The AI Feedback Loop: How Agentic AI Could Trigger a Two-Year Economic Collapse

In a provocative new analysis that’s sending shockwaves through Silicon Valley and Wall Street alike, Citrini Research has outlined a chilling scenario where agentic AI systems could trigger a self-reinforcing economic death spiral within just two years.

The scenario, published Sunday, paints a dystopian picture of 2026 America: unemployment has doubled from current levels, the S&P 500 has plummeted by over 35%, and the once-booming tech sector lies in ruins. But unlike traditional economic collapse narratives, this one isn’t triggered by war, pandemic, or financial crisis—it’s set in motion by the very AI systems we’re racing to deploy.

The Perfect Storm of Automation

The mechanism Citrini describes is deceptively simple yet terrifying in its implications. As AI agents become capable of handling complex white-collar tasks—from data analysis to client communications to strategic planning—companies begin replacing human workers en masse. These displaced workers, now unemployed or underemployed, dramatically reduce their spending. This creates margin pressure on businesses, which respond by accelerating their AI adoption to cut costs further. The cycle repeats, each iteration more destructive than the last.

“It was a negative feedback loop with no natural brake,” Citrini writes, describing what they call “one long daisy chain of correlated bets on white-collar productivity growth.” The elegance of this scenario lies in its systemic nature—it’s not about individual job losses but about the interconnected collapse of entire economic relationships.

The Death of the Middleman

What makes this scenario particularly insidious is its attack on the very architecture of modern business. Citrini identifies a fundamental vulnerability in our current economic model: the proliferation of intermediary services that exist primarily to optimize transactions between companies. These range from specialized consulting firms to software-as-a-service platforms to entire departments dedicated to vendor management.

Agentic AI threatens to eliminate the need for these intermediaries by giving companies direct, intelligent control over their business relationships. Why pay a third-party logistics optimizer when your AI agent can negotiate directly with suppliers? Why maintain a marketing agency when your AI can generate and deploy campaigns autonomously? The report suggests that businesses will rapidly realize they can replace expensive contractors with cheaper, more efficient in-house AI systems.

This mirrors the “Death of SaaS” thesis gaining traction in tech circles, which argues that AI-native applications will render traditional software subscription models obsolete. But Citrini takes this further, suggesting that the entire ecosystem of business-to-business services—worth trillions of dollars and employing millions—could be rendered redundant almost overnight.

The Velocity Problem

Perhaps the most alarming aspect of Citrini’s scenario is the speed at which it could unfold. Unlike previous technological revolutions that played out over decades, allowing time for workforce adaptation and economic adjustment, the deployment of agentic AI could happen at internet speed.

Consider the mechanics: Once a company develops or acquires capable AI agents, there’s no physical limit to how quickly they can be deployed across an organization. Unlike human workers who require training, onboarding, and time to build relationships, AI agents can be cloned instantly and begin operating at full capacity immediately. This means that once the economic logic of AI replacement becomes clear—and the report suggests this could happen as early as 2025—the transition could accelerate far faster than anyone anticipates.

The Psychological Factor

The scenario also accounts for the psychological dynamics that could accelerate the collapse. As early adopters see dramatic cost savings and efficiency gains from AI deployment, competitive pressure will force laggards to follow suit, creating a cascade effect. CEOs who hesitate risk being seen as dinosaurs by their boards and shareholders, creating what Citrini calls a “first-mover advantage” that quickly becomes a “last-survivor necessity.”

Moreover, the report suggests that the very metrics we use to measure economic health could mask the underlying rot until it’s too late. Traditional indicators like GDP and corporate profits might initially look strong as companies report massive efficiency gains from AI adoption. Only later would the structural damage become apparent as consumer spending collapses and the velocity of money grinds to a halt.

The Counterargument

Not everyone is convinced by Citrini’s apocalyptic vision. Critics point out several potential flaws in the scenario. First, the assumption that companies will trust AI agents with critical business decisions like purchasing and strategic planning may be premature. Corporate decision-making involves complex human factors—relationships, intuition, political considerations—that AI might struggle to navigate effectively.

Second, history suggests that technological unemployment often creates new categories of work even as it destroys old ones. The Industrial Revolution automated manufacturing but created entirely new sectors like retail, entertainment, and services. Why wouldn’t AI follow a similar pattern?

