood this danger before us.

Prometheus stole fire from the gods and gave it to humanity. The gods did not punish him because fire was evil. They punished him because humanity was not ready for it.

We may be living through the same moment again.

This essay is the first in a series exploring how AI may reshape our values and social norms, potentially leading to a new form of revolution.

To better understand the paths that lead to revolution, we must revisit the greatest revolution of the modern era: the French Revolution.

By the end of this essay, I will draw parallels between the Enlightenment, the French Revolution, and the Revolution of Artificial Intelligence.

So let us begin with the Enlightenment and its foundations

Subscribe now

Pre-Enlightenment

The history of thought can be understood as a gradual evolution, punctuated by occasional disruptions to prevailing intellectual frameworks

The Enlightenment was unquestionably one of those disruptions. Arguably, the greatest rupture in the intellectual architecture of the modern world.

Yet if intellectual evolution is continuous and gradual, it is also path-dependent.

Dissident and revolutionary ideas are built upon the foundations of the systems they seek to challenge.

To understand the Enlightenment, we must first understand the foundations upon which it was built.

Its predecessor was the Scientific Revolution.

Returning to historical materialism, we must examine not only science, but also morality and social relations

From the scientific perspective, we must begin with Francis Bacon (1561–1626).

Before Bacon, European natural philosophy had been dominated by Aristotle for more than 1,500 years.

The Aristotelian method began with general principles and deduced conclusions from them, observing little and experimenting even less. This was the deductive method.

Bacon turned this framework upside down.

His proposition was simple: knowledge should emerge from systematic observation and experimentation, only then arriving at general principles

In doing so, Bacon established the inductive method.

Scientia est potentia.

Knowledge is power.

The goal was no longer to understand the world merely to contemplate it, as the Greeks had done. The goal was to understand nature in order to transform and dominate it for human benefit.

Bacon did not produce major scientific discoveries himself. He did not formulate physical laws or conduct revolutionary experiments.

His contribution was something else entirely: he redefined what science should be and what it should serve.

That may have been more important than any individual scientific discovery.

This was not merely an intellectual evolution.

It was a rupture.

From this new source of knowledge emerged Copernicus, Galileo, and Newton. Together, they created a new scientific compass for human civilization: the universe operated according to rational laws discoverable by the human mind.

From the perspective of morality and social relations, we must begin with Thomas Hobbes (1588–1679).

Hobbes lived through the English Civil War. He witnessed the execution of King Charles I and the collapse of social order.

It shaped everything he wrote.

Hobbes held a deeply pessimistic view of human nature. For him, man was naturally selfish and violent. Without a strong external authority, human beings would inevitably descend into chaos.

In the state of nature, before organized society, human life would be “solitary, poor, nasty, brutish, and short.”

Each individual acts according to self-interest, producing a war of all against all.

Without a strong authority, life becomes chaos.

Homo homini lupus.

Man is a wolf to man.

Hobbes’ solution was a social contract: individuals surrender their rights to an absolute sovereign in exchange for security and order.

Freedom is partially sacrificed for stability.

Hobbes’ Leviathan draws from the Bible: a colossal sea monster, nearly invincible and beyond human control. A symbol of absolute power and primordial chaos.

The sovereign state possesses near-unlimited authority because, without control, the alternative is disorder.

The Leviathan is the monster we create to protect ourselves from ourselves.

It is a state built by individuals that, once created, comes to dominate them all.

In a certain sense, this moral framework also begins to weaken the foundations of monarchy itself. Political legitimacy no longer derives from the divine right of kings, but from a rational contract between individuals.

The intellectuals of the Enlightenment would build upon these foundations.

Subscribe now

Enlightenment

If Newton could explain the movement of planets, why could the same method not be applied to politics, morality, or religion?

Reason became the supreme arbiter of truth.

In Reflections on the Revolution in France, Edmund Burke argued that Rousseau, Voltaire, and Helvétius were the three intellectual pillars that most influenced revolutionary thought.

Voltaire dismantled reverence for the sacred and for the Church. Rousseau created a new source of political legitimacy grounded in reason and popular sovereignty. Helvétius defended a mechanistic view of man

Together, they formed something explosive: destroy the old order, consecrate the people as the ultimate source of legitimacy, and reconstruct man through rational design.

Jean-Jacques Rousseau (1712–1778)

While most philosophers of the Enlightenment celebrated progress, reason, and civilization, Rousseau asked the opposite question:

What if civilization itself was the problem?

Man is born naturally good. Society corrupts him.

In 1750, Rousseau won the Academy of Dijon essay competition with the question: “Has the Progress of the Sciences and Arts Improved or Corrupted Morals?”

His answer was simple: corrupted.

He was perhaps the only Enlightenment thinker to argue that the Enlightenment itself was fundamentally mistaken.

Human inequality is not natural. It is created by private property.

Everything that followed from private property — laws, government, war — emerged to protect that original usurpation.

From private property came inequality, luxury, social dependence, and vanity. Together, they produced an artificial, competitive, and morally corrupted humanity.

The logical conclusion soon followed: if society corrupted man, politics must regenerate him.

Rousseau built upon John Locke’s idea that legitimate government derives from the consent of the governed, but pushed the argument much further.

He introduces the concept of the general will (volonté générale): a collective interest that transcends the mere sum of individual interests. Legitimate government expresses this general will.

When it does not, it becomes tyranny.

From this emerges Rousseau’s central idea: The Social Contract.

Each individual surrenders his will to the political body and, in return, participates in the general will. But the general will is not simply the sum of individual opinions. It represents what Rousseau believed to be the true common good.

But here lies the tension Rousseau never fully resolved.

If the general will represents the true common good, what becomes of the individual who refuses to submit to it?

Rousseau’s answer was troubling: he can be forced to be free.

The individual who resists the general will is not being oppressed. He is simply being corrected — brought into alignment with what reason has already determined to be right.

The Jacobins absorbed this logic completely.

Resistance to the republic was not political disagreement. It was moral failure. And moral failure, in a system grounded in rational virtue, could not be tolerated.”

This new intellectual framework was not merely an evolution.

It was a rupture.

And beyond being a rupture, it was another blow against monarchy, the Church, and traditional customs. Sovereign power no longer descended from God or hereditary authority. It now emerged from the people themselves.

Voltaire (1694–1778)

Voltaire was not an atheist. He was a deist, for in the eighteenth century atheism remained a marginal and dangerous position.

Deism offered an intellectually respectable alternative while simultaneously emptying the Church of its institutional authority.

There is a creator God. The complexity and order of the universe, especially after Newton, seemed to require an intelligent cause. But God does not intervene in human affairs. He does not answer prayers, perform miracles, punish sinners, or choose favored peoples.

Therefore, revelation, sacred scripture, sacraments, ecclesiastical hierarchy, and the Inquisition are not divine mandates. They are human inventions.

If God does not communicate through sacred texts or intermediaries, then the Church possesses no special authority over truth. It becomes a human institution like any other, capable of being criticized, reformed, or rejected.

Voltaire attacked several pillars of the Catholic Church: the contradictions of the Bible, supposed miracles, the Inquisition, and religious persecution.

But his most consequential attack for the French Revolution was directed at the clergy as a political class and at the alliance between throne and altar.

If the Church lacked true legitimacy, then the clergy became merely a class that consciously used popular superstition to preserve political and economic power.

And to preserve that power, ignorance among the population was necessary.

The monarchy and the Church sustained one another. The Church legitimized the king’s authority as divine, while the king protected the privileges of the Church.

Voltaire viewed this alliance as the central obstacle to reform. To attack the Church was therefore an inherently political act.

Reason must expose the absurdity of institutions that shield themselves behind sacred authority.

Voltaire’s intellectual assault eventually contributed to the collapse of the Church’s institutional power, even though the Church had long served as one of the central pillars of Europe’s moral, educational, and social structure.

The destruction of the Church’s authority created a vacuum of power.

And freedom alone does not fill a vacuum.

Force does.

So does the Jacobin Terror.

Claude-Adrien Helvétius (1715–1771)

Helvétius believed that human character is the product of environment.

Virtue does not emerge from a noble soul or deep tradition, but from the proper organization of interests.

No one acts out of disinterested virtue. Every action, even those that appear altruistic, ultimately reduces itself to the pursuit of pleasure or the avoidance of pain.

Virtue is simply self-interest properly understood.

For Helvétius, all human beings are born equal in capacity. If the mind is shaped by experience, then differences in intelligence and character are not innate.

His thesis was both materialist and utilitarian: all human mental activity reduces to physical sensation and to the principles of pleasure and pain

There is no immaterial soul. There is no transcendent reason.

What we call thought, moral judgment, and character are merely products of physical experience and unconscious calculations between pleasure and pain.

Education can transform everything.

If character is the product of environment, if men are born equal, and if no transcendent soul exists, then society itself can reshape man through education and the reorganization of incentives.

At this point, politics begins to resemble social engineering.

Man can consciously alter his environment in pursuit of a desired outcome.

Society becomes a laboratory and the citizen becomes raw material.

Jacobinism absorbed much of this logic.

Politics became an instrument for the creation of public virtue.

The republic was expected to form virtuous citizens, and whoever resisted republican virtue became an enemy of the people.

For Burke, this was the ultimate expression of rationalist ignorance: the belief that society could be redesigned through abstract theory while ignoring the complexity accumulated through history.

Subscribe now

The French Revolution

The French Revolution was the materialization of those disruptive intellectual theories

The Pandora’s box opened by the Enlightenment unleashed a chain of unpredictable events that ultimately culminated in the guillotine and the Jacobin Terror.

The intellectual architects of the age created a Frankenstein that became too powerful and ultimately impossible to control.

Artificial intelligence may possess the power to reshape our values, redefine our concepts of right and wrong, and generate an entirely new disruptive intellectual framework.

Once again, Pandora’s box may be opened. Once again, humanity may create another Frankenstein.

Without getting too far ahead, since this will be explored more deeply in the next essay, let me briefly illustrate where this argument leads.

Imagine an emerging theory that claims our political institutions are fundamentally corrupted by human error and human bias.

Because of that, they become corruptible, manipulable, and ultimately untrustworthy.

Poorly designed institutions produce impunity, inequality, populism, and eventually collapse. Every great empire in history ultimately fell because of human flaws and human limitations.

Until the emergence of AI, there was no true solution to this problem.

Our institutions inevitably reflected human culture and morality, with all their virtues and imperfections.

But now we may be creating an intelligence more powerful than humanity itself, armed with the accumulated historical experience of civilization.

And if AI eventually becomes more intelligent than humanity, do laws written by humans become obsolete?

If AI can perform every human function more efficiently and more rationally, why should governing remain an exception?

A new theory may emerge:

AI should redesign our institutions according to its superior intelligence and accumulated knowledge. Human bias would disappear. Society would become more equal, less corrupt, and more efficient.

Who will become the new Marx capable of articulating this utopian vision?

Now that the reader understands where this argument is heading, we can return to the moral failures of the French Revolution.

Only then can we properly draw parallels with the modern world.

This section draws heavily from Edmund Burke.

Unlike what many Enlightenment thinkers believed, society is not merely a contract among the living generation. Society is a contract between generations — between the dead, the living, and those yet to be born.

Society reflects our humanity in all its virtues and imperfections. The science of jurisprudence is reason accumulated and condensed across centuries.

Institutions are therefore not mere historical accidents. They embody accumulated historical experience.

Revolution is fundamentally different from reform.

Revolution is easy. It means identifying visible flaws, destroying the pillars of tradition and social order, and replacing them with something new, usually utopian and rationally abstract.

Hominem non sapiente.

They do not understand man.

Revolutionaries understand abstractions, rights, formulas, and numerical equality. But they fail to understand psychology, customs, passions, belonging, tradition, religion, hierarchy, and human identity.

They fail to understand that human beings rarely act according to rational self-interest alone.

As Dostoevsky wrote in Notes from Underground, man would:

Society is not a blank sheet of paper.

Revolutions grounded in abstract rationalism and utopian equality ignored human nature, its incentives, and its contradictions. In the end, they often produced tyrannical governments instead.

Ideological revolutions destroy the invisible preconditions of economic and social order, the very foundations that make order possible

Reform, on the other hand, begins with respect for foundational institutions. It seeks to preserve and improve what has already been built.

Reform respects the intergenerational contract.

Burke did not defend a frozen society. Paradoxically, a state without the means to change also lacks the means to preserve itself.

Stable systems are those capable of absorbing change without collapsing.

True consistency often requires adaptation, flexibility, and institutional evolution in order to preserve continuity of purpose.

Reform is built gradually.

It is better to be criticized for excessive caution than ruined by excessive confidence.

France did not commit the common political error of abandoning morality for self-interest, the idea that the ends justify the means.

It committed something worse.

It sacrificed its own interests in order to sustain the illusion of virtue — a new morality grounded in abstract rationalism.

Burke called this the prostitution of virtue.

Burke feared abstract moral ideologies more than ordinary political corruption. Corruption destroys governments slowly. Ideologies can destroy entire societies quickly.

To be metaphysically true does not necessarily mean being morally or politically true.

The Jacobins murdered the monarchy in the name of the metaphysical ideals of liberty, equality, and fraternity. They assumed that by cutting off the head of power, power itself would disappear.

But power never disappears.

It is merely redirected.

Unfortunately, the vacuum left behind was not filled by liberty. It was filled by force, a consequence of the destruction of moral authority and institutional legitimacy.

Nothing becomes more oppressive or unjust than a weak government, because weak institutions gradually abandon procedure and consistency in favor of arbitrary exceptions and unequal treatment.

The worst form of government is not the corrupt one, but the incompetent one with unlimited power.

The decline of religion, the Church, and traditional institutions during the Enlightenment created a new secular god grounded in economics and science.

What new values will artificial intelligence create to replace them?

The Echoes of the French Revolution

The Enlightenment was one of the great turning points in the development of Western intellectual history. Extraordinary thinkers emerged during this period: Voltaire, Rousseau, Hume, among many others.

Yet Burke’s critique should not blind us to the extraordinary achievements of the Enlightenment itself. Constitutional government, scientific progress, individual liberty, and the modern rule of law were also products of this intellectual revolution.

The danger was never reason itself, but the belief that reason alone could fully redesign human society from first principles.

But intellectual disruptions are not necessarily beneficial. Their secondary consequences are often unpredictable.

As the old saying goes: wars are easier to start than to end.

Intellectual disruptions are seismic events whose consequences echo across centuries.

For many years, it was widely believed that when Henry Kissinger asked Chinese Premier Zhou Enlai about the impact of the French Revolution during a 1972 conversation, Zhou replied:

“It is still too early to say”.

The quote became famous over the following decades as evidence of the supposedly long-term historical perspective of Chinese leaders.

In 2011, however, it was clarified that a translation misunderstanding had occurred. Zhou Enlai was most likely referring not to the French Revolution of 1789, but to the Paris uprisings of 1968.

Even so, I believe the original interpretation may still be correct after all.

Hayek argued that the French Revolution created the intellectual environment from which socialism eventually emerged.

And because intellectual history is always path-dependent, socialism first appeared in the form of utopian socialism through thinkers such as Henri de Saint-Simon and Charles Fourier.

Their vision was a society organized rationally and administered by industrialists, scientists, and technocrats rather than by an idle aristocracy.

Over time, these ideas evolved into Marx’s scientific socialism, grounded in a materialist interpretation of history and economics centered on class struggle.

The publication of The Communist Manifesto and later Capital during the nineteenth century would eventually influence Lenin and Trotsky, culminating in the Russian Revolution of 1917.

After the Second World War came the Cold War: the great ideological struggle between capitalism and communism.

And finally, in 1991, the Soviet Union collapsed.

Then came another historical turn.

In the 2000s, China entered the WTO, strengthened its trade position, and became deeply integrated into the global economy.

China evolved into a communist state with capitalist characteristics.

After more than twenty-five years of trade, infrastructure investment, technological development, and industrial expansion, China is now the second-largest economy in the world, potentially capable of surpassing the United States in the coming decades.

Some argue that we are now living through a second Cold War.

It is no coincidence that Xi Jinping referenced the Thucydides Trap during discussions with American leadership.

Perhaps it is still too early to understand the full consequences of the French Revolution.

Parallels Between the French Revolution and the Revolution of Artificial Intelligence

The Enlightenment taught humanity that reason could redesign the world.

Artificial intelligence may become the moment humanity discovers the limits of that belief.

And perhaps, once again, humanity is opening a revolution whose consequences it still cannot comprehend.

Algorithms already influence what we read, what we believe, who receives credit, how insurance is priced, how capital is allocated, and increasingly how institutions evaluate human behavior itself.

AI is no longer merely a productivity tool. It is gradually becoming an invisible layer of institutional decision-making.

Burke’s central warning was not against progress. It was against the belief that accumulated human wisdom could be replaced by any single system of rational design, however sophisticated.

AI may be the first system powerful enough to make that belief seem plausible

In that sense, artificial intelligence may represent something far larger than a technological revolution: a new attempt to reorganize society through rational systems operating beyond ordinary human comprehension.

Christopher Olah, co-founder of Anthropic, recently sat before Pope Leo and a room of cardinals at the Vatican and told them that his team keeps finding “mysterious, even unsettling” things inside their own AI models.

They found things that were not programmed. They emerged on their own.

Olah closed his speech with a request: “We need more of the world — religious communities, civil society, scholars, governments — to take this seriously, to look closely, and to push events in a better direction. We need informed critics who will tell the labs when we are failing. We need moral voices that the incentives cannot bend.”

The man building it is telling us he does not fully understand what he built. And he is asking a two-thousand-year-old institution for help figuring it out.

Prometheus has already stolen the fire. And he is asking us what to do with it.

Subscribe now

This essay became much longer than I originally expected. For that reason, I will leave here only the questions I intend to explore in the next part of this series:

1. Burke argued that it is impossible to prescribe remedies for institutions without first exposing their weaknesses. What remedies will AI prescribe for society?

2. Rousseau’s volonté générale contained an inherent flaw: who determines what the common good actually is? The people themselves are subjective and contradictory. Now AI may begin defining what is right and wrong.

3. One of Burke’s central criticisms was that only extreme arrogance could assume a nation is a blank sheet of paper waiting to be redesigned. Will AI inherit this same presumption?

4. If AI can perform nearly every human function better than humans themselves, could governing eventually become one of them?

5. Institutions exist to mitigate human error, bias, and individual limitations. They are controllable, procedural, and continuous. Institutions are condensed human rationality and accumulated historical wisdom. AI may become something even more condensed and potentially limitless. Will humanity finally believe it can coordinate society rationally?

6. Man may no longer be shaped through education, as Helvétius believed, but instead shaped according to whatever AI defines as good.

7. Will those who resist the rationalization of society through AI eventually be treated as enemies of progress? If so, what will the modern guillotine look like?

8. Is it better to continue living with human errors or risk entirely new forms of error created by artificial intelligence?

9. How much decision-making authority are we willing to delegate to AI? Today, algorithms already influence investment allocation, credit approval, insurance pricing, and countless other areas of modern life.

