AI doomsday theory is a massive shorting opportunity

Author: The Kobeissi Letter

Translation: Shen Chao TechFlow

Shen Chao Guide: As AI tools like Anthropic demonstrate astonishing capabilities in code and workflow automation, the market has fallen into a panic of “AI doomsday,” with hundreds of billions of dollars in market value evaporating instantly. However, this article offers a highly insightful reverse perspective: the short-term shocks triggered by AI are not signs of an impending economic collapse but an inevitable process of “cognitive cost” sharply decreasing. By comparing the PC revolution of the 1980s and historical productivity data, the author points out that true “abundant GDP” era will begin when technology makes knowledge acquisition cheap and plentiful. This is not only a labor restructuring but also a necessary path toward geopolitical détente and a global productivity explosion.

Original link: It’s Too Obvious. What If AI Doesn’t Actually End The World?

The stock market just wiped out $800 billion in market value because “AI taking over the world” is becoming a consensus view. This view is too obvious. And “obvious” trades have never truly won.

The reason this doomsday scenario is spreading wildly is because it captures some instinctive elements. It depicts AI not as a productivity tool but as a macroeconomic stabilizer capable of triggering negative feedback loops: layoffs reduce consumption, reduced consumption leads to more automation, and automation accelerates layoffs.

The obvious fact is: AI is not just another software feature or efficiency tool. It is a general capability shock that touches every white-collar workflow. Unlike any revolution in history, AI is becoming proficient at “everything” simultaneously.

But what if the doomsday scenario is wrong? It assumes demand is fixed, productivity gains won’t expand markets, and system adaptation cannot outpace destruction.

We believe there is a second path, and it is greatly underestimated. The Anthropic “disassemblies” that seem like early signs of systemic collapse may ultimately be the beginning of the largest productivity expansion in history.

Before we begin, save this article and review it repeatedly over the next 12 months. While the analysis below is not necessarily the outcome, it’s important to remember that humanity always has the ability to turn defeat into victory; and free markets can always self-correct.

Anthropic’s “Disassemblies” Are Real

First, we must acknowledge the market. Anthropic is disrupting the world through Claude, causing Fortune 500 companies to lose trillions in market value.

This is a story we’ve seen several times by 2026: Anthropic releases a new AI tool, Claude makes substantial progress in programming and workflow automation, and within hours, target industry markets collapse.

If you haven’t been paying attention, here are some examples:

Market reactions to Claude announcements

IBM ($IBM) just had its worst day since October 2000, after Anthropic announced Claude can simplify COBOL code.

Adobe ($ADBE) has fallen 30% this year because generative capabilities have compressed creative workflows.

The cybersecurity sector collapsed after the release of “Claude Code Security.”

In these examples, CrowdStrike ($CRWD) stock plunged nearly at the exact moment Claude announced “Claude Code Security.”

At 1 p.m. Eastern on February 20, Claude announced “Claude Code Security,” an automated AI tool that scans for vulnerabilities in codebases.

Just two trading days later, CrowdStrike’s market cap evaporated by $20 billion due to this news.

These reactions are not irrational. The market is pricing in real-time profit compression. When AI replicates workers’ jobs, pricing power shifts to buyers. This is the first-order impact, and it is very real.

Commoditization does not mean collapse. On the contrary, it is a way for technology to lower costs and expand access. PCs commoditized computing, the internet commoditized distribution, cloud commoditized infrastructure, and AI is commoditizing cognition.

Undoubtedly, some traditional workflows will experience profit margin compression. The question is: will lower cognitive costs lead to economic collapse or enable explosive expansion?

The “Doom Loop” Assumes Demand Is Fixed

Pessimists create a simplified linear model: AI gets better, companies cut layoffs and wages, purchasing power declines, companies reinvest in AI to defend profits, and the cycle repeats. This assumes a completely stagnant economy.

History shows this is not the case. When the cost of producing something crashes, demand rarely stays unchanged; it expands. When computing costs fall, we do not consume the same amount of computing more cheaply. We consume orders of magnitude more, building entirely new industries on top.

As shown in the chart below, today’s personal computer prices are 99.9% cheaper than in 1980.

Caption: Price trend of personal computers from 1980-2015

AI reduces costs across industries, and when service costs decline, purchasing power increases regardless of wage growth.

Only if AI replaces labor without substantially expanding demand will the doomsday loop dominate. If cheap computing and productivity create new consumption categories and economic activities, an optimistic scenario emerges.

The Real Shock Is Price Collapse, Not Unemployment

Investors find it easier to promote the “obvious” story of layoffs, but the bigger news is the price compression happening in the service sector. Knowledge-based services are expensive because of their scarcity—this sounds simple, but it is true. Abundant knowledge supply leads to falling prices for knowledge work.

