Elite Overproduction and the Coming AI Disruption

The Pattern

Every empire follows the same arc. Growth creates wealth. Wealth concentrates. The elite class expands faster than elite positions. Competition intensifies. Factions form. Infighting consumes what expansion once produced. Then something breaks through from the outside—a group with stronger cohesion, forged in conditions the incumbents can no longer imagine.

This pattern is so consistent that sociologists have given it a name: elite overproduction. Peter Turchin, building on the work of Ibn Khaldun and Jack Goldstone, has documented it across civilizations. Rome. The Qing Dynasty. Pre-revolutionary France. Pre-Civil War America. The mechanism is always the same.

What’s remarkable is how precisely this pattern maps to big tech companies in 2025. The names change—SVPs instead of nobles, “alignment” instead of court intrigue, total compensation instead of land holdings—but the dynamics are identical.

And AI is the black swan that will expose it all.

The Mechanism: From Growth to Stagnation

Phase 1: Growth and Cohesion

Ibn Khaldun called it asabiyyah—the social solidarity that binds a group together. A startup has it. A scrappy team solving a hard problem has it. Google in 2004 had it. The people are aligned. The mission is clear. Resources go toward the work, not toward fighting over the work.

“Political power does not originate in wealth, ideology, or formal institutions. It originates in cohesive groups capable of coordinated action, usually forged under conditions of scarcity, danger, and struggle.” — Ibn Khaldun, Muqaddimah

This cohesion enables rapid growth. The early team captures value because they’re focused on creating it, not on extracting it from each other.

Phase 2: Wealth Concentration and Elite Expansion

Success breeds wealth. Wealth breeds hierarchy. Hierarchy breeds positions. Positions breed aspirants.

Turchin describes a “wealth pump”—a mechanism that transfers resources from productive workers to the elite class. In big tech, this looks like:

• Stock-based compensation that compounds for early employees
• Leveling systems that create ever more senior titles
• Promotion velocity that outpaces actual scope expansion
• Credential inflation (everyone needs a senior title to feel valued)

The number of VPs, Senior VPs, Directors, and Senior Directors expands. But the number of meaningful decisions doesn’t expand at the same rate. You can only have so many people setting strategy. You can only have so many budget owners.

“When there are more individuals aspiring to elite positions than there are available positions, this creates intense competition among elites.” — Peter Turchin

Phase 3: Intra-Elite Competition

This is where it gets ugly. When there are more elite aspirants than elite positions, the competition turns inward.

In historical empires, this meant:

• Court intrigue and factional warfare
• Nobles fighting nobles while the peasants watched
• Resources devoted to internal conflict rather than external threats
• Conspicuous consumption to signal and maintain status

In big tech, this looks eerily similar:

The “alignment” meetings aren’t about alignment. They’re about power. Two VPs send their organizations into battle—the outcome determines who gets the budget, the headcount, the visibility. This is medieval warfare with Slack channels.

“Aristocracies do not last. They decay not only in quantity, but also in quality.” — Vilfredo Pareto

Phase 4: Stagnation and Hubris

When elites spend more energy fighting each other than fighting external threats, the organization ossifies—innovation slows, and the institutional immune system rejects anything that might upset the power balance.

This creates hubris—the outrageous arrogance that comes with uncontested power.

“Herodotus knew that empires do not fall because of anonymous social and economic forces. In fact, the decline of the Persian Empire began when it was economically the super-power of its day.” — Peter Turchin

Big tech in 2024 exhibited classic hubris symptoms:

• “AI won’t change how we work”
• “We have the distribution”
• “Startups can’t compete with our talent”
• “We’ll just acquire anything threatening”

Meanwhile, the actual work—the building, the innovation, the customer value—gets buried under layers of process, approval, and political theater.

Phase 5: Vulnerability and Disruption

Here’s the key insight: empires don’t fall because they’re weak. They fall because their strength is misallocated.

Rome at its height could have easily repelled the northern tribes. But by the time those tribes arrived, Roman elites were too busy fighting each other to mount a coordinated defense.

“At its height, it could easily have overcome invading northern tribes; but later, once its elites began being more concerned with fighting among themselves than fighting common enemies, the Goths and Vandals found easy pickings.” — Peter Turchin

This is big tech’s vulnerability. The talent is there. The resources are there. The technology is there. But the coordination is broken. The asabiyyah is gone.