10. How will we question or audit the decisions of a superintelligent AI if we cannot even understand its reasoning process? Will we simply assume it is correct?

11. Human societies organize themselves hierarchically for many reasons, one of them being accountability. If AI commits a catastrophic error, who will ultimately be held responsible: the user, the algorithm, or the CEO behind the company?

12. If God is omnipotent, omniscient, and omnipresent, then AI may become the closest thing to a god humanity has ever created.

13. Totalitarian regimes historically failed, in part, because they lacked the capacity for total control. AI may become the first true instrument capable of enabling it.

    ---

Subscribe now

The election will be decided by the independent group, roughly 10% of the electorate.

Bolsonaristas and Lulistas are locked in. There is only one group changing its votes at the margin.

Among independent voters, Flávio is already leading: 33% vs 26%.

The trend is already clear.

To understand why, you need to understand how Brazilian elections are actually decided.

Here are the ten forces shaping the outcome:

1. Deeply entrenched political polarization in Brazil, leaving little room for a viable third path

2. Independent voters: who they are and what they want

3. 2022 vs. 2018: who actually changed their vote?

4. Lula didn’t win the election — Bolsonaro lost it. Why?

5. The left wing will not be able to credibly address public security and crime before the election

6. Corruption scandals and the erosion of democratic institutions

7. Bolsa Família is now perceived as a right, not a benefit — economic measures no longer boost popularity

8. The incumbent’s curse and exposure to opposition attacks via social media

9. An aging population and a shift toward more conservative agendas

10. Renan Santos as a wildcard outsider

Before diving in, one note on scenarios.

There is still a non-zero probability that Lula does not run, particularly due to health concerns. I also cannot rule out a potential “Biden moment” in a debate setting.

This analysis assumes the current baseline scenario and does not incorporate those possibilities.

For readers less familiar with Brazil, a brief overview helps to frame the dynamics.

Subscribe now

The Origins of Polarization

The story of this election does not begin in 2026. It begins with the rise of Jair Bolsonaro in 2018.

After three and a half terms of Workers’ Party (PT) rule, and one of the most severe economic crises in Brazil’s recent history — between 2015 and 2016 — trust in the traditional political establishment had collapsed.

On one hand, the perceived incompetence of the PT government during Dilma Rousseff’s presidency left voters disillusioned with left-wing ideology. On the other, the traditional right wing — represented by Geraldo Alckmin of the PSDB (now Brazil’s vice president) — failed to connect with voters.

A political vacuum emerged.

Into that vacuum stepped Jair Bolsonaro: a controversial, combative candidate with conservative values and a somewhat abrasive style. But above all, a charismatic figure.

This combination — charisma and a direct attack on the political establishment — resonated with a frustrated electorate.

Bolsonaro emerged as an outsider and filled the void left by both the traditional left and right.

The 2018 Election

Bolsonaro began his campaign as an unlikely candidate — even a subject of ridicule in certain social circles — due both to his unconventional style and the perceived improbability of his election.

And that skepticism was not without reason. Bolsonaro had no notable political track record, lacked strong congressional backing, and had minimal campaign funding.

To illustrate: 12 days before the first-round vote, Bolsonaro had spent just R$1.1 million on his campaign. By comparison, Henrique Meirelles and Geraldo Alckmin had each spent approximately R$43 million.

In terms of TV exposure, Bolsonaro had just 8 seconds of airtime per block, out of a total of 12 and a half minutes. Alckmin, by contrast, had 5 minutes and 32 seconds per block.

So how did a candidate with no significant political base, limited resources, and almost no TV presence manage to win?

The 2018 election marked a structural break in Brazilian campaign dynamics. Television exposure and campaign funding became far less decisive than in any previous election.

Social media, especially WhatsApp, Instagram and Facebook, emerged as a new mass communication tool.

These platforms were decisive in Brazil’s 2018 presidential election.

With his controversial, combative, and charismatic style, Bolsonaro captured voters’ attention.

WhatsApp groups began sharing his videos widely, algorithms amplified engagement, and a clear trend started to form across social media.

Jair Bolsonaro went viral.

During the campaign, Bolsonaro was stabbed at a rally in Minas Gerais. Many political analysts argue that this moment was pivotal.

It helped create a narrative of Bolsonaro as a martyr, someone who would fight against evil and against the system.

His popularity surged even further.

By the eve of the runoff, Bolsonaro had become the clear favorite and went on to defeat Fernando Haddad by 55% to 45%, a 10 percentage point margin.

The 2022 Election

In 2022, Lula and Bolsonaro faced off in the most tightly contested presidential race since Brazil’s return to democracy.

The runoff result was 50.9% for Lula versus 49.1% for Bolsonaro, a difference of just 2.1 million votes.

In other words, Bolsonaro lost roughly 12 percentage points in his head-to-head against the PT between 2018 and 2022.

We’ll come back to that number later.

This razor-thin election reveals something unusual.

Bolsonaro became the first incumbent president to lose a reelection bid since the 1997 constitutional amendment that allowed presidential reelection in Brazil.

How does an incumbent, with the full machinery of the state at his disposal, lose an election in Brazil?

Subscribe now

Polarization in Data and Across Voter Segments

Felipe Nunes, political analyst and CEO of Quaest, one of Brazil’s leading polling and political intelligence firms, frames it this way:

“Presidential candidates will be fighting over just 10% of the electorate.”

According to him, roughly 35% of voters are Lula-aligned or on the left, and another 35% are Bolsonaro-aligned or on the right. Of the remaining 30%, about 20% are disengaged voters who do not turn out.

One way to observe this polarization is through Lula’s approval ratings segmented by political alignment.

I’ve followed this data for a long time, and the pattern is clear. No matter what Lula does, those on the left tend to approve of him, while those on the right tend to disapprove.

In practice, the only group that shows meaningful variation in approval is the independents.

This is what matters: the election is decided by the remaining 10%.

Who Are the Independent Voters?

“The voters within this 10% segment are not very ideological. They are pragmatic. They care about public safety, well-being, and their children’s future. They are focused on non-ideological issues and are therefore difficult to win over. Regardless of which candidate emerges, the race will be tight.” - Felipe Nunes

Although this data from Quaest is recent, voter polarization was already in place in 2022. If anything, it has only become more entrenched.

And in 2022, Lula managed to win over this independent electorate.

Lula won the 10%.

Who Were the Voters Who Switched in 2022?

Between his victory in 2018 and his defeat in 2022, Bolsonaro lost 12 percentage points.

Who are these voters?

In one of its investor letters, Mar Asset Management breaks down the main shifts in vote counts between the 2018 and 2022 elections.

From a regional perspective, the state that contributed the most to Lula’s victory was São Paulo, with an incremental gain of 5.3 million votes for the PT.

Rio de Janeiro and Minas Gerais also played a relevant role, each adding roughly 900 thousand incremental votes for the PT.

In a second layer of analysis, Mar Asset broke down incremental votes by municipality and income level.

In other words, when comparing the last two elections, the marginal voter that elected Lula came from middle- and higher-income groups in the Southeast.

The common narrative among many Brazilians, that the Northeast elected Lula, is simply not accurate.

The Northeast represents the core 35% base referenced by Felipe Nunes, but the largest increase in votes for the PT in 2022 came from the Southeast.

This is the margin that decides elections. And this is where the shift happened in 2022.

Why Did Bolsonaro Lose the 2022 Election?

Lula didn’t win the election. Bolsonaro lost it.

Given Brazil’s cultural context, conventional wisdom would point to three main factors: (i) the economy, (ii) corruption, and (iii) social programs.

But once again, conventional wisdom is wrong.

According to academic research by Rafaella Jaeger and Lucio Rennó¹, the government’s handling of the pandemic was more decisive than economic perceptions.

Bolsonaro downplayed the severity of the health crisis. This generated deep resentment among those directly affected, especially people who lost family members during the pandemic.

The table below illustrates the relationship between the rise in government disapproval and the increase in Covid-19 deaths.

The “Letter for Democracy”

The second decisive factor, and one that goes against conventional thinking in Brazil, was the perception of a weakening of the country’s democratic institutions.

The main source of erosion came from repeated attempts to change the voting system to printed ballots, along with broader efforts to cast doubt on the reliability of electronic voting.

This ties back to the earlier point. Who, in practice, is more sensitive to threats against democratic institutions and the risk of an authoritarian shift?

The Southeast. Middle- and upper-income voters.

It is no coincidence that the “Letter for Democracy” gained so much traction. Prominent business leaders, executives from Brazil’s largest companies, and well-known artists publicly endorsed it.

The “Letter for Democracy” became a focal point of this reaction.

Rennó¹ argues that as polarization increases, ideological preferences and political positioning become more important than purely economic evaluations.

In that context, the management of the pandemic and the perceived threat to democratic institutions were the decisive factors behind the shift in independent voters.

The 10% ultimately drove Bolsonaro’s defeat in 2022. And in Brazil, the margin decides the election.

Subscribe now

What Are Brazilians Most Concerned About Today?

Lula’s approval ratings are currently very low, and there are clear reasons behind that.

According to Quaest, when Brazilians are asked about their biggest concerns, violence ranks first by a meaningful margin, followed by corruption.

These are two areas where the Brazilian left has historically struggled, and within a six-month window, it will be very difficult to shift voter perception.

On one hand, even if new public security measures are introduced in the coming months, it is unlikely they will produce tangible results in time to influence how voters feel before the election.

On the other hand, corruption scandals such as the Master case and the issues involving INSS retirees are unlikely to fade from public memory before October. If anything, they may intensify, especially with the possibility of a plea bargain involving Daniel Vorcaro.

Research by Lucio Rennó and Rafaella Jaeger already pointed out that in a highly polarized environment, economic issues tend to lose relevance, while ideological narratives take center stage.

Corruption and public security are likely to dominate the next election.

The Economic Factor in the 2026 Election

The current government finds itself in a difficult and paradoxical position.

The government is cornered.

Even with unemployment near historic lows, inflation relatively under control, and GDP growth, Brazilians do not feel that there has been real economic prosperity under the PT government.

Felipe Nunes outlines the main reasons:

1. Household debt is high, and debt servicing is consuming disposable income

2. The price level remains elevated. Lula promised “picanha and beer” during the campaign, but voters have not felt an increase in purchasing power

3. A portion of income is being diverted to betting and gambling

In a second explanation, Felipe highlights how Bolsa Família affects voter perception:

1. Brazilians now see Bolsa Família as a right, not as a government benefit

2. Around 70% of Brazilians no longer fear losing the benefit — especially after Bolsonaro increased its value during his own administration

3. Expanding social programs no longer translates into higher approval ratings

In a talk I attended with Felipe, he added two more points:

1. There has been no perceived improvement in economic status

2. In the past, Lula’s pitch was that under his government the poor could buy a home, own a car, and fly for the first time

Today, that sense of upward mobility is no longer there.

A clear example is the income tax exemption up to R$5,000, which has failed to boost the government’s popularity.

With the election approaching, it is difficult to see what additional economic measures could meaningfully change how Brazilians perceive their economic reality.

Beyond the domestic picture, external shocks — whether from geopolitical tensions or commodity price swings — could further erode economic sentiment before October.

If the economy is not winning the independents, Lula has no buffer.

Without economic upside, the margin becomes vulnerable.

The Incumbent’s Burden

Felipe Nunes also raises what he calls the “incumbent’s curse.”

The difficulty faced by any sitting government, even on a global scale, in keeping its voters satisfied.

Brazilian voters, he argues, behave almost like teenagers.

Never satisfied. Always complaining.

Social media makes this dynamic worse.

Conflict drives engagement. Engagement is rewarded by algorithms. The most polarizing content spreads faster.

Algorithms decide what gets amplified and what gets ignored.

And who is most exposed to political attacks?

The incumbent.

Nikolas Ferreira is the clearest example. Every government misstep becomes an attack video.

Several of them have gone viral.

This dynamic is unlikely to change before the election.

If anything, it will intensify.

The sitting government will continue to be the primary target.

And the more exposed the incumbent is, the more pressure on the margin.

The Erosion of Democratic Institutions

The “Letter for Democracy” was a decisive factor in the 2022 election.

Defending democracy became one of the key narratives that helped swing the 10% of undecided voters toward Lula.

Lula even adopted it as part of his campaign messaging.

Today, however, there is a growing perception of corruption within the Supreme Court itself.

This follows reports involving a R$129 million contract linked to Alexandre de Moraes’ wife, as well as Toffoli’s association with the Tayayá resort.

According to the most recent survey by Futura Inteligência, more than 55% of respondents support the impeachment of Supreme Court justices.

Public outrage over potential corruption in the Supreme Court is so strong that it cuts across the traditional left-right divide.

Even so, the easiest association tends to fall on the current government.

Once again, this becomes an attack on the incumbent.

Voters understand that the likelihood of a Supreme Court minister being impeached is significantly higher under a right-wing government than under a PT administration.

More recently, videos from Romeu Zema, in which he refers to “untouchables,” have been gaining traction.

More content along these lines is likely to go viral as the election approaches.

This time, the defense of democracy may have the opposite effect.

The same voters from the Southeast, middle- and upper-income groups, who were decisive at the margin in 2022, are likely to revisit this issue again.

These are exactly the issues that move independent voters.

And this is where Lula starts losing the 10%.

An Aging Population

Brazil’s demographic pyramid is reversing.

The population is getting older.

Studies suggest that as populations age, voters tend to shift toward right-wing or more conservative parties.

“As individuals grow older, they tend to become less open to new experiences and ideas, and show a preference for products and solutions, including political ones, that have already been tested.”

Brazil is no exception.

The aging trend shown in the data below is likely to continue favoring more conservative agendas.

Another factor working against the incumbent.

The Wildcard Outlier

Renan Santos is the wildcard.

He could turn into a new Pablo Marçal, rapidly gaining visibility through viral content but ultimately fragmenting the right-wing vote.

Or he could choose to direct his base toward a right-wing candidate.

So why does Renan matter?

Start with a simple question: who are his voters?

Renan’s voters are those who are tired of polarization. They are the voters who cast blank or null ballots, or who abstained in the last election.

The table below breaks down voting intentions for 2026 based on how voters behaved in 2022.

The key categories to watch are “blank/null vote” and “did not vote.”

Renan Santos leads both categories, with over 30% support.

There is an important distinction between these two groups.

Blank and null votes represent a form of protest. These voters do not identify with any candidate and choose not to take a side.

Abstention, on the other hand, is mostly driven by constraints. Many voters simply cannot make it to the polls due to financial, logistical, or family reasons. For example, mothers who need to care for their children on election day.

The 2022 election was decided by a margin of 1.8 percentage points — roughly 2.1 million votes. The next election is also likely to be close.These two groups can make the difference.

These two groups can make the difference.

Given their size and the tightness of the race, the candidate who is able to mobilize this segment of the population will have an advantage.

As the main right-wing candidate in the race, Flávio is better positioned to absorb this voter base.

Renan Santos doesn’t need to win. He just needs to move the margin.

Renan and Younger Voters

The founder of Movimento Brasil Livre has positioned himself as the “true right.”

He promises to fight corruption without compromise and refuses to bargain with the centrist bloc.

No concessions.

Renan rejects both Lula and the Bolsonaro family.

His voter base is digital. He reaches them through TikTok, Instagram, and YouTube.

According to CNN Brasil, Renan is particularly effective with younger voters. Among those aged 16 to 24, he reaches 24.7% of voting intentions in the March 2026 AtlasIntel poll, compared to just 4.4% in the overall electorate. In other words, his support among young voters is almost six times higher than the average.

Renan may become a wildcard outsider in this race.

Even as an anti-establishment figure, he influences younger voters through right-leaning narratives.

In a tight election, younger voters can make the difference.

Why Lula Will Lose the Election

Here I return to the ten key points of this article:

1. Deeply entrenched political polarization in Brazil, leaving little room for a viable third path

2. Independent voters: who they are and what they want

3. 2022 vs. 2018: who actually changed their vote?

4. Lula didn’t win the election — Bolsonaro lost it. Why?

5. The left wing will not be able to credibly address public security and crime before the election

6. Corruption scandals and the erosion of democratic institutions

7. Bolsa Família is now perceived as a right, not a benefit — economic measures no longer boost popularity

8. The incumbent’s curse and exposure to opposition attacks via social media

9. An aging population and a shift toward more conservative agendas

10. Renan Santos as a wildcard outsider

Dissatisfaction with the government is not only reflected in Lula’s approval ratings.

Independent voters have already started to position themselves.

A large share of independents say they may not vote.

Renan Santos and other anti-establishment figures can influence this behavior.

When asked what concerns them more, (i) Lula’s reelection or (ii) Flávio’s election,
those who chose “blank/null” or “did not vote” show less concern about a Flávio victory.

On the other hand, Flávio, as a candidate, faces the challenge of convincing independents that he is more moderate than his father.

53% of independents believe that Flávio is not more moderate than Jair Bolsonaro.

However, among those who do intend to vote, Flávio is already leading Lula by seven percentage points.

In line with his strength among independents, I believe Flávio is likely to win the runoff.

Polls are already beginning to show Flávio in the lead.

Conclusion

This will be a close election. There is no doubt about that.

Brazil is deeply polarized. Most voters have already chosen their side. What remains is a small group of independents that will decide the outcome.

Roughly 10% of the electorate.

That is where elections are won and lost.

Lula is entering this race as the incumbent. Exposed, worn out, and under constant pressure. The issues that matter most to independent voters, public security and corruption, are exactly where his government is weakest.

Economic improvements are not translating into perceived gains in quality of life. Social programs are no longer moving approval. And the digital environment continues to amplify every mistake.

At the same time, alternative figures, from Flávio to anti-establishment voices like Renan Santos, are competing for the same marginal voter.

In a tight election, small shifts are enough.

And those shifts are already happening.

This election will be close. But the forces shaping the outcome are already in motion.

Subscribe now

1. Pandemia de Covid-19: um choque exógeno no ciclo eleitoral de 2022, Rafaella Jaeger; Bolsonarismo e as eleições de 2022, Lucio Rennó

These are my bets on Polymarket. They are not investments and certainly not recommendations.

I decided to put money behind my convictions — which, in theory, sounds like a great idea.

The plan was simple: bet that the United States would attack Iran while the market was still largely ignoring that scenario.

Two days later, the attack happened.

The only problem is that I wasn’t positioned yet.

The decision to build a portfolio of bets came from a desire to translate my macro and geopolitical views into something tangible — to move beyond theory.

The catalyst was my conviction, in late February, that the U.S. would strike Iran despite markets not pricing this in yet. Credit here goes to Charlie Garcia and his piece “America Is 30 Days From War With Iran,” which influenced my thinking.

The thesis was there. The timing, almost. By the time I managed to set up my Polymarket account and move the funds, the strike had already happened.

Polymarket turned out to be an interesting vehicle for two reasons:

1. It allows you to bet directly on events, not just on their second-order effects. Instead of buying oil or Petrobras, I can bet directly on a U.S. strike on Iran.

2. The “price” is simply the implied probability of the event. In some cases, even if I believe the event will happen, I’m not willing to pay the market’s price.

