Author:David Mattin

Translation: Shenchao TechFlow

Guide by Shenchao: As the entire industry panics over Citrini Research's depiction of a "global economic collapse triggered by AI in 2028," tech thinker David Mattin stands out with a completely different interpretation. He believes we are undergoing a "global intelligence transformation," where old economic indicators (such as GDP and unemployment rate) are becoming obsolete. This article delves into the implications when intelligence becomes as cheap and abundant as air; although income may suffer, costs will collapse even faster, ushering in a new era driven by "intelligence output per unit energy." This is not only a crisis, but a radical evolution towards a "post-human economy."

The full text is as follows:

Everyone is talking about Citrini Research's paper, "The 2028 Global Intelligence Crisis." It is a brilliant thought experiment: a speculative report from June 2028 envisioning a scenario where artificial intelligence (AI) triggers a chain economic collapse.

The following content will serve as a response to that article. You can view it as a creation consistent with the original spirit of Citrini: a speculative "reverse scenario." It seeks new ways of observation rather than claiming to have all the answers (no one does). This article draws from years of research and analysis published by Raoul Pal and me in the Global Macro Investor and our jointly operated tech-focused research service, The Exponentialist.

Citrini Research's paper has garnered significant attention, and rightly so. It is a cleverly conceived thought experiment: a speculative briefing from June 2028, rehearsing a chain economic meltdown triggered by AI. The S&P 500 index drops by 38%. Unemployment rises to 10.2%. High-quality mortgages collapse. The private credit complex is shattered by a series of bets on white-collar productivity growth.

This scenario is logically self-consistent, and the study of financial mechanisms is extremely detailed, with a provocative core argument—that an abundance of intelligence destroys the consumer economy it was supposed to enhance. Some of its contents will likely prove prescient. There is indeed real turmoil ahead, even extreme crises. The transition to an abundant intelligence era will be anything but smooth.

For over five years, I have immersed myself in this way of thinking. I have been constructing frameworks to understand what happens when intelligence becomes abundant, AI-energy flywheels begin to spin, and we transition from a human-centric economy to something entirely new. In related articles I have written, I have described this as a shift towards a fundamentally new economic system: a form of "post-human economics." Based on this perspective, I want to provide a thoughtful response to Citrini's argument—grounded in my years of analysis—and I have reached a completely different conclusion.

Citrini's argument posits that abundant intelligence destroys the income side of the economy—wages, jobs, consumer spending—triggering a financial crisis. My argument is that abundant intelligence simultaneously destroys the cost side of the economy even faster. When the prices of goods and services collapse along with wages, you are not facing a crisis. You are undergoing a transition to a completely new system; in this system, all old norms, rules, and measures become incoherent.

So, what is the core mistake in Citrini's article? Their analysis measures the "post-human economy" using the instruments of the "human economy." Then, they misinterpret the disordered readings of the instrument as signs of systemic collapse.

No one has a crystal ball, and no one has all the answers. We are all piecing together a seven-dimensional puzzle that no one can fully comprehend. But I believe that despite its sophistication, Citrini's article may have made a profound and enlightening mistake, and my own work points towards this.

My timeframe is also longer than that of Citrini. Their scenario unfolds in two years. I am observing over a span of ten to twenty years. I acknowledge that serious turmoil may lie ahead: a "Fourth Turning"-style moment of chaos, social unrest, and institutional breakdown. Some version of what they describe may indeed come to pass. But my argument is that the broader forces of AI and the "Exponential Age" can ultimately carry us into a whole new economy. An economy that works well in many ways, one that is better than anything we currently know.

Faulty Metrics

This is the core argument I wish to make; if I am right, it will reconstruct everything.

Every data point used to construct the arguments in Citrini's article—10.2% unemployment, a 38% drop in the S&P 500, skyrocketing mortgage delinquency rates in San Francisco, stagnation in money circulation velocity—are all priced in the old system. Every metric originates from the economy we have been living in. That economy revolves around human labor input, conditions of material scarcity, and using GDP as a scoreboard.