And something is coming.

AI as the Black Swan

Every empire’s collapse has a catalyst—a force that exposes the underlying fragility. For Rome, it was the migrations. For the Qing Dynasty, it was the Taiping Rebellion, led by “disgruntled well-educated young men” who couldn’t find positions worthy of their credentials. For pre-revolutionary France, it was a fiscal crisis that forced the elites to confront their own dysfunction.

For big tech, it’s AI.

Not because AI is inherently destructive. Because AI changes the terms of competition in ways that ossified organizations cannot adapt to.

The Economics Shift

AI inverted the software development cost structure: the expensive part (building) became cheap, while the cheap part (knowing what to build) became expensive.

This matters because big tech’s organizational design assumes the old economics:

• Large teams to build complex systems → Now a small team with AI can match output
• Specialists for each function → Now generalists with AI can span domains
• Process to coordinate hundreds of people → Now unnecessary overhead

The VPs fighting over headcount are fighting over the wrong resource. The Directors getting “alignment” are aligning on obsolete assumptions. The entire power structure is optimized for constraints that no longer bind.

The Talent Exodus

Here’s what the elite overproduction researchers observed about empire collapse: the talented elites leave first.

“If the governing elite does not find ways to assimilate the exceptional individuals who come to the front in the subject classes, an imbalance is created in the body politic.” — Vilfredo Pareto

In big tech, the exceptional ICs—the ones who actually build things—are the first to leave. They see the dysfunction most clearly, and they have the most options. And now, for the first time, they have leverage that doesn’t require scale.

A senior engineer who spent years navigating political theater can now:

• Build in weeks what would take their org months
• Serve a niche that their employer ignored
• Work with a cohesive team instead of a feudal hierarchy

The talent that made big tech valuable is walking out the door. What remains is the political class—the courtiers who are expert at the game of alignment but haven’t shipped anything in years.

The Coordination Advantage Flips

Big tech’s historical advantage was coordination. Thousands of people working toward common goals. This justified the overhead. Yes, there was politics, but the scale effects outweighed the friction.

AI flips this. When small teams can produce what large teams used to, coordination becomes a liability, not an asset. Every VP in the approval chain is latency. Every alignment meeting is delay. Every political battle is energy not spent on customers.

The companies that will win aren’t the ones with the most resources. They’re the ones with the strongest asabiyyah—the cohesive groups capable of coordinated action, unencumbered by internal warfare.

Historical Precedents: The Succession Pattern

When empires fall, new powers rise. The succession pattern is instructive.

Persia → Greece

The Achaemenid Persian Empire was the largest the world had seen. Wealthy beyond measure. But Herodotus noted the hubris—the arrogance that came with uncontested dominance.

Alexander’s Macedonians weren’t more numerous or wealthier. They had stronger cohesion. A small, unified force defeated an empire fragmented by internal politics.

Rome → Medieval Europe

Roman decline wasn’t a single event—it was centuries of elite infighting punctuated by external pressures. The “barbarian” tribes that eventually prevailed weren’t more sophisticated. They were more cohesive, carrying the asabiyyah that Rome had lost.

Spain → Dutch → British

Spain’s empire fell to internal dysfunction: weak kings, court intrigue, overextension, and an economy dependent on colonial extraction rather than productive innovation.

The Dutch rose because they were a cohesive commercial nation, forged through struggle against Spanish hegemony. Their advantage wasn’t size—it was coordination and focus.

Britain eventually succeeded the Dutch not through superior culture, but through better geographic position and larger population. But they maintained it through continuous adaptation—something the ossified Spanish court couldn’t manage.

The Pattern Applied to Tech

The pattern suggests that big tech’s successors won’t be bigger. They’ll be more cohesive. They’ll be:

• Small teams with shared purpose (asabiyyah)
• Forged through constraint (the struggle of building something new)
• Unencumbered by political overhead (no VPs fighting for budget)
• Focused on customers, not internal games

This is what AI enables. Not through technology alone, but through economics. When the cost of building drops, the advantage shifts from scale to cohesion.

What This Means for Strategy

For Big Tech Refugees

If you’re inside a large tech company watching the political theater, you have options your predecessors didn’t.

The talented nobility of falling empires had to choose: stay and play the game, or leave and build something new. Many chose the latter. The most capable Romans became the founders of medieval institutions. The most capable Spaniards built new ventures in the colonies.