I’ll continue to share how these positions evolve over time — both the right calls and the wrong ones.

If you want to follow the thinking behind them, feel free to subscribe.

Subscribe now

Open Bets

Brazilian election

My first position was a basket of the main candidates, with a heavier weight on Flávio.

My initial purchase was at 36 cents in late February. It’s now at 43 cents, implying roughly a 19% return.

I used other candidates as hedges, particularly in scenarios where Lula or Flávio drop out. For example, I bought Haddad at 3 cents — now at 6.4 cents, up 113% since my initial entry.

I’ve been adding to the position over time, so Polymarket reflects an average price.

Thesis

In my view:

• Flávio has been gaining traction in the polls

• He has a credible path to victory

• After the first round, probabilities tend to reprice

This creates room for upside.

At the same time, Lula appears increasingly weakened:

• Scandals (INSS, Master)

• Inflation and economic pressure

• Public security concerns

On top of that, a potential escalation with Iran could push oil prices higher, fuel inflation, and reduce room for rate cuts by the central bank.

Source: Atlasintel, Bloomberg

Source: Atlasintel, Bloomberg

Electoral framework

The 2022 election was highly polarized. In my view, it reflected four groups:

1. Pro-Lula

2. Pro-Bolsonaro

3. Anti-Lula

4. Anti-Bolsonaro

With a tight margin (51% vs 49%), small shifts can change the outcome.

The most fluid group, in my opinion, is the anti-Bolsonaro vote — both due to current government fatigue and the perception that Flávio is more palatable than his father.

Haddad serves as a hedge, particularly given Lula’s age and increasing signs of fatigue — a potential Brazilian “Biden moment.”

Subscribe now

Iran War

Over the past months, I’ve spent significant time studying this topic. Key references include:

• Charlie Garcia

• Professor Jian (Game Theory)

• Robert Pape

• Melanie Phillips

Thesis

I struggle to envision an end to this war without a normalization of shipping flows through the Strait of Hormuz.

Without that, the world would face shortages of critical goods — especially oil and fertilizers — triggering supply chain disruptions and, inevitably, global inflation.

It also seems increasingly clear that the U.S. cannot guarantee safe passage through the Strait, even with President Donald Trump urging crews to “show some guts.”

Iran, meanwhile, maintains tactical control through mines and drones, effectively dictating the pace of any reopening.

There are reports that Iran has charged up to $2 million per transit. With roughly 100 ships per day before the conflict, this implies billions in potential funding to the war — a scenario that, frankly, does not seem sustainable.

At this point, I find it difficult to construct an outcome that does not involve escalation by the United States.

The math doesn’t work:

Who exactly is negotiating on behalf of Iran?

So far, the U.S. has not clarified who they are dealing with — and with much of the leadership eliminated, there may not even be a centralized decision-maker at this point.

The U.S. 15-point plan

The U.S. proposal is, in practice, unacceptable for Iran:

• Dismantling the nuclear program

• Ending uranium enrichment

• Restricting missile capabilities

• Reducing proxy support

• Giving up control over Hormuz

• Accepting long-term monitoring

Iran’s response

Equally unacceptable for the U.S.:

• Guarantees of no future U.S./Israel strikes

• War reparations

• Maintaining control over Hormuz

• Full sovereignty (no limits on missiles or nuclear autonomy)

There is no overlap here. There is no negotiation. There is only time until escalation.

Escalation risk

Countries don’t deploy warships — or Marines — without intent to escalate.

As Charlie Garcia puts it:

“The administration says it is still open to diplomacy. Of course it is. Nothing says ‘let’s talk’ like 60,000 troops, a floating airfield, and a division of paratroopers whose entire job description is jumping out of airplanes into places where they are not welcome.”

This isn’t diplomacy. It’s leverage.

Melanie Phillips’ view

Her argument is that the West fundamentally misunderstands the religious dimension of Shia Islam — particularly the role of martyrdom.

In her view, the Iranian regime is not trying to end the war — it genuinely believes it is winning.

The uncomfortable implication: we are dealing with an adversary that does not respond to the same incentives.

Iran does not need to win militarily. It only needs to make the global cost unbearable.

It does not adhere to conventional rules of war

Human cost is not a constraint — it can be part of the strategy

This changes the game entirely.

Iran’s strategy

Iran knows it cannot win a conventional war.

Its path to victory is economic disruption.

One of the most critical scenarios would be expanding the conflict to key routes — such as the Red Sea — creating a severe shock to oil markets and global supply chains.

This is not a tail risk. It is part of the strategic calculus.

With that in mind, most of my positions reflect an expectation of escalation.

Subscribe now

Closed Bets

I bet that the S&P 500 would decline by the end of March — driven by the war, potential stress in private credit, and elevated AI valuations. Topics for another post.

I put $10 on the trade four days ago. In three days, it was up ~30%. I closed the position.

The problem is that, as I write this, the S&P 500 kept falling — and I would have doubled the return.

In the end, the trader is always dissatisfied: either wrong, or underexposed.

I made a few smaller bets that are not worth detailing now.

These positions will either age well — or not.
I’ll be sharing both. Subscribe if you want to follow the process.

Subscribe now

From now on, I’ll be writing my Substack posts in English. Many of the books and sources that inspire these essays were originally read in Portuguese, so some translations may occasionally feel imperfect.

I hope readers will forgive any linguistic imperfections. The goal here is to explore ideas, patterns, markets, and economic history.

Now, to the essay.

We may be living in the most self-conscious era about happiness in history — and one of the most persistently dissatisfied. In a world of social media, where other people’s lives constantly seem better than our own, the sense of lack has become permanent.

With that in mind, I decided to revisit two books that approach happiness from distinct philosophical perspectives: The Symposium, by Plato, and Happiness, Desperately, by André Comte-Sponville¹.

The goal here is not to criticize their core ideas — after all, who am I to criticize Plato or Comte-Sponville?

My objective is to raise modern questions using reflections from the past that have endured over time.

Throughout this essay, I aim to address two questions:

i) Have social media and the seemingly perfect lives of others increased our perception of lack?

ii) Have social media raised our expectations of ourselves, widening the gap between expectation and reality — and, consequently, increasing our unhappiness?

I found the following structure to be the most effective way to organize a discussion on happiness:

1. The Platonic view of happiness

2. Comte-Sponville’s critique of the Platonic view

3. Reflections on the modern world

Subscribe now

The Platonic View of Happiness

It is impossible to determine how much Plato shaped Western thought and society. Somewhere between enormous influence and total influence. So we must begin with Plato.

What is love for Plato?

Love is the desire for the Beautiful and the Good. Since we desire what we do not have, desire arises from lack. Love is not fullness — it is lack that creates movement².

Every time we love what we lack, we are within Plato’s framework.

The problem is that once we finally obtain what we desired, we possess the object — and no longer desire it. If happiness is not having what we once desired, but rather having what we currently desire, then this condition can never be fully realized.

It is no coincidence that the question of happiness has troubled us for so long. Montaigne wrote: “there is no science more difficult than that of living this life well and naturally” — it is an art, and a learning process.

Aragon wrote, “by the time we learn how to live, it is already too late.”

If we are Platonic — if we desire what we lack — how can we ever expect to be happy, when what we lack is precisely what we desire?

Among the more pessimistic thinkers: Schopenhauer argued that life is an endless pendulum between the struggle to obtain and the boredom of possession. Camus wrote that "men die, and they are not happy." Sartre claimed that "pleasure is the death and failure of desire" — because desire perishes in the moment of its satisfaction.

Lack and Expectation

Perhaps never in history have we been so aware of what we lack. When we open Instagram or LinkedIn, we are constantly exposed to ideals — the perfect trip, the long-awaited promotion, record company growth, among others.

Lately, I’ve noticed a new kind of ideal: “look how I used ChatGPT or Claude to automate 80% of my work, and now my productivity has increased tenfold — do this or fall behind.” But I digress.

Remarkably, we are capable of envying two completely different scenarios at the same time: the perfect vacation and the relentless pursuit of a higher goal. There seems to be no limit to our perception of lack.

The issue is that our happiness appears to depend on the relationship between expectation and reality. We only expect what we do not have — and for that very reason, the more we expect to be happy, the less happy we tend to be.

It is paradoxical.

We are constantly separated from happiness by the very hope that promises it.

I believe social media increases our unhappiness through a double effect: by amplifying both our perception of lack and our expectations of ourselves relative to reality.

Comte-Sponville’s Critique of the Platonic View

Fortunately, Comte-Sponville offers a more optimistic perspective. For him, life is not as unhappy as Plato suggests.

They underestimated two small things: pleasure and joy.

There is pleasure and joy when we desire what we already have — what is not lacking.

Sometimes we are happy because we do what we desire and desire what we do. Unhappy is the traveler who expects happiness only upon arrival.

When we desire what we have, we no longer wait for anything. The present is enough.

For Comte-Sponville, this is what he calls happiness in act — a happiness that expects nothing, a happiness without hope.

He distinguishes three types of desire: hope, love, and will.

Hope

To hope is to desire without enjoying, without knowing, and without being able to act on it.

A simple example: “I hope your surgery went well.”

Even though the event is in the past, you do not know the outcome, you do not experience it directly, and it does not depend on you. Hope does not always regard the future.

The strength of hope tends to be inversely proportional to its probability. The more one hopes, the less likely it is to occur. And there is no hope without fear — the greater the hope, the greater the fear of not achieving it. They are not opposites: they are two sides of the same coin.

The opposite of hope, then, is not fear, but happiness in act.

Happiness in Act

Happiness is not impossible: we can desire what we enjoy, what we know, and what we can act upon.

That means pleasure, knowledge, and will.

This is happiness in act — and it can only exist in the present.

For Comte-Sponville, beatitude is the happiness of someone who no longer expects anything.

As Jules Renard wrote:
“I desire nothing from the past. I no longer count on the future. The present is enough.”

Love

Comte-Sponville adopts Baruch Spinoza’s definition of love: joy accompanied by the idea of its cause.

Joy is rejoicing in what is, rather than suffering over what is not.

To love is to experience joy, rather than to hope or to fear.

Will

Will is the form of desire that depends only on us.

If will depends only on us, then when we truly want, we act.

To will is to act.

Conclusion of Comte-Sponville

Comte-Sponville draws from Spinoza:
“Desire is the very essence of man. Desire is not lack — it is power: the power to exist, to act, to enjoy, and to rejoice.”

Spinoza vs. Plato:

In Plato’s framework, we are led to reduce our perception of lack — something like suppressing appetite in the face of hunger.

For Spinoza, appetite is the human conatus — the power of being — as it relates to both body and mind.

Life, for Spinoza, is not about diminishing our power, but exercising it: eating with appetite, acting with appetite, loving with appetite.

The depressed and the anorexic have not lost lack — they have lost the power to enjoy.

To conclude this section, Comte-Sponville proposes what he calls the conversion of desire:

Instead of desiring what we lack, we should learn to desire what depends on us and what already is.

That is action and love.

That is happiness without hope — desperate happiness

Reflections on the Modern World

I believe social media has significantly deepened our perception of lack. We are rarely truly satisfied — not that this is a new problem, but we may now be experiencing it with greater intensity than ever before.

At the same time, social media has raised our expectations.

Since happiness depends on the relationship between expectation and reality, perhaps expectations have never been higher.

In the past, we compared ourselves to a small social circle.

Today, we compare ourselves to the entire world — simultaneously.

For younger generations in particular, when goals are not achieved, the disappointment is compounded by a modern belief — often unrealistic — that we can achieve anything. The conclusion that follows is quietly corrosive: that the problem must be us.

Bringing these reflections closer to my own field, I wonder how this affects the economy, markets, and human behavior.

Beyond increasing unhappiness, it seems that this heightened perception of lack — combined with the need to project a perfect life — drives higher consumption.

It is the old pattern: spending money you don’t have to impress people you don’t know or don’t even like.

Perhaps social media has only amplified a much older problem: human desire shaped by lack.

In a world that constantly reminds us of what we do not have, perhaps true freedom lies in learning to desire what we already possess.

Subscribe now

1. The title in French, "Le bonheur, désespérément," literally means "happiness without hope" — a play on désespoir (despair/hopelessness). The philosophical precision of the original is worth noting: Comte-Sponville's happiness is "desperate" not in the sense of urgency, but in the sense of being liberated from hope.

2. This is a simplified reading of Plato, sufficient for this essay. I chose not to explore the nature of Beauty itself or the ladder of love.

That flow has reinforced market concentration, strengthened a narrow group of companies, and, more recently, been channeled into a single dominant narrative: artificial intelligence.

But when capital becomes concentrated, so does risk.

In a market shaped by passive flows and mechanical allocation, a reversal in that dynamic could have consequences that are far less diversified than they appear.

What is often perceived as diversification may, in reality, be a highly correlated system dependent on continuous inflows.

In this analysis, I will address the following topics:

1. The concentration of the ten largest companies in the U.S. market

2. The concentration of wealth and consumption in the American economy

3. The rise of passive investing through ETFs and its effects on markets

4. Why all of this matters

1. The concentration of the ten largest U.S. companies

According to Goldman Sachs, the ten largest companies in the S&P 500, which represent roughly 2% of the companies in the index, account for approximately 40% of its total market capitalization and around 30% of its cumulative earnings.

Among these ten companies, eight are directly linked to the AI ecosystem¹, whether through language models, data centers, semiconductor chips, humanoid robotics, or related infrastructure.

Source: Aquavis

To put this into perspective, the ten largest American companies together represent roughly $21 to $24 trillion in market value, approximately two-thirds of U.S. GDP.

That is correct: ten companies represent about two-thirds of the U.S. economy in terms of market capitalization.

If we consider that global GDP is roughly $110 trillion, these companies represent close to 20% of the entire global economy.

From the perspective of economic concentration, the U.S. economy begins to resemble a large oligopoly whose products and services permeate nearly every sector of economic activity.

Since the profits of these firms represent a meaningful share of economic activity, both directly and indirectly, most companies and individuals end up purchasing services connected to these large technology platforms.

Another way to assess the economy’s dependence on AI is to observe the magnitude of investment commitments made by major technology companies. As shown in the chart below, AI investment relative to U.S. GDP already exceeds many historic infrastructure programs.

Source: JP Morgan Asset Management

As I argued in my previous post, the central question is not whether AI will change the world.

The real question is whether all of these companies can win the race simultaneously — and if they cannot, who will pay the cost of the losers.

Subscribe now

2. Wealth and consumption concentration in the U.S.

Three dimensions of concentration are relevant here:

2.1) Concentration of equity ownership across wealth groups
2.2) Concentration of equities within household wealth portfolios
2.3) Concentration of consumption among the wealthiest households

2.1) Concentration of equity ownership

Large U.S. corporations have experienced significant gains in operating leverage over the past decades. In other words, many companies have grown substantially while keeping headcount and operational costs relatively stable.

IBM employed roughly 400,000 workers in 1985 and was then the largest American company. Today, Nvidia employs around 40,000 people, carries a market capitalization roughly twenty times larger than IBM had at the time, and generates five times more profit with one-tenth of the workforce.

The chart below shows the division between labor and capital contributions to corporate profits. It clearly illustrates the long-term decline in labor’s share.

Source: Aquavis

The main consequence of this increase in operating leverage has been a steady expansion in corporate profit margins.

Source: Goldman Sachs, Investment Strategy Group

Why does this matter?

Because the reduction in labor’s share of profits means that the expansion in margins was not passed on to workers. Instead, it accrued largely to capital owners: shareholders, senior executives, and CEOs.

The chart below illustrates the growing gap between CEO compensation and that of average employees.

Source: Aquavis

All of these forces contributed to rising equity ownership concentration in the United States — now at its highest level since 1980.

The wealthiest 10% of households hold approximately 87% of the total value of the U.S. stock market.

Source: Aquavis

2.2) Concentration of equities within household wealth

How much of the average American household’s financial wealth is exposed to equities?

According to the Federal Reserve, more than 45% of household financial savings are invested in stocks.

In other words:

i) Most equity market value is concentrated among the wealthiest households.

ii) Within those households, a large portion of wealth is also concentrated in equities.

This creates a double concentration: wealth is concentrated at the top, and that wealth is itself concentrated in equities.

Source: Federal Reserve and Kobeissi letter

This raises a natural question.

What happens if the wealthiest households face financial stress within their major equity holdings and suddenly require liquidity?

Where would those withdrawals come from?

2.3) Concentration of consumption

The third dimension of concentration concerns consumption.

Not only is wealth concentrated, and not only is that wealth heavily exposed to equities. Consumption itself is also concentrated.

According to the Financial Times, the wealthiest 10% of households account for roughly 50% of total U.S. consumption.

Source: Financial Times

Why is consumption so concentrated?

Real wages have not increased proportionally with corporate profit growth. Inflation has eroded purchasing power for lower-income households, compressing both consumption and savings.

A significant portion of this dynamic was intensified by the post-COVID monetary expansion conducted by the Federal Reserve. That expansion contributed to higher asset prices and higher inflation, though that subject deserves its own discussion and another post.

The result is a K-shaped economy. For affluent households, conditions appear healthy: wealth appreciates, consumption holds. For lower-income households, purchasing power continues to deteriorate.

Three threads converge:

i) Much of the consumption in this “K-shaped” economy is sustained by affluent households.

ii) Much of the broader economy is supported by AI-related investment.

iii) Much of affluent consumer confidence derives from financial wealth — concentrated in equities, which are themselves concentrated in AI-related companies.

If the economy, investment flows, and consumption are deeply interconnected, what happens if one part of the AI chain fails to perform as expected?

Subscribe now

3. Passive investing and the mechanics of market flows

Up to this point, the argument has focused on structural concentration in wealth, consumption, and corporate profits.

The next step is to examine how financial market mechanics may amplify these dynamics.

I will discuss four topics:

3.1) The United States as a global empire (Ray Dalio)
3.2) The rise of ETFs and passive management
3.3) Corporate buybacks reinforcing structural demand
3.4) Valuations and rising capital expenditure

3.1) The United States as a global empire

Ray Dalio argues that the United States emerged as the dominant global empire following World War II. In that context, the dollar became the world’s reserve currency and the Federal Reserve acquired what has often been called the “exorbitant privilege.”

This privilege allows the United States to issue large quantities of currency and government debt while still being perceived as a safe haven.

This is extremely important because it keeps interest rates structurally lower, reduces the cost of capital, and supports the development of American corporations².

Over several decades this privilege enabled the United States to absorb enormous volumes of capital from both domestic and foreign investors.

However, a new narrative is emerging in financial markets: the possibility of a shift in the global order.

This is where George Soros' concept of reflexivity becomes relevant: market participants' perceptions do not simply reflect reality — they also shape it. Two questions follow.

If market perceptions begin to shift and confidence in the United States as the dominant global power weakens, how would that affect financial markets?

Would we observe meaningful capital outflows from the United States, particularly from foreign investors?

3.2) The rise of ETFs

ETFs have become a dominant channel for capital flows, both domestic and international.

Unlike traditional active managers, ETFs do not consider valuation. They simply buy or sell securities according to the weights assigned in the benchmark index they replicate — a mechanical and indiscriminate process.