The authors of this article look at these readings and see disaster, which is understandable. But what if these indicators do not record the death of the economy? What if they record the death of an "economic measurement framework" that can no longer describe the reality of what is happening?

Let's think from another angle. At the core of Citrini's article is a powerful concept: "Ghost GDP." This refers to outputs that appear in national accounts but never circulate in the real economy. They regard it as evidence of dysfunction. But I would completely reverse this perspective. Ghost GDP is not a bug, but a signal. It tells us that GDP itself, as a meaningful measure of the current situation, is collapsing. The instrument has failed, yet Citrini treats the readings of the failed instrument as the real condition of the patient.

In my research on post-human economics, I have argued that as we transition to an economy based on automation inputs and extreme abundance, GDP becomes incoherent. It cannot capture an economy where the costs of many goods and services are approaching zero—though this decline is uneven across different sectors. It cannot capture the substantial elevation in human welfare when intelligence is abundant and nearly free. Nor can it capture the emergence of "Autonomous Economic Activity"—i.e., AI trading with other AIs—completely independent of the human labor market.

In a post-human economy, GDP is not a coherent measure of anything. So, what metrics should we observe?

Intelligence Output per Unit Energy

This is my answer; this idea is at the core of my thinking about the future post-human economy.

In the coming economy, the most coherent metric for measuring prosperity is intelligence output per unit energy. How efficiently does our civilization convert energy into useful intelligence?

This is the metric that addresses the core paradox of the Citrini scenario. For just as their scenario shows GDP shrinking, the S&P index falling, and unemployment soaring, intelligence output per unit energy is skyrocketing.

Think about what is driving the crisis predicted by Citrini. AI models are becoming increasingly powerful, the cost of computing is declining, and inference costs are falling through the floor. Energy systems managed by AI are becoming increasingly efficient. Every force—the very ones that are destroying the old economic indicators—are simultaneously propelling "intelligence output per unit energy" to new heights.

This is the key insight: there are two lines on the graph. One line—GDP, employment, consumer spending—is declining; the other line—intelligence output per unit energy—is rising exponentially. Citrini's article only focuses on the declining line and concludes that we are in a crisis. My position is that the rising line is the true signal, while the falling signals are merely noise from the death of the old system.

In a world of extreme intelligence abundance, everything is downstream of better and more abundant intelligence. Scientific breakthroughs, new materials, advanced medicine, cheaper energy, better infrastructure, more efficient manufacturing—all come from the same source: our capacity to convert energy into intelligence is relentlessly improving.

Citrini's article sees the GPU clusters in North Dakota and might say: that machine just destroyed 10,000 white-collar jobs in Manhattan. I look at the same GPU clusters and say: that machine just collapsed the costs of drug development, materials science, legal services, education, energy management, and software development. Both observations are true, but the article only focuses on the income side of the balance sheet while nearly ignoring the expense side.

And that is the deeper error.

Radical Prosperity

Yes, output is decoupling from the labor market. Citrini is right on this point. But the same forces that destroy wages are also destroying costs. When AI drives the price of legal services towards near zero, you no longer need a $180,000 salary to hire a lawyer; when AI collapses the costs of medical diagnosis, you do not need expensive health insurance to get a diagnosis. When coding agents make software nearly free, the $500,000 annual SaaS renewal costs worrying Citrini are not just a hassle for the vendors—they are huge savings for the buyers.

Looking through the lens of GDP, this appears as a collapse of the consumer economy; but from another perspective, this is the birth of deflationary prosperity. It is wealth brought about by abundance. Even if nominal incomes decline, real purchasing power is exploding. The ability of ordinary people to acquire is surging in ways that traditional metrics cannot capture.

If a person earns $50,000, but in their world, AI has driven the costs of healthcare, education, legal consulting, financial planning, software, entertainment, and creative services to near zero, are they richer or poorer than the person earning $180,000 in 2024?

Citrini's paper never considers this. It tracks the decline in wages but does not track the simultaneous decline in "cost of living expenses."