Finally, there’s the question of whether the economic system has sufficient resilience to absorb such shocks. Governments could intervene with stimulus packages, universal basic income, or regulations limiting AI deployment. Companies might recognize the collective suicide implicit in mass automation and voluntarily limit their AI adoption.

The Counter-Counterargument

Yet Citrini’s defenders argue that these objections miss the point. The report isn’t predicting that AI will be perfect or that companies will immediately trust it with everything. Rather, it suggests that even partial AI deployment in key areas could trigger the feedback loop. If AI handles just 30% of white-collar work effectively, that might be enough to start the spiral.

As for new job creation, the report notes that previous technological transitions took place over much longer timescales. The AI revolution is happening too fast for the economy to adapt organically. By the time new industries emerge, the old ones might already be gone.

And while government intervention is possible, the report suggests that by the time the crisis is apparent, it might already be too late. The interconnected nature of the global economy means that a collapse in one major economy could quickly spread to others, overwhelming any single government’s ability to respond.

The Human Element

What makes Citrini’s scenario particularly compelling is its grounding in human behavior rather than just technological capability. It recognizes that companies aren’t deploying AI out of malice but out of rational self-interest. Each individual decision to replace workers with AI makes sense in isolation, even as the aggregate effect becomes catastrophic.

This mirrors real-world examples of the “tragedy of the commons,” where individual rational behavior leads to collective disaster. Just as overfishing depletes a shared resource or excessive carbon emissions destroy the climate, the rational deployment of AI could destroy the consumer economy that businesses depend on.

The Timeline

The two-year timeframe is perhaps the most controversial aspect of the report. Many experts believe that true agentic AI capable of handling complex business decisions is still years away. However, Citrini argues that we’re closer than we think. With the rapid advancement in large language models, autonomous agents, and AI reasoning capabilities, systems that can handle sophisticated business tasks may emerge as early as 2025.

The report suggests that the critical period would be 2025-2026, as early deployments prove successful and create competitive pressure for rapid adoption. By 2027, the feedback loop would be in full effect, and by 2028, the economy could be in free fall.

The Stakes

The implications of this scenario extend far beyond economics. If Citrini is even partially correct, we’re facing not just a recession or market correction but a fundamental restructuring of how our economy functions. The report suggests that the companies best positioned to survive would be those that recognize the danger early and develop strategies to maintain human economic participation even as they adopt AI.

This might involve deliberately limiting AI deployment, creating new economic models that don’t rely on traditional employment, or developing hybrid systems that combine AI efficiency with human economic activity. The companies that figure this out first might not just survive the AI transition—they might emerge stronger than ever.

The Bottom Line

Citrini’s scenario is deliberately provocative, designed to spark discussion rather than serve as a definitive prediction. The firm itself describes it as “a scenario, not a forecast,” acknowledging the many uncertainties involved.

Yet the core insight remains compelling: the deployment of agentic AI represents not just a technological shift but a potential economic singularity. For the first time in history, we’re creating machines that can not only perform tasks but make decisions about resource allocation and economic relationships. The consequences of this shift are so profound and so unprecedented that traditional economic models may not apply.

Whether the two-year timeline is accurate or the specific mechanisms differ from what Citrini describes, the fundamental question remains: are we prepared for an economy where intelligent machines make increasingly sophisticated decisions about how resources are allocated? The answer to that question may determine whether the next decade brings unprecedented prosperity or economic collapse.

Tags: AI economic collapse, agentic AI, technological unemployment, AI feedback loop, white-collar automation, future of work, AI agents, economic singularity, corporate AI adoption, technological disruption, AI timeline 2025, economic scenarios, AI and employment, business automation, AI revolution

Viral sentences: “AI capabilities improved, companies needed fewer workers, white-collar layoffs increased, displaced workers spent less, margin pressure pushed firms to invest more in AI, AI capabilities improved… It was a negative feedback loop with no natural brake.” “The system turned out to be one long daisy chain of correlated bets on white-collar productivity growth.” “It’s not Skynet-style misalignment but the gradual unspooling of the economy itself.” “Once the economic logic of AI replacement becomes clear, the transition could accelerate far faster than anyone anticipates.” “We’re facing not just a recession or market correction but a fundamental restructuring of how our economy functions.”

,