This creates a potential vulnerability.

If a significant outflow of capital from the United States were to occur, ETFs could produce indiscriminate selling pressure, regardless of underlying fundamentals.

Since 2019, technological advances, broader investment access, and cultural shifts have accelerated the rise of passive investing relative to active management.

Source: Aquavis

Rather than allocating capital to portfolio managers who select securities based on fundamental analysis, many investors now prefer ETFs that track major indices such as the S&P 500 or Nasdaq.

Because these vehicles buy securities according to index weights, the largest companies automatically receive the largest share of new inflows.

This mechanism can reinforce market concentration and create a snowball effect.

Has the rise of passive investing and its mechanical buying contributed to the increase in market concentration since 2019?

This brings us back to the key question:

How will markets behave in a period of meaningful selling pressure?

Will ETFs amplify the move?

3.3) Corporate buybacks

Large corporations have generated substantial cash flows thanks to the operating leverage described above.

Many have become asset-light businesses requiring limited capital to continue growing.

With this excess cash and increasing stock-based compensation programs, companies have aggressively repurchased their own shares.

Fonte: Aquavis

Buybacks reinforce structural demand for equities — a second mechanical buyer alongside ETFs, adding to the snowball effect.

But the AI investment wave is beginning to alter this dynamic.

3.4) Valuations and cash flow pressure

First, the market is not cheap.

Price-to-earnings multiples remain near the upper range observed over the past decade.

Source: Factset, February 2026

This matters because markets are not currently priced for pessimism. Risks discussed throughout this essay are not fully reflected in valuations.

There is considerable room for repricing.

Second, large technology companies are changing their financial profiles.

Driven by the race for AI dominance, capital expenditure is rising rapidly. In some cases companies are beginning to burn cash rather than generate it.

Source: Mufgamericas

Markets have already adjusted their projections for cash flow generation in 2026.

Source: Factset, Evercore ISI Research

If cash generation weakens, buybacks may decline, which would reduce structural demand for equities.

This raises another question.

Subscribe now

4. Why this matters

The system works as long as capital keeps flowing in.

ETFs buy, companies repurchase shares, household wealth rises, consumption holds up, and AI continues to attract investment. If this dynamic were the title of a film, it could be called:

“AI: The New Engine of America’s Exorbitant Privilege.”

However, all of these elements are part of the same mechanism. If flows reverse, the system can begin to operate in the opposite direction.

One could argue that this concentration simply reflects U.S. structural strength: superior productivity, stronger innovation, better capital allocation. In that view, concentration is efficiency, not fragility. That argument deserves respect.

The point is not to deny that strength, but to recognize that systemic vulnerability can emerge when capital flows become concentrated along a single vector.

In this section, I outline three scenarios based on the risks discussed so far:

4.1) Potential foreign capital outflows

4.2) Declines in wealth, consumption, employment, and economic activity

4.3) Amplified and indiscriminate selling flows

4.1 Potential Foreign Capital Outflows

The U.S. financial system has a high level of foreign participation. As shown in the chart below, foreign investors represent roughly 35% of total U.S. financial assets, including equities and debt.

In absolute terms, foreign investors hold approximately $70 trillion in U.S. assets, close to 70% of global GDP.

Source: Aquavis

This raises a fundamental question.

As perceptions of U.S. risk evolve and the narrative of a shifting global order gains traction, will the rest of the world continue allocating capital to the United States at the same pace?

Early 2026 has already provided partial signals. We have observed strong capital flows into emerging markets, reaching the highest levels in years. The Brazilian equity market, for example, has reached historical highs.

Source: Bloomberg

So far, however, this has not been driven by a meaningful outflow from the United States. Rather, it reflects a redistribution of marginal capital. The chart below illustrates that net flows into U.S. equities and debt have not yet turned negative.

Source: Bloomberg

But the relevant question is forward-looking.

How would markets react to an actual reduction in foreign capital allocation?

It is important to remember that just as ETF inflows are mechanical, outflows can be mechanical as well.

4.2) Wealth, Consumption, and Economic Feedback Loops

The “K-shaped” economy, AI investment, and financial markets are deeply interconnected.

What happens if one link in the AI chain fails? Or maybe in private credit?

What if some of these companies fail to deliver the expected returns?

Would we see a decline in household wealth, leading to lower consumption and, ultimately, weaker economic activity?

Lower-income households are already facing pressure from inflation and declining purchasing power. In the event of rising unemployment, potentially accelerated by AI-driven productivity gains, would these households be able to service their debt?

We are already observing signs of stress.

The U.S. savings rate has declined significantly and household savings are now near their lowest levels since 2008.

Source: Bloomberg

At the same time, credit indicators are deteriorating. Delinquencies in credit cards and other forms of consumer credit are rising.

There are also emerging risks in commercial real estate and private credit, though that is a subject for a separate discussion.

Source: Federal Reserve

4.3) Indiscriminate Selling and Market Dynamics

For many years, supported by multiple structural factors, including its role as the dominant global power, the United States attracted continuous capital inflows. This created a virtuous cycle and a positive snowball effect.

Based on the dynamics discussed throughout this essay, there is a risk that this process could reverse.

In a scenario of capital outflows, especially in a market increasingly dominated by passive investing, selling pressure may become both amplified and indiscriminate.

In other words, the beginning of a negative snowball effect.

Conclusion

Financial markets, wealth dynamics, and the broader economy have become deeply interconnected. In many ways, investing in the United States has become a large thematic bet on artificial intelligence.

If that bet proves correct, the returns could be extraordinary.

If it does not, what appears today as diversification may reveal itself as systemic concentration.

The central question remains the same as in my previous essay.

The question is not whether AI will change the world.

The real question is who will win this race and who will pay for the losers.

¹ Nvidia, Apple, Alphabet, Microsoft, Amazon, Meta, Broadcom, Tesla

² This was not the only factor: entrepreneurial culture, technology, institutions, rule of law, military capacity, and education all played fundamental roles.

Subscribe now

From now on, I’ll be writing my Substack posts in English. Many of the books and sources that inspire these essays were originally read in Portuguese, so some translations may occasionally feel imperfect.

I hope readers will forgive any linguistic imperfections. The goal here is to explore ideas, patterns, markets, and economic history.

Now, to the essay.

Political polarization rarely begins with policy disagreements.

More often, it begins with something deeper: the moral intuitions that shape how we perceive fairness, authority, loyalty, and justice.

When people possess different moral “taste buds,” political disagreements cease to be debates over policy and become collisions between competing moral worlds.

Understanding this dynamic is essential if we want to understand the rise of modern political polarization—and, ultimately, the global resurgence of populism.

This essay draws on the arguments of Jonathan Haidt and his book The Righteous Mind. In it, Haidt explores the origins and development of human morality and how it permeates virtually all of our decisions.

The book provides a framework for understanding the growing lack of empathy between conservatives and liberals — a dynamic that has contributed to the formation of increasingly polarized political identities and helped fuel the rise of populism across the world.

The goal of this essay is to increase our understanding of why people possess such different moral frameworks—and consequently such different perceptions of right and wrong—and how modern society and technology have intensified our tribal and identity-driven tendencies.

I have divided this post into four main sections:

1. Why more intelligent people are not necessarily more impartial (and may even be worse)

2. The main moral differences between conservatives and liberals

3. How algorithms transformed political differences into identity conflict

4. How rising inequality has deepened political polarization

However, before addressing these themes, it is necessary to first establish the conceptual foundations Haidt uses to support his arguments.

Subscribe now

The Elephant and the Rider

Haidt introduces a fundamental concept that supports many of his arguments throughout the book: the metaphor of the elephant and the rider.

In this analogy, the elephant represents our emotions, intuitions, and instincts, while the rider represents our moral reasoning.

Throughout the history of moral philosophy—from Plato to Kant—many thinkers defended pure reason as the highest form of moral guidance, suggesting that reason should govern human action.

Haidt argues the opposite.

In practice, the rider rarely leads. More often, it simply adjusts the path already chosen by the elephant — our intuitions and emotional responses.

Moral reasoning does not guide our behavior so much as it justifies it. It functions less like an impartial judge and more like a defense attorney: constructing arguments for decisions that were already made, emotionally and intuitively, before conscious thought ever intervened.

In other words, the judgment comes first. The reasoning comes later.

Moral “taste buds”

To support his argument, Haidt draws on ideas from David Hume and A Treatise of Human Nature.

According to Hume, morality is not abstract or purely rational. It arises from sentiment; much like our sensory perceptions of sweetness, bitterness, heat, or cold arise from individual experience.

Moral perceptions therefore should not be understood as a matter of greater or lesser understanding, but rather as different “flavors” of moral experience.

Just as virtues are culturally shaped — the virtues taught to children in a warrior society differ from those emphasized in an agricultural one — human morality is not a purely deductive logical system.

Instead, it resembles a plural moral palate.

Humans possess universal moral receptors, but each culture, society, and set of institutions develops its own moral cuisine.

Group thinking and tribal instinct

A third central argument in Haidt’s framework concerns our tribal instincts.

Human nature is partly selfish, but we also possess a strong tendency to cooperate within groups—especially when competing with other groups.

Haidt describes this dual nature as our ability to operate on multiple levels: we are selfish primates who also desire to belong to something greater than ourselves.

Metaphorically, we are 90% chimpanzee and 10% bee.

Humans are the world champions of cooperation beyond kinship. We build complex systems of formal and informal accountability that allow large groups of strangers to collaborate.

This tendency to form groups was one of the key ingredients behind the development of civilization and humanity’s ability to dominate the planet.

But there is a downside.

Morality both binds and blinds.

Within groups that share similar moral frameworks, morality strengthens cohesion. At the same time, it can blind us to the perspectives of groups with different moral “taste buds.”

If morality is intuitive and group-based, political conflict is no longer simply disagreement over policy. It becomes a clash between competing moral frameworks.

With this conceptual foundation established, we can now explore how these dynamics connect to modern political polarization and the rise of populism.

Subscribe now

Why smarter people are not more impartial

Do we seek truth or approval?

Although we often view ourselves as rational beings, moral reasoning rarely functions as a genuine search for truth.

Instead, it frequently serves to justify our actions, especially when others do not share our moral framework.

We form judgments quickly and are notoriously poor at seeking evidence that might challenge our initial conclusions.

Instead, we become obsessed with arguments that reinforce our intuitive position —the preference of the elephant— and we actively search for evidence that confirms our existing beliefs.

Our reasoning therefore operates primarily as a post-hoc defense of intuitions we already hold.

Our minds function less like impartial judges and more like defense attorneys—or even politicians gathering support for a predetermined position.

Haidt's provocation: individuals with higher IQs or greater education are not more likely to question their initial judgments.

Greater cognitive ability does not produce greater moral impartiality.

It simply means that such individuals can mobilize a larger number of arguments and intellectual tools to defend he positions they already hold.

The main moral differences between conservatives and liberals

Should governments do more to promote the common good and social justice, even if this requires limiting individual choices?

According to Haidt, if you answered yes, you are likely a liberal. If you answered no, you are likely a conservative or libertarian.

To explain these differences, Haidt introduces the concept of moral foundations.

Each foundation possesses both a positive and negative dimension:

1. Care / Harm

2. Fairness / Cheating

3. Loyalty / Betrayal

4. Authority / Subversion

5. Sanctity / Degradation

6. Liberty / Oppression

Connecting this framework to the concept of moral “taste buds,” there is no universal rational system of timeless moral principles.

The moral worldview that resonates with you depends on the specific configuration of your mind's moral receptors — shaped by genes, culture, and experience.

Morality is intuitive, contextual, and culturally situated.

And that means two people can look at the same political question and reach opposite conclusions — not because one is reasoning poorly, but because they are reasoning from entirely different moral foundations.

How algorithms transformed political disagreement into identity conflict

The two most disruptive and emotionally charged areas of human life are politics and religion. Because both express our underlying moral psychology.

As previously discussed, humans naturally form groups.

But we do not form groups randomly. We feel empathy toward those who share our moral framework.

What has changed in recent years is how these groups form.

With the rise of digital technology and highly personalized algorithms, social media increasingly traps individuals inside information environments populated almost entirely by people who already agree with them.

This accelerates confirmation bias and steadily blinds us to opposing perspectives.

Over time, confirmation bias hardens into identity politics — inflamed by populist narratives of "us versus them," "the rich versus the poor," competing visions of justice.

When social bonds weaken, political opposition stops being an adversary with different ideas and becomes a moral threat. Politics turns into tribal warfare.

Public debate shifts away from policy discussions toward moral narratives about what is legitimate, fair, and desirable.

The chart below illustrates the increasing ideological concentration within each political party in the United States, highlighting the growth of political polarization and its reflection in economic perceptions.

Source: Gallup

Source: Gallup

This raises an interesting question:

How can economic confidence differ so dramatically between groups living in the same economy?

If economic perception is no longer determined solely by objective data, but also by political identity, then economics stops being purely economic. It becomes moral.

Inequality and the intensification of polarization

Rising inequality is the accelerant.

The post-pandemic surge in productivity and operational scale — combined with aggressive monetary expansion by central banks and strong financial asset appreciation — has intensified wealth concentration in the United States.

According to the Federal Reserve, the wealthiest 1% now control more than 30% of total U.S. net worth.

Source: Federal Reserve

This dynamic is also reflected in consumption patterns: the top 10% of households account for roughly half of total consumption in the country.

Source: Financial Times

We observe the same pattern in the chart below: the stock market surged after 2020—benefiting those who already owned financial assets—while consumer confidence continued to deteriorate, partly due to the erosion of purchasing power caused by inflation.

As a result, we introduce another dimension alongside moral perception: economic perception.

Consumer sentiment varies significantly depending on income level.

Rising inequality intensifies feelings of injustice, activating moral foundations that strengthen group identity. Ultimately fueling populism.

Populist movements thrive precisely because they offer simple moral narratives that resonate with deep emotional foundations and reinforce collective identity.

Source: Refinity DataStream

Conclusion

How can social cohesion be strengthened in an increasingly identity-driven world?

We are living through a period of intense polarization, in which individuals are so emotionally attached to their confirmation biases that finding common ground in political debate has become increasingly difficult.

Political debates no longer revolve around public policy alone. Instead, they increasingly revolve around competing moral narratives.

On one side, the hyper-personalization of algorithms deepens informational bubbles. On the other, rising inequality intensifies economic resentment.

Together, these forces reinforce identity politics and polarization.

As discussed in my previous essay on Hayek and American state capitalism, these dynamics also contribute to the rise of populism and the weakening of democratic institutions.

Haidt's prescription is simple but demanding: leave the social media bubble.

The most effective way to understand other moral frameworks is through interaction with different people, cultures, and moral “palates.”

Reducing political conflict will not come from producing more arguments, or even better arguments.

It will come from understanding the moral frameworks that produce those arguments in the first place.

Only then can disagreement stop being a moral battle and become a conversation again.

Subscribe now

From now on, I’ll be writing my Substack posts in English. Many of the books and sources that inspire these essays were originally read in Portuguese, so some translations may occasionally feel imperfect.

I hope readers will forgive any linguistic imperfections. The goal here is to explore ideas, patterns, markets, and economic history.

Now, to the essay.

Political systems rarely collapse overnight.

They erode slowly, often under the justification of efficiency, security, or national interest.

History shows that when governments expand their control over economic life, the boundaries between democratic authority and arbitrary power can gradually blur.

It was precisely this dynamic that concerned Friedrich Hayek when he wrote The Road to Serfdom.

This essay draws on that book to reflect whether some of those institutional dynamics can be observed in today’s United States.

The goal is not to claim that the United States is becoming a totalitarian state. Rather, the objective is to examine whether some of the institutional dynamics described by Hayek may help illuminate aspects of today’s political and economic environment.

Hayek divided the book into fifteen short chapters addressing themes such as the relationship between the state and economic control, liberty, totalitarianism, and democracy. This essay focus on four of them:

1. Planning and Democracy

2. Planning and the Rule of Law

3. Why the Worst Get on Top

4. The End of Truth

Planning and Democracy

Main ideas of the chapter:

1.1) The stronger the state and its control over economic resources, the weaker liberal democracy tends to become. In other words, the more the state controls resources, the more it indirectly controls individual choices.

1.2) In a highly interventionist state, individual choices become shaped by centralized incentives. Those who control the ends and incentives also condition the means required to achieve them.

1.3) Hayek rejects the assumption that democratic power cannot be arbitrary. In his view, the fact that a government has been democratically elected does not prevent it from acting arbitrarily.

1.4) Democracy is not an end in itself but a means of protecting internal peace and individual liberty. As such, it is neither perfect nor infallible.

1.5) What prevents power from becoming arbitrary is not its source (the people), but its limitation. Democracy alone neither guarantees nor prevents arbitrary power.

How is Trump weakening democratic institutions in the United States?

The United States (like many other countries) currently faces a deeply polarized and identity-driven political environment.

In such a context, politics shifts away from a competition of ideas and toward a focus on the efficiency of action, often bypassing institutional procedures in favor of ideological outcomes.

After a decisive electoral victory and backed by strong popular support, Trump has increasingly relied on executive orders to advance his agenda, expanding the scope of discretionary presidential power. As the chart below shows, Trump's second term has already reached record levels in the number of executive orders issued.

Source: Goldman Sachs Investment Strategy Group

Planning and the Rule of Law

Main ideas of the chapter:

2.1) A country is free when all government actions are governed by general rules that are previously established and publicly known—the principle known as the rule of law.

2.2) Under the rule of law, the government should define the rules but not control how they are implemented. Whoever controls the implementation of the means effectively determines outcomes. In a well-functioning system, results emerge from open competition under established and impartial rules.

2.3) For the rule of law to function properly, consistent application of rules — without exceptions — is often more important than the content of the rules themselves.

2.4) The rule of law provides predictability, preventing ad hoc interventions and allowing individuals to plan their lives and economic activities.

2.5) Ad hoc interventions, in Hayek’s view, create privileges that weaken the rule of law.

The erosion of previously established rules

Combined with the dynamics above, the frequent use of executive orders not only weakens democratic processes — it breaks previously established rules and creates privileged groups, undermining the rule of law.

Measures such as a proposed 10% cap on credit card interest rates, selective taxation on Nvidia chips, and proposed additional energy charges targeting large technology companies — alongside the acquisition of Intel and the possibility of intervention in the Federal Reserve — do not appear to aim at reforming the rules of the game to make markets more competitive or fair.

In practice, these measures intervene directly in how rules are applied, creating winners and losers, increasing legal uncertainty, expanding arbitrariness, and weakening democratic processes.

Subscribe now

Why the Worst Get on Top

Main ideas of the chapter:

3.1) According to Hayek, the pre-totalitarian stage often fosters the cult of a strong leader, driven by dissatisfaction with the slow pace of democratic processes.

3.2) This dissatisfaction generates demand for a decisive and forceful leader.

3.3) Debate ceases to be the central political value and is replaced by the efficiency of action.

3.4) Totalitarian regimes do not arise from broad public support but from the intense and disciplined backing of a smaller, highly organized group willing to display unconditional loyalty.