I can hear some readers shouting at me. I am not naive. The costs of some essential goods and services will not drop quickly, or might not drop at all, like housing, physical food, and (at least for a time) energy. This process will be extremely uneven. Some areas will see costs collapse within years, while others may take a decade or longer. This transition will be painful for many, and this is a key social reality we must face, with depths beyond the scope of this article, but I have written about this elsewhere. I have written about the "sharp turns" ahead and warned that the "Fourth Turning" moment is likely to arrive. There will be social unrest and political upheaval, and I do not dispute this.

The Foundation Layer Flywheel: The True Brake Mechanism

But Citrini's scenario describes this transition as a one-way spiral towards destruction. They assert there are no natural brake mechanisms, and the displacement loop has no bottom line.

I disagree. The brake mechanism is abundance itself.

This leads to what I call the Foundation Layer Flywheel.

As early as 2023, I wrote about the profound symbiotic relationship between AI and clean energy. AI requires vast amounts of energy, but AI is also the only technology capable of managing the extremely complex, distributed energy systems we are building. More AI unlocks more energy, and more energy powers more AI. A cyclical process.

This flywheel is the foundation of the entire exponential age. It underpins everything happening above. That’s also why there is a natural brake in Citrini's alternative spiral—something they did not consider in their model.

As intelligence output per unit energy rises, the flywheel spins faster. Cheaper, more abundant AI makes energy systems smarter; smarter energy systems provide cheaper energy; cheaper energy makes AI cheaper. Cheaper AI then permeates everything downstream: cheaper materials science, cheaper manufacturing, cheaper healthcare, cheaper infrastructure.

Citrini's article envisions a negative feedback loop: AI destroys jobs -> unemployed workers consume less -> companies buy more AI -> repeat, with no natural brake.

But there is also a parallel positive feedback loop, at least equally powerful: AI becomes smarter -> energy becomes cheaper -> intelligence output per unit energy rises -> all costs downstream of intelligence decline -> material conditions of living improve even as nominal GDP shrinks.

Which loop will dominate? That is the question. In my view, the positive loop is supported by the laws of physics. It is driven by the exponential improvement in energy-to-intelligence conversion—a curve that has steepened over the years, showing no signs of slowing down. In contrast, the negative loop is driven by institutional and political inertia: slow-moving mortgage markets, fiscal policy, and labor market adjustments. These are real and bring genuine pain, but they are not immutable natural laws. They are human constructions that humans can change.

AI and Robotics as Part of Demographics

There is one more point that Citrini's article completely overlooks, which is one of the most significant macro forces of our time.

Demographics.

Developed countries are reducing their labor forces. The working-age populations in the U.S., Europe, Japan, South Korea, and China are sharply declining. This is what I often refer to as the demographic doomsday cycle. Fewer babies, longer lifespans, and the unprecedented height of the population pyramid throughout human history.

As Raoul has long pointed out, the golden rule is: GDP growth = population growth + productivity growth + debt growth. Population growth has disappeared. It has been gone for some time. This means the only way to keep the GDP game going is to increase debt. We are borrowing tomorrow's money to keep today’s party going.

Now think about what happens when AI and humanoid robots enter this environment. Citrini's article describes the arrival of machine intelligence as an invasion of a healthy labor market. AI storms in, displacing millions of workers.

This is the economy emerging on the other side of the "singularity." It is not a desolate zone filled with massive unemployment, but a world where the old economy has been turned into fertilizer to nurture some new, strange, and in many ways more abundant thing.

But that’s not the case. AI is entering a world that urgently needs it. We are short-handed. The working-age population in the global North is rapidly decreasing; without AI and robots, GDP growth would head toward structural decline anyway.

Kevin Kelly refers to what is about to happen as a "handoff." With the peak and decline of the human population, billions of AI agents and tens of millions of humanoids will flood in to fill this void. We are handing the economy over to non-human actors.

This does not eliminate the pain of individual transitions. Those who truly lose real jobs face real difficulties, and we need to confront this. But from a macro perspective, AI and robotics are not replacing workers but are filling a demographic void that is about to consume the entire economy.

The Citrini scenario envisions a world where AI destroys the job market, and no one can find work. But what if by 2028, the reality resembles this: AI and humanoids fill millions of vacancies created by a labor shortage, while humans displaced from knowledge jobs move on—though painful, and with support—to the emerging economies I am about to describe?