Today’s equivalent: take your skills, find a cohesive team, build something for a specific audience that your employer ignored. The economics favor you.

For New Builders

The opportunity isn’t to replicate big tech at smaller scale. It’s to not replicate the dysfunction.

• Maintain asabiyyah: Cohesion is your advantage. Don’t hire faster than you can acculturate. Don’t create hierarchy before you need it.
• Encode taste early: Your judgment is your competitive advantage. If you don’t encode it, you’ll lose it as you grow.
• Stay focused on customers: The moment internal politics becomes more interesting than customer problems, you’ve started the decline.

For Organizations Trying to Adapt

The historical record isn’t hopeful here. Empires rarely reform themselves. The elite class that benefits from the status quo will resist changes that threaten their position.

But there are exceptions. Turchin notes that the New Deal generation successfully “turned off the wealth pump” for a time. Deliberate intervention can reset the dynamics—but it requires:

• Acknowledging the dysfunction (most won’t)
• Accepting that current elites will lose status (they’ll resist)
• Creating new sources of asabiyyah (hard to manufacture)

For most large organizations, the honest answer is: the talent will leave, the organization will decline, and something new will emerge. The question is whether you’re building the new thing or defending the old one.

Connecting to Our Theses

This historical analysis reinforces several beliefs we’ve articulated:

Taste as Infrastructure

Big tech can’t encode taste because the organization is too fragmented. Whose taste? Which VP’s judgment? The politics prevent coherence.

Small, cohesive teams can encode taste. They have a shared point of view. They can make opinionated choices. This becomes their differentiation.

“AI without encoded taste is commodity. Encoded taste compounds into durable leverage.”

Software Inversion

The constraint shifted from “can we build” to “do we know what to build.” Big tech’s org structure assumes the old constraint. All those engineers, all those approval processes, all those coordination mechanisms—optimized for a world where building was hard.

Now building is cheap. Knowing what to build is expensive. And that knowledge lives in small, focused teams—not in 50-person alignment meetings.

Factory Thinking

Big tech became feature factories, not capability factories. They optimized for output, not for building the infrastructure that produces output.

The companies that will win are building factories:

• Factories for encoding judgment
• Factories for deploying taste at scale
• Factories for compounding capabilities

The Ratio Shift

Human attention is fixed. AI capability is expanding. The organizations that add more humans (more VPs, more Directors, more political overhead) are moving in the wrong direction.

The winning organizations will have fewer humans with more leverage—each one supported by AI infrastructure that embodies encoded taste.

Niche-Making Economics

Big tech tried to be everything to everyone. That’s why they needed massive coordination. That’s why they developed political complexity.

The future is opinionated products for specific audiences. You don’t need a feudal court to serve a thousand true customers perfectly. You need a cohesive team with clear taste.

The Historical Moment

We’re at an inflection point that occurs rarely in history. The economic terms of competition are shifting. The incumbents are optimized for a world that’s disappearing. New entrants have structural advantages the incumbents can’t replicate.

This isn’t a technology story. It’s a sociological story that happens to involve technology. The same dynamics that drove the fall of Rome, the decline of Spain, and the end of the Qing Dynasty are playing out in corporate form.

The question isn’t whether disruption will occur. The historical pattern is too consistent. The question is who builds what comes next.

History suggests it won’t be the organizations currently fighting over alignment. It will be the cohesive teams—forged in constraint, focused on customers, unencumbered by political overhead—who build the tools that make the old way obsolete.

Ibn Khaldun wrote that “the term of life of a dynasty does not normally exceed three generations.” By that measure, big tech—born in the 1990s, matured in the 2000s, ossified in the 2020s—is right on schedule.

The Goths and Vandals are at the gate. They have MacBooks instead of swords, and they’re building things instead of conquering them. But the dynamic is the same.

Asabiyyah wins. It always does.

References

• Turchin, Peter. End Times: Elites, Counter-Elites, and the Path of Political Disintegration
• Ibn Khaldun. The Muqaddimah: An Introduction to History
• Pareto, Vilfredo. The Rise and Fall of Elites
• Goldstone, Jack. Revolution and Rebellion in the Early Modern World
• Herodotus. Histories

This analysis was developed for Goose Group as part of our ongoing work understanding the dynamics of AI-driven disruption. We believe that understanding historical patterns helps us—and our customers—navigate the current moment more effectively.

← All analysis