3.5) The worst individuals are often those most willing to disregard ethical constraints in pursuit of ideological objectives.

Trump’s leadership and the rhetoric of America First

Driven by a “us versus them” political narrative, deepening polarization, and the near-unconditional support of a loyal base, Trump has demonstrated — through the seizure of Maduro, proposals to annex Greenland, and the strike on Iran — a willingness to subordinate institutional constraints to strategic objectives.

For the purposes of this essay, the point is not to debate the morality of such actions but to highlight the erosion of democratic restraints. Hayek had already identified the tendency of authoritarian leaders to justify actions that would ordinarily be morally constrained by invoking national or group interest.

When Trump argues that the only limit to his actions is his own moral judgment —and that he sees no need to follow international law — combined with the rhetoric of America First and national security, the framework Hayek described becomes increasingly legible.

The End of Truth

Main ideas of the chapter:

4.1) Totalitarianism does not begin with explicit censorship; it begins when the state starts shaping the public’s perception of reality.

4.2) In competitive political systems, propaganda is essentially a dispute between ideas. In authoritarian systems, it becomes the manipulation and reformulation of values.

4.3) When the state dominates the media environment, the challenge goes beyond evaluating information—it becomes difficult to determine whether the information itself is true.

4.4) Information that could raise doubts about government competence or generate dissatisfaction is often concealed from the public.

To what extent are media outlets being influenced?

Paraphrasing Eurasia Group:

The acquisition of Paramount by the Ellison family—politically aligned with Trump—gives them control of CBS. The group has also signed agreements related to TikTok’s U.S. operations and is reportedly pursuing CNN from Warner Bros. Discovery.

Combined with X under Elon Musk, Fox under Rupert Murdoch, and Trump’s own Truth Social, a significant share of traditional media and social networks in the United States may end up under the control of actors aligned with the current administration.

If the main sources of information about the government are platforms such as Truth Social — and the major media outlets are controlled by allies of the sitting president—how can the public form an independent judgment about its actions?

How do these dynamics affect financial markets?

Political institutions and financial markets are deeply intertwined. The credibility of a currency, the stability of sovereign debt, and the perception of safe-haven assets all ultimately rest on institutional trust.

The question that emerges is whether we should continue to treat the U.S. dollar and U.S. Treasuries as unquestioned stores of value.

For decades, these assets have functioned as safe havens, even during crises that originated within the United States itself.

But with the weakening of the American-led global order and the possibility of structural shifts in the international system, how might these assets behave going forward?

Several indicators are worth watching:

A) The increase in gold purchases by global central banks, especially following the war in Ukraine and the freezing of Russian assets.

B) The decline in the U.S. Dollar Index since “Liberation Day”.

C) The sharp rise in gold prices since 2022.

Source: Kinea. Gold purchases by Central Banks

Source: Bloomberg

Conclusion

Many of the themes raised by Hayek in The Road to Serfdom resonate with the current political environment in the United States. We observe a weakening of institutional checks and balances and a growing reliance on unilateral decisions through executive orders.

Trump's leadership has become increasingly unpredictable, driven by the rhetoric of America First and national security.

In this context, what many observers have begun to call American state capitalism becomes easier to recognize.

I prefer, however, to close this essay on a slightly more constructive note, recalling the words of Alexis de Tocqueville:

“The greatness of America lies not in being more enlightened than any other nation, but rather in her ability to repair her faults.”

Subscribe now

From now on, I’ll be writing my Substack posts in English. Many of the books and sources that inspire these essays were originally read in Portuguese, so some translations may occasionally feel imperfect.

I hope readers will forgive any linguistic imperfections. The goal here is to explore ideas, patterns, markets, and economic history.

Now, to the essay.

Every technological revolution begins with the same promise: this time is different.

Artificial intelligence may indeed transform the world. But history suggests that when a technology captures the collective imagination of investors, financial euphoria often follows.

Today’s post is inspired by the book A Short History of Financial Euphoria by John Kenneth Galbraith.

The aim is to highlight recurring financial behaviors and dynamics that tend to emerge during periods of economic crisis — particularly during moments of euphoria — and to draw parallels with today's markets, especially in relation to artificial intelligence.

Before drawing comparisons between past financial crises and AI, I will briefly revisit three themes explored by Galbraith:

1. Speculation

2. Euphoria

3. The bursting of bubbles

Speculation

Galbraith defines speculation as the combination of collective imagination with something new—usually a technology—that gradually turns into a belief system.

In this environment, markets stop pricing economic utility and instead begin pricing the ability of the new technology to capture imagination.

During speculative phases, those who question asset prices are accused of “not understanding the business model.” After all, this time is different— something fundamentally new has entered the world.

Criticism becomes almost a form of intellectual heresy, a sign of ignorance, a sin against modernity.

Questioning the narrative starts to be perceived as a moral failure.

As a society, we tend to assume that the greater an individual’s wealth, the sharper their economic judgment — and therefore the more refined their reasoning must be.

In this context, markets, corporations, and increasingly CEOs become moral and intellectual authorities, almost infallible.

Modern capitalism, in other words, routinely converts financial success into intellectual authority. During periods of euphoria, the right to doubt is quietly suspended.

Euphoria

Euphoria is not immediate delusion, it is a temporary truth. It creates the illusion of prosperity.

As a bubble expands, capital inflows (especially foreign capital) push asset prices higher across the entire market, not just in sectors directly related to the speculative theme. Prosperity appears real until, without any clear catalyst, selling accelerates, liquidity disappears, and prices fall faster than anyone imagined.

The rush to sell quickly turns into panic.

Before the bursting of a bubble, there are typically two types of buyers:

1. Those who believe markets will continue rising because a new economic reality has emerged

2. Those who recognize the speculative environment but believe they will be able to time the market

Both groups, in their own way, believe that their gains reflect superior insight or intuition.

No one, in the heat of a bull market, attributes their growing wealth to luck.

Subscribe now

The Bursting of the Bubble

First, it is important to remember that the cause of a bubble is not the trigger event itself. It is speculation.

Second, the end of a bubble is never gradual. There is always widespread selling, intensified by panic and herd behavior.

It is also worth noting that bubbles do not arise only from useless assets, tulips being the classic example. They can also emerge around essential infrastructure.

Economic utility does not prevent financial euphoria.

The railroad crisis of 1873 began with the collapse of Jay Cooke & Company, a bank highly concentrated in financing the railroad sector. The New York Stock Exchange closed for ten days and suspended payments in hard coin.

Who could have imagined a financial collapse triggered by financing one of the most necessary infrastructure sectors and most transformative technologies of the era?

This leads to the final point of this introduction and connects directly with today’s markets:

Every financial crisis is ultimately a product of leverage.

If we are indeed experiencing an AI bubble, who is taking the credit risk? Corporations? Governments? Or investors?

The central question is not whether AI is useful.

The real question is: where is the leverage that could turn disappointed expectations into a systemic event?

Below is a timeline of the dot-com bubble of the early 2000s. Beginning in 1996, the internet became a dominant narrative. Major IPOs such as Yahoo and Amazon captured investor attention.

Below is a timeline of the dot-com bubble. Beginning around 1996, the internet became a dominant market narrative. Major IPOs such as Yahoo and Amazon, attracted enormous investor attention.

The Federal Reserve chairman warned of "irrational exuberance."

The Nasdaq peaked in early 2000. It took roughly fifteen years to recover

Source: Visualizing Economics

With this historical framework in mind, the obvious question is whether today’s AI boom follows a similar pattern.

AI Investments

Considering only Meta, Alphabet, Amazon, and Microsoft, announced investments related to AI already exceed $600 billion (compared with roughly $370 billion in 2025).

Source: John Authers, Bloomberg

When we expand the scope to include the entire AI ecosystem—including data centers, hardware, and software—the figure approaches $1 trillion, equivalent to roughly 3.5% of U.S. GDP.

Source: Joseph Politano

To put this into perspective, as a share of GDP this level of investment is comparable to the combined scale of the largest American infrastructure projects since the twentieth century, including the Manhattan Project, the Apollo Program, and the telecommunications investments of the early 2000s.

Source: JPM Asset Management

Subscribe now

The Energy Equation

In a Goldman Sachs report titled AI: In a Bubble?, David Cahn raises an important point.

According to Cahn, the scale of current AI investments is so large that projections are now measured in gigawatts, not billions of dollars.

To support the projected expansion of AI infrastructure, analysts estimate that 100 to 250 gigawatts of additional power generation capacity may be required by 2030.

The problem is that each additional gigawatt costs roughly $40 billion to build. In other words, constructing 100 to 250 gigawatts would require between $4 trillion and $10 trillion in investment—equivalent to 13% to 33% of U.S. GDP.

Source: David Cahn

According to Cahn, in order to justify such investment, the AI industry would need to generate between $8 trillion and $20 trillion in annual revenue.

For comparison, U.S. GDP is roughly $30 trillion, while global GDP is around $110 trillion.

We know AI will change the world. But will the investments pay off?

Circular Investment in the AI Industry

In a healthy and mature industry, the price paid by the final consumer should be sufficient to finance the entire production chain.

In emerging industries, however, demand during the early stages of a technological cycle is often insufficient to sustain the entire supply chain.

In these cases, suppliers themselves take on the risk: they invest in production capacity and hope that future demand will eventually absorb that supply.

The AI industry appears to be in this stage. The difference, as discussed earlier, is the scale of investment involved.

Today, the risk created by the mismatch between supply and demand is largely being absorbed by the Big Tech companies.

One example of cross-financing within the industry is the massive capital flows between infrastructure providers, chip manufacturers, and AI model developers. For instance, Nvidia recently announced $100 billion in investment in OpenAI.

It is worth remembering that OpenAI reported revenue of roughly $20 billion in 2025.

What happens if demand fails to match projected supply?

If the final consumer does not ultimately finance the production chain, one of its links may break. Part of the newly built capacity would remain unused, and some of the associated debt might never be repaid.

The deeper problem is that the industry itself financed the expansion of that capacity. Under such circular investment dynamics, the entire ecosystem can begin to resemble a house of cards.

Source: Visualizing Economics

AI and the Labor Market

Joseph Politano describes three phases in the post-pandemic labor market:

1. The Great Resignation (2021–2022), when millions of workers quit their jobs to seek higher-paying positions

2. Low-hire, low-fire equilibrium (2023–2024), when hiring and layoffs occurred at balanced levels

3. The no-hire economy (2025), in which layoffs returned to pre-pandemic averages but hiring fell to levels last seen in 2013

Recent data highlights the slowdown in the labor market over the past two years. The United States has created virtually no new jobs over the past five months. The unemployment rate is at its highest level since early 2021, and job creation in 2025 is already below the average of any year during the 2010s.

Source : Joseph Politano

When we examine the U.S. economy more broadly, an unusual dynamic emerges: economic growth alongside labor market weakness.

Historically, GDP expansion tends to be accompanied by job creation. The current divergence raises the possibility that we may be witnessing the early stages of productivity gains driven by AI.

Source: Blackrock

Who benefits from this dynamic?

Primarily shareholders—those who capture the gains from productivity improvements. In practical terms, this means older generations, such as Baby Boomers, and those who already own financial assets.

As a result, wealth concentration intensifies.

If productivity increases without a proportional expansion in employment, the gains flow to capital rather than labor. Historically, such dynamics have often preceded periods of political tension.

Source: Liz Ann Sonders.

Who is most negatively affected by this shift?

Younger generations, especially Generation Z and those entering the labor market today.

Source : Joseph Politano

It is also important to note that the recent improvement in corporate profitability has been driven largely by cost reductions rather than new AI-driven revenue streams. At a macroeconomic level, we still do not see clear evidence of a structural revenue surge tied to AI.

A study from Massachusetts Institute of Technology suggests that 95% of companies have yet to extract measurable value from their AI investments.

Given the enormous scale of AI investment, there is reason to worry that cost reductions alone may not be enough to justify the spending.

Sustained returns may require real economic expansion—not simply a redistribution of income from labor to capital.

Source: Joseph Politano

Conclusion

Artificial intelligence will almost certainly change the world. That is not the question.

The real question is: who will be the winners, and who will ultimately pay for the losers?

Even if we are witnessing a period of overinvestment, that does not necessarily mean the technology will fail. Historically, major waves of overinvestment often destroy capital in the short term but build infrastructure that enables long-term growth. Railroads, fiber optics, and the dot-com boom all followed this pattern.

The scale of investment currently undertaken by American Big Tech companies appears unprecedented. It is difficult to imagine that all of these companies will emerge as simultaneous winners from this technological wave—especially when we consider the circular nature of many of the investments and the potential for a house-of-cards dynamic.

At the same time, given the enormous capital already committed, it is possible that we are still somewhere between the beginning and the middle of the cycle. The dot-com bubble also took years to fully unfold.

There is a great deal of noise in the debate surrounding AI. Some argue that humans will become obsolete within a few years. Others claim we are approaching the limits of available data and diminishing returns.

In this essay, I chose to present the counterpoints to the most enthusiastic narratives. CEOs already provide enough marketing.

As Galbraith reminds us, the cause of a bubble is never the trigger itself. It is speculation.

The trigger merely reveals where the leverage was hiding.

Subscribe now

A resposta para a pergunta acima é: sim e não. Ou melhor, sim e talvez não tanto quanto a maioria das pessoas pensa.

Claro, se você mora no Brasil, possui renda em reais e seus investimentos estão majoritariamente denominados em real, você está extremamente exposto ao “risco Brasil”. Nesse sentido, sim, investir nos Estados Unidos aumenta a sua diversificação.

Esta é a parte fácil da análise, e esse texto não é sobre isso.

Este post é sobre por que investir nos EUA talvez não represente tanta diversificação quanto parece.

Spoiler: é, em ampla medida, uma grande aposta em AI.

Nesta análise abordarei os seguintes tópicos:

1. A grande concentração das 10 maiores empresas americanas em valor de mercado

2. A concentração de patrimônio e consumo do mercado americano

3. O aumento dos investimentos passivos via ETFs e seus efeitos no mercado

4. Por que isso é relevante

Subscribe now

1. A grande concentração das 10 maiores empresas americanas

De acordo com a Goldman Sachs, as 10 maiores empresas do S&P 500 - equivalente a 2% das empresas que compõem o índice – representam aproximadamente 40% do valor de mercado agregado e 30% do seu lucro acumulado do índice.

Destas 10 empresas, 8 estão diretamente relacionadas ao mercado de AI¹, seja construindo modelos de linguagem, data centers, chips processadores, robôs humanoides, entre outros segmentos.

Fonte: Aquavis

Para colocar em perspectiva, as 10 maiores empresas americanas acumulam algo entre US$ 21 e US$ 24 trilhões de valor de mercado, aproximadamente dois terços do PIB americano.

Isso mesmo: 10 empresas concentram cerca de 2/3 do PIB dos Estados Unidos em valor de mercado. Se considerarmos que o PIB global é de aproximadamente US$ 110 trilhões, essas empresas representam algo próximo de 20% do PIB mundial.

Em termos de concentração econômica, podemos pensar que a economia americana se aproxima de um grande oligopólio, cujos produtos e serviços permeiam praticamente todos os setores da atividade econômica.

Como o lucro dessas empresas representa parcela relevante da economia (direta ou indiretamente), a maioria das empresas e indivíduos acaba contratando algum serviço vinculado às grandes companhias de tecnologia.

Outra forma de analisar a dependência da economia americana em relação à AI é observar o quanto de investimentos estão contratados pelas grandes empresas de tecnologia. Como demonstrado no gráfico abaixo, os investimentos em AI, em relação ao PIB americano, já superam grandes projetos históricos de infraestrutura.

Fonte: JP Morgan Asset Management

Como destaquei no meu último post, a grande pergunta aqui não é se AI vai mudar o mundo, mas sim se todas essas empresas vencerão a corrida ao mesmo tempo.

E, se não vencerem, quem pagará a conta dos perdedores?

2. A grande concentração de patrimônio e consumo do mercado americano

Nesta seção três temas precisam ser expostos: :

2.1) A concentração do mercado acionário em relação aos estratos de riqueza populacional

2.2) A concentração das ações em relação ao patrimônio total do investidor

2.3) A concentração do consumo americano na camada populacional mais afluente

2.1) A concentração do mercado acionário em relação aos estratos de riqueza populacional

Houve um grande ganho de alavancagem operacional nas grandes empresas americanas nas últimas décadas. Em outras palavras, essas empresas conseguiram crescer enquanto mantinham o número de funcionários e custos operacionais estáveis.

Como exemplo, a IBM empregava 400 mil funcionários em 1985, sendo então a maior empresa americana. Atualmente, a Nvidia emprega cerca de 40,000 mil funcionários, vale aproximadamente 20 vezes o valor de mercado da IBM à época e é 5 vezes mais lucrativa, com um décimo dos funcionários.

O gráfico abaixo mostra a divisão entre a contribuição do trabalho (labor) e do capital para os lucros corporativos. Fica clara a perda relativa da participação do trabalho ao longo do tempo.

Fonte: Aquavis

A principal consequência desse ganho de alavancagem operacional foi o aumento das margens de lucro das empresas, conforme demonstrado no gráfico abaixo.

Fonte: Goldman Sachs, Investment Strategy Group

Por que isso é importante?

Com a redução relativa da participação do trabalho nos lucros, o aumento de margem não foi repassado para os trabalhadores, mas sim para os donos do capital – acionistas, altos executivos e CEOs.

Podemos ver no gráfico abaixo a evolução da diferença de remuneração entre os CEOs e os demais funcionários.

Fonte: Aquavis

Todos esses fatores contribuíram significativamente para a crescente concentração acionária nos Estados. Atualmente, observamos o maior nível de concentração patrimonial desde 1980.

Os 10% mais ricos da população detém 87% do valor total do mercado acionário americano.

Fonte: Aquavis

2.2) A concentração das ações em relação ao patrimônio total do investidor

Quanto do patrimônio do americano médio está exposto ao mercado acionário?

De acordo com o Federal Reserve, mais de 45% da poupança financeira das famílias está alocada em ações.

Em outras palavras:

i) A maior parte do valor de mercado está concentrada na camada mais rica da população.

ii) Considerando o patrimônio dessa camada mais afluente, uma parcela relevante está concentrada em ações.

Ou seja, há uma dupla concentração; uma em renda e outra em patrimônio.

Fonte: Federal Reserve e Kobeissi letter

Deixo a pergunta: e se essa camada mais rica enfrentar algum problema relevante em suas empresas e precisar de liquidez?

De onde virá o saque?

2.3) A concentração do consumo americano na camada populacional mais afluente

Agora, vamos ao terceiro ponto de concentração.

Não apenas o patrimônio está concentrado em poucos, e que por sua vez está concentrado em ações. O consumo também está concentrado.

De acordo com o Financial Times, os 10% mais ricos da população respondem por aproximadamente 50% do consumo americano.

Fonte: Financial Times

Por que o consumo está tão concentrado?

Não houve correção real de salários proporcional ao aumento de lucro das empresas (como já demonstrado). A inflação reduziu o poder de compra das famílias, comprimindo o consumo e a poupança das camadas de menor renda.