Human Residue

Because this is what Citrini’s article never considered. As the old economy contracts, a new economy is self-motivating from the grassroots.

I have written about the rise of independent industrialists. Sam Altman has mentioned this billion-dollar company run by one person. In some areas, AI tools and agents allow a high-productivity individual to produce what used to require hundreds of employees. We will see millions of such new economic participants—independent operators and micro-teams managing large AI agents—creating tremendous value in ways that the old economic framework cannot foresee.

Anthropic’s insights into how people use Claude reveal the contours of this future. Software development. Consulting. Financial services. Marketing. Content creation. In every domain, highly capable people wielding AI are increasingly becoming one-person enterprises. This is new economic activity. And much of this will happen outside the structures monitored by Citrini's work.

But a deeper transformation is underway. As machine intelligence takes on all intellectual work—coding, legal documents, financial analysis, data processing—the economic value migrates up Maslow's hierarchy to levels that only humans can provide.

I call it "human residue." The role that requires a person to be a person in value creation. Rather, it is the attention, empathy, and recognition of another who truly sees you. It is the art and narratives from real, lived experiences. It is the consultant who helps you through the stressful process of moving, the guide who helps you navigate a life crisis, the community builder who creates a place where you feel a sense of belonging.

What remains in scarcity after AI takes over all clerical work? Emotion. Connection. Meaning. Around this irreducible human output, a vast new economy will form. This will bring tremendous value. But this will not be reflected in GDP, nor captured by the metrics tracked by Citrini’s article.

This is the economy that emerges on the other side of the singularity. Not a wasteland of mass unemployment. But one where the old economy has been composted to nourish a new, strange, and more abundant world in many ways.

System Transition

Let’s integrate all of this.

Citrini’s article poses a core question: What happens when the scarce input (intelligence) becomes abundant?

This is a very correct question. Throughout modern economic history, human intelligence has been the kind of scarce input that commands a premium. They assert that this premium is dissipating, which is also true. In an increasing number of tasks, machine intelligence has become a competent and rapidly evolving substitute for human intelligence. On this point, we agree.

But Citrini concludes that the dissolution of the human intelligence premium represents a "crisis." And I see it as "transformation." They are fixated on the process of the caterpillar dissolving, screaming that this creature is dying. In a sense, they are not wrong—the caterpillar is indeed dying. But something else is forming inside the chrysalis.

What is forming is a post-human economy. In this economy, intelligence is no longer scarce but is as abundant as air. In this economy, the costs of knowledge work and ultimately many material productions will approach zero—this will not happen overnight, and will not be evenly distributed across sectors, but the process is relentless. In this economy, the fundamental measure of prosperity is no longer how much nominal economic output we have produced, but how efficiently we convert energy into intelligence. In this economy, the value exchanged among humans will migrate from intellectual labor to deeper places: empathy, meaning, connection, creativity, and the pure experience of living alongside other conscious beings.

We are not heading toward a "global intelligence crisis," but moving toward a "global intelligence transformation." We are entering a new economic system, one that we are all struggling to comprehend. Yes, the period of transition will be filled with bumps, and may even involve severe upheaval. There will be chaos, pain, and political turbulence. The "Fourth Turning" is likely a real phenomenon. Some scenarios depicted by Citrini—unemployment, the collapse of the SaaS industry, zero friction—are likely on the way and may arrive faster than most people envision.

But viewed from the longer time span I am observing—ten to twenty years, not just two years—the conclusions they draw begin to falter. A recession rivaling the global financial crisis (GFC), with declines of 57% and no natural brake mechanism? This conclusion entirely hinges on an assumption: that those old metrics can still reflect the truth of the system.

I do not believe they can anymore. There will be real pain, but this pain is a feature of the transformation process, not evidence that the destination is doomed to disaster.

There are two lines on the graph:

• The GDP is declining.
• The intelligence output per unit energy is rising.

One of these lines is a real signal, while the other is merely noise from a dying measurement system.

If we want to understand everything happening around us, we need to ensure we are keeping an eye on both of these lines simultaneously.