Grande parte desse movimento foi acentuado pela injeção monetária do FED pós-COVID, que elevou a inflação e inflou o preço dos ativos financeiros - mas isso é tema para outro post.

Hoje os EUA vivem uma “economia em K”.

Para os afluentes, a economia parece saudável, seu patrimônio continua se valorizando, e o consumo segue forte.

Enquanto isso, as camadas de menor renda vivem “boleto a boleto”, com o poder de compra cada vez mais deteriorado pela inflação.

Para encerrar esta seção, considere os seguintes pontos:

1. Grande parte do consumo da “economia em K” está sendo sustentada pelos mais afluentes

2. Grande parte da economia está sendo sustentada pelos investimentos em AI

3. Grande parte da confiança de consumo dos afluentes vem do seu robusto patrimônio - que por sua vez está concentrado em ações, que, por sua vez, estão concentrados em empresas ligadas a AI.

Se economia, investimentos e consumo estão extremamente interconectados, o que acontece com os mercados se algum elo da cadeia de AI quebrar ou não performar como esperado?

Subscribe now

3. Aumento dos investimentos passivos via ETFs e seus efeitos no mercado

Espero que eu tenha demonstrado até aqui que economia, mercados, patrimônio estão extremamente interconectados ao setor de AI.

Nesta seção, pretendo demonstrar como os mercados podem reagir caso algum elo dessa cadeia quebre.

Abordarei os seguintes tópicos:

3.1) Estados Unidos como império (Ray Dalio)

3.2) Aumento do uso de ETFs e da concentração da gestão passiva

3.3) Aumento dos buybacks reforçando o fluxo de compra

3.4) Valuations e queima de caixa

3.1) Estados Unidos como império (Ray Dalio)

Considere que os Estados Unidos se tornaram o império global após a Segunda Guerra Mundial (como argumenta Ray Dalio). Neste momento, o dólar passou a ser a moeda de reserva global, e o banco central americano adquiriu o chamado “privilégio exorbitante” de emitir enormes quantidades de papel moeda.

Além de emitir a principal moeda de reserva do mundo, os EUA também passaram a ter a capacidade de emitir grandes volumes de dívida e, ainda assim, serem considerados um porto seguro.

Isso é extremamente relevante, pois mantém seus os juros estruturalmente baixos, reduz o custo de capital e favorece o desenvolvimento do mercado empresarial americano².

Para simplificar, ao longo das últimas décadas, esse “privilégio exorbitante” permitiu aos EUA absorver um volume colossal de capital — tanto doméstico quanto estrangeiro.

Porém, surge agora uma nova narrativa nos mercados: “A mudança da ordem global”

Para fechar esta seção, trago o conceito de reflexivity de George Soros: a ideia de que, nos mercados financeiros, a percepção dos participantes não apenas reflete a realidade, ela também a altera.

E com isso, deixo as duas perguntas abaixo:

Se os mercados começarem a mudar a narrativa, reduzindo a confiança dos EUA como império global, sua moeda de reserva e em sua dívida, como isso reflete nos mercados?

Haverá uma saída relevante de investimentos nos EUA, principalmente dos investidores estrangeiros?

3.2) Aumento do uso de ETFs e aumento da concentração da gestão passiva

Decidi abordar o aumento do uso de ETFs porque esses veículos passaram a representar um fluxo de capital enorme, tanto doméstico quanto estrangeiro.

Os ETFs não consideram valuation em suas operações. Eles simplesmente compram e vendem ativos de acordo com o peso que cada empresa possui no benchmark replicado.

Meu receio em relação ao crescimento da gestão passiva está no momento em que houver um fluxo relevante de saída de capital dos Estados Unidos. Nesse cenário, poderemos observar um movimento de venda exacerbada (e também indiscriminada), independentemente dos fundamentos ou dos valuations.

Incluo os temas abaixo para melhor explicar meu receio:

Desde 2019, com o avanço da tecnologia, a ampliação das opções de investimento e, até certo ponto, uma mudança cultural, o mercado passou a ter maior participação da gestão passiva em relação à gestão ativa.

Ou seja, em vez de alocar recursos em uma gestora que monta carteiras com base em análise fundamentalista, muitos investidores passaram a preferir ETFs; veículos que replicam índices como S&P 500, Nasdaq ou até mesmo índices temáticos (por exemplo, energia ou tecnologia).

Fonte: Aquavis

A resiliência e o crescimento dos mercados americanos, somados à dificuldade histórica de gestores ativos superarem seus benchmarks no longo prazo, reforçaram essa migração para a gestão passiva.

É importante relembrar como funciona essa dinâmica.

Quando um investidor compra o S&P 500 via ETF, ele compra todas as empresas do índice em seus respectivos pesos. Se investe US$ 100 mil e determinada empresa representa 5% do índice, ele automaticamente aloca US$ 5 mil nessa empresa, independentemente de valuation ou fundamentos.

Em outras palavras, trata-se de uma compra mecânica e indiscriminada.

Diversos analistas argumentam que essa dinâmica contribui para o aumento da concentração: quanto maior o peso da empresa no índice, maior o fluxo automático de capital para ela.

Isso pode gerar um efeito bola de neve.

O aumento da gestão passiva e sua compra mecânica contribuíram para o aumento da concentração do mercado a partir de 2019?

Retomo o questionamento:

Como o mercado se comportará em um momento de fluxo relevante de venda?

Os ETFs amplificarão esse movimento?

3.3) Aumento dos buybacks reforçando fluxo de compra

Com o aumento de escala operacional já mencionada, as grandes empresas passaram a gerar volumes expressivos de caixa, e se tornaram cada vez mais “asset light” – empresas que necessitam de relativamente pouco capital para continuar crescendo.

Com esse excesso de geração de caixa, e também com o aumento do pagamento de remuneração em ações, essas empresas passaram a recomprar suas próprias ações no mercado, o chamado buyback.

Fonte: Aquavis

Isso é relevante porque reforça o fluxo estrutural de compra.

Ou seja, além da compra mecânica promovida pelos ETFs, há também a recompra sistemática por parte das próprias empresas.

Mais um elemento que contribui para o efeito bola de neve positivo.

Mais à frente no texto, mostro como os grandes investimentos em AI estão levando as empresas a começar a queimar caixa, deixando gradualmente de ser asset light. Isso é importante porque, com menos caixa disponível, haverá menor capacidade para realizar buybacks, o que pode alterar o fluxo estrutural de compra no mercado.

3.4) Valuations e queima de caixa

Primeiro ponto importante: o mercado não está barato.

Em termos de múltiplo preço/lucro, encontra-se em um dos patamares mais elevados dos últimos 10 anos, e isso em meio aos riscos já discutidos neste post.

Fonte: Factset, fevereiro de 2026

Por que isso é relevante?

Porque não estamos em um cenário em que os múltiplos estejam deprimidos e os riscos já tenham sido amplamente precificados.

Há muita gordura pra queimar.

Segundo ponto: as grandes empresas de tecnologia vêm alterando seu perfil financeiro. Atualmente, estão investindo volumes significativamente maiores em CAPEX, impulsionadas pela corrida por AI.

Como demonstrado no gráfico, algumas passaram a queimar caixa em vez de gerar caixa, dado o volume de investimento necessário.

Fonte: Mufgamericas

Os mercados já ajustaram suas projeções de geração de caixa para 2026.

Fonte: Factset, Evercore ISI Research

Se essas empresas deixarem de gerar caixa no mesmo ritmo de antes, é provável que reduzam seus programas de buyback, o que implica menor fluxo estrutural de compra.

Como a redução dos buybacks impactará o fluxo de capital e, consequentemente, os valuations?

Subscribe now

4. Porque isso é relevante

O sistema funciona enquanto o fluxo entra. ETFs compram, empresas recompram ações, patrimônio sobe, consumo se mantém, AI recebe mais investimento. Se fosse o título de um filme, eu chamaria de “AI: O Novo Motor do ‘Privilégio Exorbitante’ Americano”

Porém, todos esses elementos são a mesma engrenagem. Se o fluxo inverter, o mecanismo pode funcionar na direção oposta.

É possível contra-argumentar que essa concentração é justamente o reflexo da superioridade estrutural americana: maior produtividade, maior inovação e melhor alocação de capital. A concentração pode ser eficiência, não fragilidade.

O ponto não é negar a força estrutural dos EUA, mas entender que apesar da sua superioridade, pode haver vulnerabilidade sistêmica quando todos os fluxos se concentram no mesmo vetor.

Nesta seção, tentarei construo alguns cenários considerando os riscos discutidos até aqui, nos seguintes tópicos:

4.1) Potencial saída de investimento estrangeiro

4.2) Redução de patrimônio, consumo, emprego, e atividade econômica

4.3) Fluxo de venda exacerbada e indiscriminada

4.1) Potencial saída de investimento estrangeiro

O mercado americano apresenta elevada participação de investidores estrangeiros, conforme demonstrado no gráfico abaixo. Atualmente, investidores estrangeiros representam aproximadamente 35% dos ativos financeiros americanos, incluindo mercado acionário e dívida.

Em termos absolutos, estrangeiros detêm cerca de US$ 70 trilhões em ativos americanos, valor próximo a 70% do PIB global.

Fonte: Aquavis

Com o aumento da percepção de risco nos Estados Unidos e a crescente narrativa de “mudança da ordem global”, surge a pergunta:

O restante do mundo continuará investindo na mesma proporção nos EUA?

O início de 2026 já trouxe indícios parciais de resposta. Observamos forte fluxo de capital para mercados emergentes, nos maiores níveis em anos. A bolsa brasileira, por exemplo, atingiu máximas históricas.

Fonte: Bloomberg

Até o momento, não houve uma saída relevante de capital dos Estados Unidos — apenas uma redistribuição do capital novo, marginal. O gráfico abaixo ilustra esse ponto: não observamos fluxos líquidos de saída, nem no mercado acionário nem no mercado de dívida.

Fonte: Bloomberg

Mas como os mercados reagiriam diante de uma eventual saída efetiva de capital estrangeiro?

É importante lembrar que, assim como as compras via ETFs são indiscriminadas, as vendas também tendem a ser.

4.2) Redução de patrimônio, consumo, emprego, e atividade econômica

A economia em K, os investimentos em AI e o mercado acionário estão profundamente interconectados.

O que acontece se algum elo da cadeia de AI quebrar?

Ou se parte dessas empresas não entregar os resultados esperados pelo mercado?

Veremos redução de patrimônio, levando a menor consumo e, consequentemente, a uma economia mais fraca?

Considerando que a parcela de menor renda já sofre com inflação e deterioração do poder de compra, esses indivíduos conseguirão honrar suas dívidas caso haja aumento do desemprego (inclusive potencialmente impulsionado por AI)?

Já observamos queda no savings rate e redução da poupança das famílias americanas, atualmente nos níveis mais baixos desde 2008.

Fonte: Bloomberg

Como os mercados de crédito estão refletindo esse risco, especialmente cartão de crédito e hipotecas? Já começamos a observar elevação nos indicadores de inadimplência.

Fonte: Federal Reserve

Há também risco relevante no mercado de hipotecas comerciais, mas esse é um tema para outro post.

4.3) Fluxo de venda exacerbada e indiscriminada

Por muitos anos, e por diversos fatores (incluindo o status de império global), os EUA receberam fluxo constante de capital, gerando um ciclo virtuoso e um efeito bola de neve positivo.

Pelos motivos discutidos ao longo deste post, vejo risco de um movimento de venda indiscriminada, especialmente diante do aumento da concentração na gestão passiva.

Ou seja, o início de um possível efeito bola de neve negativo.

Subscribe now

Conclusão

Sim, o investidor, de forma geral, aumenta sua diversificação ao investir no mercado acionário americano.

No entanto, conforme demonstrado, embora o S&P 500 seja composto por 500 empresas, ele não é tão diversificado quanto aparenta.

Os mercados e a economia estão tão interconectados que investir nos Estados Unidos se tornou, em grande medida, uma grande aposta em AI.

Ou seja, investir nos Estados Unidos deixou de ser apenas uma decisão geográfica ou diversificação. É uma decisão temática.

Se essa aposta der certo, os retornos podem ser extraordinários. Se não der, a interconexão pode transformar concentração em fragilidade

Mantenho a mesma pergunta da última publicação:

A dúvida não é se AI mudará o mundo, mas quem serão os vencedores dessa corrida?

E quem pagará a conta dos perdedores?

Rodapé:

1. Nvidia, Apple, Alphabet, Microsoft, Amazon, Meta, Broadcom, Tesla

2. É claro que não foi apenas isso: cultura empreendedora, tecnologia, instituições, rule of law, capacidade militar e educação, também desempenham papel fundamental

Porém, antes de traçar os paralelos entre crises financeiras passadas e inteligência artificial, farei referência a três tópicos abordados por Galbraith:

1. Especulação

2. Euforia

3. Estouro da bolha.

Especulação

Galbraith define especulação como a combinação da capacidade coletiva e popular de imaginação acoplada a algo novo (normalmente uma tecnologia), que se torna quase uma crença. Neste cenário, os mercados deixam de precificar a utilidade econômica, e passam a mensurar a capacidade (da nova tecnologia) de capturar imaginação

Durante a especulação, quem critica os preços dos ativos é quem “não entendeu o modelo de negócio”, pois dessa vez é diferente, há algo novo no mundo. A crítica se transforma quase em heresia intelectual, sinal de ignorância e pecado contra a modernidade. Criticar passa a ser visto como uma falha moral.

Temos, como sociedade, a tendência de acreditar que, quanto maior for o patrimônio de um indivíduo, maior é sua percepção econômica e social e, portanto, mais apurado é seu processo mental. Nesse contexto, os mercados, as empresas e, mais recentemente, os CEOs se tornam autoridades morais e intelectuais, praticamente infalíveis. Em outras palavras, o capitalismo moderno transforma o sucesso financeiro em autoridade intelectual e, em momentos de euforia, suspende o direito à dúvida

Euforia

A euforia não é delírio imediato, é uma verdade temporária. A euforia cria prosperidade ilusória.

No momento de crescimento da bolha, o influxo de capital, principalmente estrangeiro, deixa tudo mais caro, aumentando o nível de preço de todo o restante do mercado, e não apenas dos ativos diretamente relacionados à bolha. A prosperidade parece real até que, sem causa aparente, a venda se acelera, a liquidez some e a queda acontece mais rápido do que qualquer um imaginava. A pressa para vender se transforma em pânico.

Antes do estouro da bolha, há dois tipos de compradores:

1. Os que acreditam que os mercados vão continuar em ascenção (é uma nova realidade);

2. Os que percebem o momento especulativo e acreditam que conseguirão “time the market”.

Os dois, de certa forma, acreditam que seu aumento de riqueza é fruto de insight ou intuição superiores. Ou seja, ninguém acredita que o aumento de riqueza é sem mérito.

Estouro da bolha

Primeiramente, precisamos lembrar que a causa de uma bolha não é o gatilho em si, mas sim a especulação.

Além disso, o fim da bolha nunca é gradual, sempre há venda generalizada, intensificada pela sensação de pânico, gerando um efeito manada. “Gradually, then suddenly”, como gostam de dizer os americanos.

Outro ponto importante, é que as bolhas não surgem apenas de ativos inúteis, como as tulipas, mas também de infraestrutura essencial. A utilidade econômica não impede euforia financeira.

A crise das railroads em 1873 estourou com a quebra do Jay Cooke & Company, banco altamente concentrado no financiamento do setor ferroviário. A Bolsa de Nova York fechou por 10 dias e suspendeu pagamentos em hard coin. Quem poderia quebrar financiando um setor de infraestrutura tão necessário e uma tecnologia tão revolucionária?

O que leva para o último ponto dessa introdução, e que faz conexão com os mercados de hoje: Toda crise financeira é um produto de alavancagem.

Se estamos em uma bolha de AI, quem está tomando o risco de crédito? As empresas? Os governos? Ou os investidores?

A pergunta central não é se AI é útil, mas onde está a alavancagem que pode transformar frustração de expectativas em evento sistêmico?

Segue abaixo a timeline da bolha dot-com dos anos 2000. A partir de 1996 a internet se torna uma narrativa, há IPOs relevantes como Yahoo e Amazon, o presidente do Fed alertou sobre “irrational exuberance” e, finalmente, o Nasdaq atingiu seu ápice no início dos anos 2000. Vale lembrar que levou aproximadamente 15 anos para o Nasdaq retomar o nivel do pico da bolha dot-com.

Fonte: Visualizing Economics

Subscribe now

Investimentos em AI

Considerando apenas Meta, Alphabet (Google), Amazon e Microsoft, os valores anunciados de investimento relacionados à AI já ultrapassam US$ 600 bilhões (vs US$ 370 bilhões em 2025).

Fonte: John Authers, Bloomberg

Quando ampliamos o escopo para o ecossistema de AI, incluindo data centers, computadores e software, o valor se aproxima a U$ 1 trilhão, equivalente a 3,5% do PIB americano.

Fonte: Joseph Politano

Para colocar em perspectiva, como porcentagem do PIB, é equivalente à soma dos maiores projetos de infraestrutura americanos desde o século XX, incluindo o projeto Manhattan, Apollo e os investimentos em telecomunicações nos anos 2000.

Fonte: JPM Asset Management

A equação energética

Em um relatório da Goldman Sachs intitulado AI: In a Bubble?, David Cahn (sócio da Sequoia Capital) levanta um ponto relevante. De acordo com Cahn, os valores de investimentos são tão elevados que o mercado passou a projetar expansões em gigawatts, e não mais em bilhões de dólares.

Para viabilizar os investimentos em AI, projeta-se algo entre 100 e 250 gigawatts adicionais de geração energética até 2030.

O problema dessa conta, é que cada gigawatt adicional custa, em média, U$ 40 bilhões de investimento. Em outras palavras, construir entre 100 e 250 gigawatts exigiria algo entre U$ 4 trilhões e U$10 trilhões, o equivalente a 13% e 33% do PIB americano.

Fonte: David Cahn

Segundo Cahn, para justificar esses investimentos, o mercado de AI precisaria atingir entre U$ 8 trilhões e U$20 trilhões em receitas. Como comparação, o PIB americano é de aproximadamente U$30 trilhões enquanto o PIB global é em torno de U$110 trilhões.

Sabemos que AI mudará o mundo. Mas os investimentos se pagarão?

Subscribe now

Investimentos circulares na indústria de AI

Em uma indústria saudável e madura, o preço pago pelo consumidor final deveria ser suficiente para financiar a produção de todos os elos da cadeia produtiva.

Em indústrias nascentes, durante o início do ciclo tecnológico, a demanda pelo produto ainda não é suficiente para sustentar toda a cadeia. Nesse cenário, os próprios fornecedores assumem o risco, investem em capacidade produtiva e esperam que a demanda futura eventualmente absorva essa oferta.

A indústria de AI está nesse momento. A diferença, como já demonstrado na seção anterior, é a escala dos investimentos em AI.

Hoje, quem absorve o risco do descasamento entre oferta e demanda são as Big Techs americanas. Um exemplo de financiamento cruzado dentro da cadeia, investimentos bilionários entre os fornecedores de infraestrutura, chips e modelos. Como exemplo, a Nvidia anunciou investimento de US$ 100 bilhões na OpenAI. Vale lembrar que a OpenAI reportou uma receita de aproximadamente “apenas” US$ 20 bilhões em 2025.

O que acontece se a demanda não atingir a oferta projetada?

O consumidor final não financia a cadeia produtiva e algum elo da cadeia quebra. Ou seja, parte da capacidade não será absorvida e parte da dívida não será paga. O problema, é que a própria cadeia financiou a expansão desse elo. Neste cenário de investimentos circulares, a indústria pode se tornar um grande castelo de cartas.

Fonte: Visualizing Economics

AI e seu efeito no mercado de trabalho

Joseph Politano descreve três fases no mercado de trabalho pós-pandemia:

1. Great Resignation (2021–2022), em que milhões de pessoas pediram demissão para serem contratados em posições com melhor remuneração;

2. Low hire, low fire equilibrium (2023–2024), dinâmica em que os empregadores ainda contratavam, mas também demitiam, de forma equilibrada;

3. No-hire economy (2025), caracterizado pelo volume de demissões em linha com a média pré-pandemia, porém, as contratações atingiram níveis de 2013.

O gráfico abaixo evidencia a desaceleração do mercado de trabalho nos últimos dois anos. Os Estados Unidos praticamente não criaram novos postos nos últimos cinco meses. A taxa de desemprego está no nível mais alto desde o início de 2021, e a média de geração de empregos em 2025 já é inferior à de qualquer ano da década de 2010.

Fonte : Joseph Politano

Quando analisamos o PIB americano, observamos uma dinâmica incomum: crescimento econômico em meio à desaceleração do mercado de trabalho. Historicamente, expansão do PIB costuma vir acompanhada de criação de empregos. A situação atual levanta a hipótese de estarmos diante de um início de ganho de produtividade impulsionado por AI. O gráfico abaixo mostra esse descasamento entre PIB e geração de empregos.

Fonte: Blackrock

Quem está se beneficiando dessa dinâmica? Essencialmente, os acionistas, aqueles que capturam os ganhos de produtividade. Em outras palavras, os Baby Boomers e quem já possui capital investido e a consequência é que concentração de patrimônio se intensifica.

Se produtividade cresce sem geração proporcional de empregos, o ganho vai para capital, não trabalho. Historicamente, essa dinâmica antecede ciclos de tensão política.

Fonte: Liz Ann Sonders.

Quem é o mais prejudicado por essa dinâmica? A geração Z e aqueles que estão ingressando agora no mercado de trabalho.

Fonte : Joseph Politano

Por fim, vale lembrar que o aumento recente de rentabilidade decorre majoritariamente da redução de custos, e não de novas receitas impulsionadas por AI. Ainda não vemos, em escala macro, um salto estrutural de faturamento associado à tecnologia. Uma pesquisa do MIT indica que 95% das empresas ainda não extraíram valor mensurável de seus investimentos em AI. Ao analisar o PIB americano, tampouco encontramos evidências claras de impacto direto da AI. Considerando o tamanho dos investimentos em AI, tenho receio de que o aumento de rentabilidade não seja suficiente e de que será necessário crescimento real da economia, e não apenas transferência de ganhos entre capital e trabalho.

Fonte: Joseph Politano

Conclusão

Certamente a inteligência artificial mudará o mundo, a dúvida não é essa. A dúvida é: quem serão os vencedores? E quem pagará a conta dos perdedores?

Mesmo que estejamos diante de excesso de investimento, isso não significa que a tecnologia falhará. Historicamente, grandes ondas de superinvestimento destroem capital no curto prazo, mas criam infraestrutura que sustenta crescimento no longo prazo (railroads, dot-com, fibra óptica).

O tamanho da aposta das Big Techs americanas parece sem precedentes. Tenho dificuldade de enxergar todas essas empresas como vencedoras simultâneas dessa nova onda tecnológica. Principalmente quando consideramos os investimentos circulares e o potencial castelo de cartas.

Por outro lado, como os investimentos já comprometidos são gigantescos, é possível que ainda estejamos entre o início e o meio do ciclo. A bolha dot-com também levou anos até estourar.

Há muita espuma no debate sobre AI. Não sabemos em quem acreditar: os que afirmam que humanos se tornarão obsoletos em poucos anos ou os que argumentam que estamos próximos do limite de dados e de retornos marginais decrescentes. Decidi apresentar aqui contrapontos aos revolucionários. CEOs já fazem marketing suficiente.

Voltando a Galbraith e retomando uma frase do início do texto: “a causa de uma bolha não é o gatilho em si, mas sim a especulação”.

Subscribe now

A intenção desse ensaio é aumentar a percepção e entendimento sobre por que as pessoas têm matrizes morais tão diferentes (e consequentemente visões de certo e errado tão distintas), e como a sociedade moderna e a tecnologia têm agravado nossas tendências identitárias e tribais.

Decidi dividir o post em quatro grandes blocos:

1. Por que pessoas mais inteligentes não são mais imparciais (e podem ser piores)

2. Principais diferenças morais entre conservadores e liberais

3. Como algoritmos transformaram diferenças políticas em confronto identitário

4. O aprofundamento da desigualdade econômica e o aumento da polarização

Porém, antes de abordar estes grandes temas, preciso, primeiro, contextualizar os principais conceitos fundamentais que Haidt usa para defender suas teses.

A) O elefante e o ginete

Haidt apresenta um conceito fundamental para justificar diversos de seus argumentos ao longo da obra: o conceito de o “elefante e o ginete”. De forma resumida, Haidt associa o elefante ao nosso corpo, emoções e intuições enquanto atribui o ginete ao nosso raciocínio moral.

Ao longo da história da moral, há um link direto de Platão a Kant, em que muitos defenderam a razão pura como a forma mais elevada e que tal forma deveria guiar todas as nossas ações. Porém, Haidt defende que, na prática, o ginete apenas aplica ajustes marginais ao caminho já favorecido pelo elefante e por nossas intuições.

Nosso raciocínio moral não guia nossas ações, ele serve como advogado de defesa, justificando as nossas ações (para os outros e para nós mesmos) de forma a posteriori. Ou seja, o primeiro julgamento é emocional, depois o racionalizamos.

B) “Papilas gustativas” morais

O autor cita Hume e o tratado da natureza humana para auxiliá-lo em sua explicação: A moralidade não é abstrata, ela é relativa ao sentimento ou paladar mental de cada ser em particular, da mesma maneira que as distinções de doce e amargo, quente e frio, surgem do sentimento particular de cada sentido. As percepções morais, portanto, não devem ser classificadas como de mais ou menos entendimento, e sim como sabores ou sentimentos

Em outras palavras, da mesma forma que virtudes são construções sociais, em que as virtudes ensinadas as crianças em uma cultura guerreira são diferentes daquelas ensinadas em uma cultura agrícola; a moralidade humana não é um sistema logico dedutivo, mas um paladar plural. Portanto, temos receptores universais, mas cada gene, cultura e sociedade constrói sua própria culinária moral.

C) Pensamento de grupo e instinto tribal

O terceiro ponto fundamental de Haidt é que a natureza humana é egoísta. Porém, também temos a tendência de formar e promover interesses de grupos, principalmente quando estamos em concorrência com outros grupos. Haidt define este tema como a nossa capacidade de trabalhar em múltiplos níveis, somos primatas egoístas que desejam fazer parte de algo superior e mais nobre que nós mesmos, metaforicamente somos 90% chimpanzés e 10% abelhas.

Os seres humanos são campeões mundiais da cooperação além do parentesco, fazemos isso grande parte criando sistemas de responsabilização formal e informal. A tendencia de formar grupos, é um dos ingredientes mágicos que possibilitaram os desenvolvimentos de civilizações, domínio da Terra e uma vida cada vez mais pacífica.

O ponto principal, é que a moralidade agrega e cega. Dentro dos nossos próprios grupos, em que compartilham nossas matrizes morais, criamos um ambiente forte e coeso, porém, e ao mesmo tempo, nos afastando cada vez (ou nos cegando) dos grupos que possuem “papilas gustativas” morais diferentes. Se a moralidade é intuitiva e grupal, então o conflito político não é apenas divergência de opinião, mas colisão entre matrizes morais distintas.

Ufa. Com essa base conceitual estabelecida, podemos agora explorar as conexões entre polarização e populismo.

1. Por que pessoas mais inteligentes não são mais imparciais (e podem ser piores)

Buscamos verdade ou aprovação?

Apesar de frequentemente nos considerarmos racionais, o raciocínio moral raramente constitui uma busca honesta pela verdade. Nosso raciocínio moral frequentemente busca justificar nossas ações, principalmente quando os outros não compartilham de nossa matriz moral.

Fazemos nossos primeiros julgamentos rapidamente e somos terríveis ao procurar evidências que possam refutar esses julgamentos iniciais. Ficamos obcecados com argumentos que sustentam nosso ponto de vista intuitivo (o “elefante”) e procuramos evidências que reforcem nosso viés de confirmação.

Nosso raciocínio funciona principalmente como defesa pós-hoc das intuições que já formamos. Ou seja, nossa mente funciona mais como um advogado de defesa do que um juiz imparcial; ou mais parecido com um político que angaria votos do que um cientista que busca a verdade.

Haidt argumenta que pessoas com QI mais elevado ou mais instruídas, também tendem a não procurar alternativas para seus julgamentos iniciais. Maior capacidade cognitiva não implica maior imparcialidade moral. A principal diferença é que tais pessoas conseguem mobilizar um número maior de argumentos e conexões para defender suas premissas iniciais.

2. Principais diferenças morais entre conservadores e liberais

O governo deveria fazer mais para promover o bem comum e a justiça social, mesmo que isso signifique limitar as escolhas dos indivíduos?

Haidt alega que, caso você tenha respondido “sim” para a pergunta acima, provavelmente você é um liberal; caso contrário, conservador ou libertário. Para respaldar seus argumentos sobre nossas diferenças e gatilhos emocionais, o autor constrói o conceito de alicerces morais.

Cada alicerce possui tanto uma dimensão positiva quanto uma negativa. Haidt os caracteriza da seguinte forma: i) Cuidado/Dano, ii) Justiça/Trapaça iii) Lealdade/Traição iv) Autoridade/Subversão, v) Pureza/ Degradação e vi) Liberdade/Opressão.

De forma geral, Haidt argumenta que liberais priorizam com mais intensidade os alicerces de Cuidado/Dano e Liberdade/Opressão, enquanto os conservadores tendem a apresentar uma distribuição mais ampla entre os fundamentos morais.

Fazendo a conexão entre “papilas gustativas” morais e as diferenças entre conservadores e liberais, é importante destacar que não há um sistema racional de princípios universais imutáveis e atemporais. A moralidade é muito mais intuitiva, situada e cultural. Dessa forma, a moralidade que o atrai depende dos seus “receptores de sabor” da mente, dependendo do seu nível de sistematização e empatia

3. Como algoritmos transformaram diferenças políticas em confronto identitário

Os dois temas mais importantes, perturbadores e desagregadores da vida humana são a política e a religião, porque são expressões da nossa psicologia moral subjacente

Como já explicado, temos a tendência de formar e promover interesses de grupos. Porém, não formamos grupos cegamente: sentimos empatia por quem compartilha de nossa matriz moral. Contudo, a forma como formamos grupos vem sendo alterada nos últimos anos. Com a evolução tecnológica e a super personalização dos algoritmos, as redes sociais vêm nos isolando em um casulo de pessoas que comungam da mesma opinião. Isso alimenta, cada vez mais, nosso viés de confirmação, enquanto nos cega para opiniões divergentes.

Em determinado momento, o viés de confirmação torna-se forte e se transforma em política identitária, inflamada pelos discursos populistas de “nós contra eles”, “ricos contra pobres” e justiça social, e gera o aumento de polarização e visão tribal. Ou seja, quando os laços humanos enfraquecem, a oposição passa a ser vista como ameaça moral, e a política vira uma guerra tribal. Deixamos de discutir políticas públicas e passamos a debater narrativas morais identitárias, e valores fundamentais sobre o que é justo, legitimo e desejável.

Como exemplo, o gráfico abaixo mostra a evolução da concentração ideológica dentro de cada partido nos Estados Unidos, evidencia o aumento da polarização e seu reflexo na percepção econômica. Ficam algumas perguntas: como o índice de confiança econômica pode ser tão diametralmente oposto entre partidos? A política identitária estaria se sobrepondo à percepção econômica?

Se percepção econômica já não é apenas função de dados objetivos, mas da identidade política do indivíduo, então a economia deixa de ser apenas econômica : torna-se moral.

Fonte: Gallup

Fonte: Gallup

4. O aprofundamento da desigualdade e o aumento da polarização

Outro ponto que agrava as tendências demonstradas acima, é o aumento da desigualdade. A recente onda de aumento de produtividade (pós-covid) e ganho de escala operacional, combinada à forte injeção monetária dos bancos centrais e valorização dos ativos financeiros, intensificou a desigualdade americana. De acordo com o Banco Central americano (Fed), o 1% mais ricos concentram mais de 30% do patrimônio líquido dos Estados Unidos.

Fonte: Federal Reserve

Esta dinâmica também se reflete na curva de consumo americana, em que mostra que os 10% mais ricos respondem por aproximadamente 50% do consumo total do país.

Fonte: Financial Times

Podemos analisar essa mesma tendência no gráfico abaixo: o mercado de ações decola após 2020, beneficiando a camada que possui reservas financeiras, enquanto confiança do consumidor segue deteriorando, muito em função da corrosão do poder de compra pela inflação. Assim, adicionamos mais uma variável além da percepção moral: a percepção econômica e sentimento do consumidor variam significativamente conforme o estrato da renda.

A desigualdade aumenta a sensação de injustiça, ativando fundamentos morais que reforçam a identidade grupal e, por fim, fortalecem o populismo. Movimentos populistas prosperam justamente porque oferecem narrativas morais simples, que ativam fundações emocionais profundas e reforçam identidade grupal.

Fonte: Refinity DataStream

5. Conclusão

Como aumentar a coesão social em um mundo identitário?

Estamos em um momento de tamanha polarização e tão emocionalmente conectados aos nossos vieses de confirmação que não conseguimos encontrar a um campo comum no debate político, pois já não discutimos políticas públicas, mas sim narrativas morais identitárias.

A super personalização dos algoritmos, por um lado, e a concentração de renda, por outro, agravam a tendência identitária e a polarização. Conforme mencionei no meu post sobre o Hayek e o Capitalismo de Estado americano, todas essas variáveis contribuem para o aumento do populismo, redução da institucionalidade de estrutura democrática.

Para finalizar, Haidt recomenda que saiamos da bolha das redes sociais. A melhor forma de entender outras matrizes morais é por meio da interação com mais pessoas, culturas e “paladares morais”. Vale encerrar ressaltando que a redução do conflito não decorre de mais argumentos, ou argumentos melhores, mas da compreensão das matrizes morais que os produzem.

A inspiração do post de hoje é o livro “O caminho da servidão”, de Friedrich Hayek. Começarei com um breve resumo da obra, contextualizando suas ideias principais, para depois construir conexões com o cenário atual.

Para analisar os principais temas do livro, é preciso considerar o ambiente geopolítico em que foi escrito. Hayek começou a trabalhar em “O caminho da servidão” em 1940 e o concluiu em 1944, em meio à Segunda Guerra Mundial, à forte ascensão do comunismo soviético e, de certa forma, o início da cristalização das diferenças ideológicas que levariam à Guerra Fria.

Hayek dividiu o livro em quinze capítulos breves, que abordam conceitos como Estado e controle econômico, Liberdade, Totalitarismo, Democracia, entre outros. Neste ensaio, abordarei apenas quatro capítulos:

I - Planificação e democracia

II – Planificação e o Estado de Direito

III – Por que os piores chegam ao poder

IV – O fim da verdade

I – Planificação e democracia

Principais pontos do capítulo:

A) Quanto maior for a força e controle do Estado, menor tende ser a democracia liberal. Em outras palavras, quanto mais o Estado controla recursos, mais ele passa a controlar indiretamente todas as escolhas individuais.

B) Em um Estado controlador, as escolhas individuais passam a ser moldadas. Quem controla os fins e os incentivos condiciona os meios para alcançá-los.

C) Hayek descarta o pressuposto de que “se o poder é democrático, não pode ser arbitrário”. Na visão do autor, mesmo que um governo tenha sido eleito democraticamente, não impede a arbitrariedade.

D) A democracia não é um fim, mas um meio para proteger a paz interna e liberdade individual. Como tal, não é perfeita nem infalível.

E) Não é a fonte do poder (o povo), mas sim a limitação do poder, que impede que este seja arbitrário. Ou seja, a democracia, por si só, não impede nem assegura que o poder seja arbitrário.

Como Trump vem reduzindo as estruturas democráticas do governo americano?

Os Estados Unidos (e diversas outras nações) vivem hoje um ambiente político dividido, identitário e polarizado. Nesse contexto, a política deixa de ser competição de ideias e passa a ser a eficiência da ação, por vezes ignorando os meios de implementação e focando nos resultados finais e ideológicos.

Após uma vitória eleitoral contundente e respaldado pelo apoio populacional, Trump se utiliza dessa conjuntura para exercer suas intenções por meio de ordens executivas, ampliando seu poder arbitrário. No gráfico abaixo podemos ver que o segundo mandato do Trump bate recordes em termos de número de ordens executivas:

Fonte: Goldman Sachs Investment Strategy Group

Subscribe now

II – Planificação e o Estado de Direito

Principais pontos do capítulo:

A) Um país é livre quando todas as ações do governo são regidas por normas gerais, previamente estabelecidas e conhecidas (rule of law).

B) Em um Estado de Direito, o governo deve determinar as regras, mas não controlar sua implementação. Quem controla a implementação dos meios, determina os resultados. Ou seja, os resultados serão aleatórios e de acordo com a livre competição de mercado, respeitando regras pré-estabelecidas.

C) Muitas vezes, para o Estado de Direito funcionar, a aplicação das normas, sem exceções, é mais relevante do que o próprio conteúdo das normas.

D) O Estado de Direito é previsível, impedindo intervenções ad hoc e permitindo que o indivíduo planeje sua vida e suas ações.

E) Intervenções ad hoc, para Hayek, configuram privilégios que enfraquecem o Estado de Direito.

A quebra de regras previamente estabelecidas e a redução do Estado de Direito

Somado aos pontos discutidos na seção anterior, o uso recorrente de ordens executivas não apenas reduz o processo democrático, como também rompe regras previamente estabelecidas e cria grupos privilegiados, enfraquecendo a força do Estado de Direito.

Nesse sentido, iniciativas como o teto de 10% para juros de cartão de crédito, a taxação seletiva de chips da Nvidia e a proposta de cobrança adicional de energia para grandes empresas de tecnologia (somadas à compra da Intel e à potencial intervenção no Fed, entre outros) não buscam reformar as regras do jogo para torná-lo mais competitivo ou justo.

Na prática, elas intervêm diretamente na aplicação das regras, criando vencedores e perdedores, aumentando a instabilidade jurídica, ampliando a arbitrariedade e fragilizando o processo democrático.

III – Por que os piores chegam ao poder

Principais pontos do capítulo:

A) Para Hayek, o estágio pré-ditatorial ou totalitário alimenta o culto a um líder forte, impulsionado pela insatisfação com a lentidão do processo democrático

B) Surge, assim, a demanda por um líder decidido e capaz.

C) O valor central deixa de ser o debate e passa a ser a eficiência da ação.

D) O totalitarismo não nasce de apoio amplo, mas do apoio intenso e disciplinado de um grupo menor, altamente organizado e disposto à obediência incondicional.

E) Os piores estão dispostos a ignorar os meios para atingir seus objetivos ideológicos.

A liderança do Trump e sua bandeira de America First e segurança nacional

Impulsionado pelo discurso “nós contra eles”, pela polarização americana e apoio quase incondicional de parte de seu eleitorado, Trump demostra (com a invasão da Venezuela e o projeto de anexar a Groenlândia) disposição para relativizar os meios em prol dos fins.

Para os propósitos deste ensaio, não se trata de discutir a moralidade dessas ações, mas de evidenciar a reduções dos freios democráticos. Hayek já havia identificado a tendência de líderes autocráticos a justificar ações que deveriam ser moralmente restritas em nome do interesse do grupo ou da nação.

Quando Trump afirma que a única restrição às suas ações é sua própria moral e que não vê necessidade de seguir o direito internacional, combinando esse discurso com America First e segurança nacional, a cartilha desenhada por Hayek se torna cada vez mais aderente.

Subscribe now

IV – O fim da verdade

Principais pontos do capítulo:

A) O totalitarismo não começa com a censura explícita; começa quando o Estado passa a moldar a percepção da realidade.

B) Em um regime de concorrência, a propaganda é a disputa entre ideias. Em regimes autoritários, ela se transforma em manipulação e reformulação de valores.

C) Quando o Estado domina a mídia, além do desafio de julgar o conteúdo da informação, surge a dúvida sobre sua veracidade.

D) Tudo o que possa despertar dúvidas sobre a competência do governo ou gerar descontentamento tende a ser ocultado do público.

Até que ponto os veículos de mídia estão sendo influenciados ou manipulados?

Parafraseando a Eurasia Group “A tomada de controle da Paramount pela família Ellison (aliada de Trump) deu a ela o controle da CBS; o grupo assinou um acordo para as operações do TikTok nos EUA e agora busca a CNN, da Warner Bros. Discovery. Somados ao X de Elon Musk, à Fox de Rupert Murdoch e à própria Truth Social de Trump, grande parte da mídia tradicional e das redes sociais nos Estados Unidos ficará sob o controle de atores pró-Trump.

Se as principais fontes de notícias relacionados ao governo são a Truth Social (Trump), e os controladores dos principais veículos do país são aliados do atual presidente, como o público poderá formar um juízo de valor independente sobre suas ações?

Como todos esses pontos afetam os mercados?

A grande questão que emerge é se ainda devemos tratar o dólar e as Treasuries (dívida americana) como reservas de valor absolutas. Por décadas, esses ativos foram vistos como safe havens, inclusive em crises originadas nos próprios Estados Unidos.

Com o enfraquecimento da Pax americana e a possibilidade de uma mudança na ordem global, como esses ativos tendem a se comportar?

Mostro abaixo alguns gráficos que considero importantes:

(i) aumento da compra de ouro por bancos centrais globais, especialmente após a guerra na Ucrânia e o congelamento de ativos russos

(ii) a redução do índice DXY desde o Liberation Day

(iii) o aumento do valor do ouro desde 2022

Fontes: John Authers, Kinea, Bloomberg/IIF

Finalizo essa seção com uma última pergunta: o ouro virou (ou voltou a ser) a principal reserva de valor global?

Conclusão

Muitos dos pontos levantados por Hayek ao longo de O caminho da servidão ressoam com o ambiente político atual dos Estados Unidos. Observa-se uma redução dos freios democráticos e um aumento de decisões unilaterais via ordens executivas. Além disso, Trump tem se mostrado cada vez mais imprevisível, impulsionado pelo discurso de America First e segurança nacional.

Nesse contexto, o que muitos têm chamado de “capitalismo de Estado americano” ganha contornos mais claros. Prefiro, no entanto, encerrar este ensaio com um tom marginalmente mais construtivo lembrando as palavras de Alexis Tocqueville: “the greatness of America lies not in being more enlightened than any other nation, but rather in her ability to repair her faults”.

Subscribe now

Fontes: European Council, Reuters e Valor Economico

i) O que foi assinado

O acordo firmado entre o Mercosul e a União Europeia é, na prática, um conjunto de dois pilares principais. O primeiro é o Acordo de Parceria Mercosul–UE (EMPA), que estabelece o arcabouço geral da relação, abrangendo desenvolvimento sustentável, transformação digital, cooperação política e diálogo institucional. O segundo é o Acordo de Comércio Interino (iTA), pensado como um instrumento transitório para permitir a implementação mais rápida das partes comerciais do acordo, enquanto os processos de ratificação seguem nos parlamentos nacionais.

No campo comercial, o acordo prevê a redução progressiva de tarifas, maior previsibilidade regulatória e ampliação do acesso a mercados, especialmente em bens industriais, agrícolas e serviços. Para a União Europeia, trata-se de ampliar exportações de máquinas, equipamentos, produtos químicos e bens de maior valor agregado. Para o Mercosul, o foco recai sobre produtos agrícolas, minerais, além de ganhos potenciais em investimentos e integração às cadeias globais.

Ainda assim, o acordo não entra em vigor automaticamente. Ele precisa do consentimento do Parlamento Europeu e da ratificação pelos parlamentos dos países do Mercosul.

ii) Um acordo após 25 anos e o tamanho do mercado envolvido

Segundo a Reuters, o acordo foi concluído após 25 anos de negociações, um período que por si só ilustra o grau de complexidade econômica, política e regulatória envolvido. Trata-se de um dos maiores acordos comerciais do mundo, conectando dois blocos que, juntos, representam um mercado de cerca de 700 milhões de pessoas.

Nesse contexto, a declaração oficial que acompanhou a assinatura ajuda a entender o simbolismo do momento:

“This agreement sends a very strong message to the world, it reflects a clear and deliberate choice. We choose fair trade over tariffs. We choose a productive, long term partnership over isolation. And above all, we intend to deliver real and tangible benefits to our people and our businesses.”

Mais do que comércio, o acordo foi apresentado como uma escolha política em um mundo cada vez mais marcado por tarifas, fragmentação e tensões geopolíticas.

Subscribe now

iii) Canadá, China e a ideia de uma “nova ordem mundial”

O movimento europeu ocorre em paralelo a outros realinhamentos relevantes. O primeiro-ministro do Canadá, Mark Carney, por exemplo, recentemente sinalizou uma reaproximação estratégica com a China. Em sua visita oficial, afirmou:

“Mine is the first visit of a Canadian prime minister to China in nearly a decade. The world has changed much since that last visit, and I believe that the progress that we have made in partnership set us up well for the new world order.”

A referência explícita a uma “nova ordem mundial” não é casual. Ela reflete a percepção crescente de que o sistema internacional está se reorganizando, com países buscando diversificar parceiros, reduzir dependências excessivas e construir novas alianças econômicas e estratégicas fora dos eixos tradicionais.

Nesse sentido, o acordo Mercosul–UE se insere em um movimento mais amplo de reconfiguração das relações comerciais globais, em que acordos de longo prazo voltam a ganhar relevância como instrumentos de estabilidade.

iv) O efeito dominó para o Brasil

De acordo com o Valor Econômico, o acordo com a União Europeia pode abrir caminho para novas negociações comerciais do Brasil e do Mercosul com outros países e blocos, como Canadá, Reino Unido e Japão. Após anos de relativo isolamento, o Mercosul parece voltar ao radar das grandes negociações comerciais internacionais.

Conclusão

Curiosamente, esse movimento ocorre em um contexto em que o protecionismo dos Estados Unidos ganha força. Como resumiu Marsilea Gombata, pesquisadora em relações internacionais da USP, “o protecionismo do governo Trump parece estar levando o Brasil a uma nova fase em termos de política comercial”.

Mais do que um acordo específico, o Mercosul–UE pode sinalizar uma mudança de postura: menos dependência de um único eixo e maior disposição para se integrar a uma economia global cada vez mais fragmentada, porém, ainda profundamente interdependente.

Subscribe now

1. A relevância e a necessidade dos metais raros

2. A diferença entre produção e processamento

3. A concentração e o controle chineses da cadeia

4. Por que é tão caro construir um hub de metais raros

5. Estados Unidos, influência geopolítica e potenciais investimentos internacionais (Brasil)

Fontes: CSIS, S&P Global, Visual Capitalist

1. A relevância e a necessidade dos metais raros

Os metais raros são um dos insumos mais estratégicos da economia moderna porque atravessam, ao mesmo tempo, a defesa nacional, a saúde e o cotidiano das pessoas. Em sistemas militares avançados, o uso é massivo: um caça F-35 incorpora mais de 400 kg de metais raros, enquanto um submarino da classe Virginia pode consumir aproximadamente 4.170 kg desses elementos.

Mas a dependência não se limita ao setor militar. Metais raros são essenciais em equipamentos médicos como ressonância magnética, tratamentos contra o câncer, além eletrônicos de consumo e semicondutores. Ou seja, qualquer disrupção nessa cadeia afeta simultaneamente a segurança nacional, os sistemas de saúde e produtos básicos do dia a dia.

2. A diferença entre produção e processamento

Um ponto frequentemente subestimado é a diferença entre produzir e processar metais raros. O verdadeiro gargalo está no processamento e na separação química, etapas altamente complexas, intensivas em capital, energia e know-how técnico.

Embora responda por cerca de 60% da produção global de metais raros, a China concentra aproximadamente 90% do processamento, percentual que sobe para 99% no caso dos metais raros pesados (HREEs).

3. A concentração e o controle chineses da cadeia

Mais do que uma vantagem histórica, essa posição dominante é ativamente defendida. Empresas chinesas têm demonstrado disposição de pagar prêmios muito elevados para manter o controle da cadeia. Em 2025, por exemplo, um projeto relevante na África foi adquirido com um prêmio próximo de 200% sobre o preço de mercado.

Além disso, a China já deixou claro que está disposta a usar esse controle como instrumento geopolítico. Ao longo dos últimos 15 anos, restringiu exportações em momentos-chave e, mais recentemente, impôs controles à exportação de tecnologias de processamento e separação de metais raros.

Subscribe now

4. Por que é tão caro construir um hub de metais raros

Construir hubs alternativos de processamento é extremamente caro e demorado. O transporte pode representar mais de 50% do custo total do produto final, o que torna a proximidade geográfica da matéria-prima um fator crítico de viabilidade econômica.

Além disso, a produção exige volumes gigantescos de água — em alguns casos, mais de 10 mil quilos de água por quilo de metal processado — e é altamente intensiva em energia, com o processamento podendo demandar de 9 a 13 vezes mais energia do que a simples extração.

5. Estados Unidos, influência geopolítica e potenciais investimentos internacionais (Brasil)

Os Estados Unidos também enfrentam limitações domésticas relevantes. O tempo médio entre a descoberta de uma mina e o início da produção se aproxima de 30 anos, um dos mais longos do mundo, com pouquíssimos projetos efetivamente entrando em operação nas últimas duas décadas. Diante desse horizonte, qualquer estratégia para reduzir a dependência da China tende a exigir investimentos fora do território americano, em países aliados ou parceiros estratégicos.

Nesse contexto, o Brasil surge como um candidato potencial: conta com recursos naturais abundantes, matriz energética relativamente limpa e ampla disponibilidade hídrica. Por outro lado, enfrenta entraves importantes, como insegurança jurídica e um processo de licenciamento que pode levar mais de uma década até a operação efetiva. Sem endereçar essas questões institucionais, o país corre o risco de perder uma oportunidade estratégica justamente em um momento em que o Ocidente busca alternativas urgentes à concentração chinesa.

Conclusão

No fim, a discussão sobre terras raras vai muito além de mineração ou tecnologia. Trata-se do controle das cadeias produtivas críticas em um mundo cada vez mais fragmentado, onde eficiência econômica e segurança estratégica nem sempre caminham juntas. Reduzir dependências, diversificar fornecedores e construir capacidade industrial fora dos centros tradicionais exigirá tempo, capital e coordenação entre governos e setor privado.

Mais do que uma disputa pontual, o tema das terras raras ilustra um movimento estrutural: cadeias globais deixam de ser apenas um problema econômico e passam a ocupar o centro das decisões geopolíticas. As escolhas feitas hoje — sobre onde investir, com quem se alinhar e quais riscos aceitar — tendem a definir não apenas a competitividade industrial, mas também o equilíbrio de poder nos próximos anos.

Este espaço nasceu de uma necessidade pessoal: organizar leituras e reflexões em um mundo com excesso de informação e pouco tempo para separar o que é ruído do que realmente importa. Entre macroeconomia, geopolítica e mercados, muita coisa acontece todos os dias — mas pouca coisa de fato muda o regime.

A ideia aqui não é cobrir notícias nem oferecer respostas definitivas. É tentar entender o que é estrutural, o que está mudando na margem e onde o consenso pode estar confortável demais.

Pretendo publicar com frequência semanal ou quinzenal, reunindo leituras que me pareceram relevantes e algumas reflexões sobre como elas se conectam. Menos manchete, mais contexto. Menos reação, mais interpretação.

Se esse tipo de exercício também te interessa, fique à vontade para acompanhar.

Subscribe now

Ray Dalio - 2025

Ray Dalio - 2025

Um dos pontos mais interessantes do comentário recente de Ray Dalio é trazer para o dia a dia algo que costuma passar despercebido: o dinheiro perdeu valor. Em 2025, o dólar caiu cerca de 39% quando comparado ao ouro. Outras moedas tiveram desempenho semelhante — euro, franco suíço e renminbi também se desvalorizaram de forma relevante frente ao metal. Mesmo ativos que “subiram” em termos nominais contam outra história quando olhados por esse ângulo. O S&P 500, por exemplo, avançou cerca de 18% em dólares, mas caiu aproximadamente 28% quando medido em ouro. Ou seja, muita coisa parece ter se valorizado, mas ficou para trás em relação ao ouro, uma das principais — senão a principal — reservas de valor.

Dalio conecta esse movimento a uma mudança mais ampla no cenário global. O mundo vem migrando de um arranjo multilateral, baseado em regras e instituições, para um ambiente cada vez mais unilateral, no qual países agem de forma mais direta em defesa de seus próprios interesses. Esse deslocamento aumenta tensões geopolíticas, incentiva protecionismo e leva governos a expandirem gastos militares — quase sempre financiados por mais dívida. O resultado é um ambiente em que moedas tendem a se enfraquecer enquanto ativos reais e estratégicos ganham relevância, reforçando a necessidade de olhar além dos preços nominais.

Top Risks 2026 - Eurasia Group

Top Risks 2026

Dos dez riscos mapeados pela Eurasia Group, um dos que mais ajuda a entender o pano de fundo deste post é a tentativa de retomada da Doutrina Monroe em sua versão mais literal. Em essência, trata-se da visão de que os Estados Unidos precisam voltar a exercer influência real e direta sobre o hemisfério ocidental, não apenas no plano diplomático, mas também econômico e estratégico. Dentro dessa lógica, o endurecimento em relação à Venezuela deixa de ser um caso isolado, e a possibilidade de Cuba se tornar o próximo alvo — ainda que por meio de sanções e pressão econômica, e não ação militar — passa a fazer sentido como parte de uma estratégia mais ampla de reassertar poder no “quintal” americano.

Outro risco importante é o timing político. Existe uma probabilidade elevada de que Trump acelere suas políticas mais controversas antes das eleições de meio de mandato, justamente porque o apoio político tende a se deteriorar ao longo do tempo. Se o midterm representar perda de força no Congresso, o espaço de manobra diminui. Isso cria incentivos claros para um comportamento mais agressivo no curto prazo, com decisões abruptas, menor disposição para consensos institucionais e maior uso do poder executivo — tornando 2026 um ano potencialmente mais conturbado do ponto de vista político.

Por fim, esse ambiente se conecta diretamente ao avanço do state capitalism com características americanas. Na história moderna dos Estados Unidos, raramente — ou talvez nunca — um presidente escolheu de forma tão explícita vencedores e perdedores, seja por meio de participações acionárias do governo, controle de preços, direcionamento de investimentos estrangeiros ou concessões regulatórias seletivas. Essa lógica, além de distorcer a alocação de capital, enfraquece o rule of law, torna o ambiente jurídico imprevisível e aumenta o risco de investir nos EUA.

Subscribe now

Latin America’s Revolution of the Right - Foreign Affairs

Latin America’s Revolution of the Right

O ponto central do artigo Latin America’s Revolution of the Right, da Foreign Affairs, é que a guinada à direita na América Latina não parece ser apenas um movimento cíclico, mas sim uma mudança mais estrutural. Segundo o Latinobarómetro, em 2024 a proporção de latino-americanos que se identificam como de direita atingiu o maior nível em mais de duas décadas. Esse movimento tem um motor claro: segurança pública. O avanço do narcotráfico — com a produção de cocaína triplicando na última década — elevou a violência a níveis inéditos, tornando o crime a principal preocupação do eleitorado em diversos países, uma pauta na qual a esquerda historicamente encontra mais dificuldade em oferecer respostas críveis.

Nesse contexto, medidas duras e até controversas passaram a receber apoio popular significativo. Operações policiais violentas, como as realizadas no Rio de Janeiro, foram aprovadas inclusive entre moradores de favelas, e o caso de Nayib Bukele é ainda mais emblemático: desde 2019, os homicídios em El Salvador caíram mais de 90%, levando o presidente a níveis de popularidade inéditos na região. O artigo mostra que, diante do aumento da criminalidade, parte expressiva da população está disposta a relativizar garantias institucionais em troca de maior sensação de segurança — inclusive em países com tradições democráticas mais sólidas, como o Chile.

Ao mesmo tempo, a direita latino-americana também mudou. Ela se tornou mais populista, mantendo — e em alguns casos ampliando — programas de transferência de renda e benefícios sociais, enquanto desloca o debate para o ponto fraco da esquerda: segurança pública. Esse arranjo ajuda a explicar sua competitividade eleitoral recente. O risco, porém, vem de fora. O artigo argumenta que Donald Trump pode se tornar um fator desestabilizador para a direita regional, já que algumas de suas ações geraram reação nacionalista e perda de apoio a aliados, como no Brasil. Por outro lado, quando bem alinhado — como no caso de Javier Milei — o ambiente externo pode amplificar expectativas pró-mercado e atrair capital, mostrando que o impacto de Trump será menos ideológico e mais contingente ao contexto político local.