Think of healthcare administration, legal documents, tax filings, compliance checks, marketing production, basic programming, customer service, and tutoring. These services consume significant economic resources, largely because they require trained human attention. AI reduces the marginal cost of this attention.

In fact, as shown below, the US service sector accounts for nearly 80% of US GDP.

If operating costs decline, small businesses become more accessible; if service access costs fall, more households participate. To some extent, AI’s progress acts as a “hidden” tax cut.

Companies relying on high-cost cognitive labor may suffer losses, but the broader economy benefits from lower service inflation and higher real purchasing power.

From “Ghost GDP” to “Abundant GDP”

Pessimists’ argument relies on “Ghost GDP,” which shows output in data but does not benefit households. The optimistic counter is what we call “Abundant GDP,” combining output growth with declining living costs.

“Abundant GDP” does not require nominal income to skyrocket; it requires prices to fall faster than incomes decline. If AI lowers the costs of many essential services, even if household wages slow, their real income increases. Productivity gains are not lost; they are transmitted through lower prices.

This may explain why productivity has outperformed wage growth over the past 70+ years:

The internet, electricity, mass manufacturing, and antibiotics all provided new ways to expand output and reduce costs, despite being disruptive and volatile. But looking back, these changes permanently improved living standards.

A society that spends less time navigating complex systems and paying redundant services becomes functionally wealthier.

Labor Market Is Reshaping, Not Disappearing

A core concern is that AI disproportionately impacts white-collar jobs, which drive discretionary consumption and housing demand. This is true and a reasonable worry, especially given the already vast wealth gap.

However, AI faces more difficulties in the physical world and in human identity recognition. Skilled trades, hands-on healthcare, advanced manufacturing, and experience-driven industries still have structural demand. In many cases, AI complements these roles rather than replaces them.

More importantly, AI lowers barriers to entrepreneurship. When one can automate accounting, marketing, support, and programming tasks, starting small businesses becomes easier. We are optimistic about small enterprises.

In fact, removing entry barriers via AI might be a solution to the current wealth gap problem.

The internet killed some professions but created entirely new ones. AI may follow a similar pattern, compressing certain white-collar functions while expanding self-directed economic participation in others.

Received, continuing with the modular translation of Part 3 (final part). This section will explore the evolution of SaaS business models, AI’s reshaping of market structures, the actual performance of productivity data, and an underestimated perspective: how AI-driven “Abundance” reduces global conflicts.

The “Decline” of SaaS

AI clearly pressures traditional SaaS (Software as a Service) business models. Negotiations with procurement teams become tougher, and some long-tail software products face structural resistance. But SaaS is just a delivery mechanism, not the end of value creation.

Next-generation software will be adaptive, agent-driven, outcome-based, and deeply integrated. Winners will not be static tool providers but those best able to adapt to change.

Every technological revolution rearranges the stack; companies pricing static workflows will inevitably struggle. Those with data, trust, computing power, energy, and verification may thrive.

A layer of profit compression does not mean the collapse of the entire digital economy; it signals transformation.

AI Reshaping the Market

Pessimists believe that “Agentic Commerce” will eliminate middlemen and transaction fees. To some extent, this is true. When friction decreases, extracting fees becomes harder.

As shown below, even before AI became what it is today, stablecoin trading volume was already surging. Why? Because markets always favor efficiency.

Lower systemic friction also expands trading volume. When price discovery improves and transaction costs fall, more economic activity occurs. This is a bullish trend.

Intelligent agents representing consumers may compress platform profits based on “habits.” However, they can also increase overall demand by reducing search costs and improving efficiency.

Productivity Is the Key Variable

The ultimate determinant of an optimistic outcome is productivity. If AI can sustain productivity improvements in healthcare, government management, logistics, manufacturing, and energy optimization, the result will be abundance and lower entry barriers for all.

Even a 1–2% incremental productivity growth sustained over ten years can generate enormous compound effects.

The macroeconomic shifts driven by AI have already created some of the best investment opportunities in history. This is a field we have studied extensively and continue to lead in.

As shown below, productivity, influenced by AI, is beginning to accelerate rapidly. US labor productivity surged in Q3 2025, reaching its strongest growth in two years:

The pessimistic view assumes productivity gains fully benefit AI model builders without broader distribution. The optimistic view believes that price compression and new markets will more broadly transmit these gains.

Abundance Reduces Conflict, Not Just Costs

One of the least discussed impacts of AI-driven “Abundance” is geopolitics. For most of modern history, war has been fought over scarce resources: energy, food, trade routes, industrial capacity, labor, and technology. When resources are limited and growth feels like a zero-sum game, nations compete. But abundance changes everything.

If AI substantially lowers the costs of energy, manufacturing design, logistics, and services, the global economic pie will grow. As productivity rises and marginal costs fall, reliance on resource plunder for growth diminishes. This could end wars and usher in the most peaceful era in human history.

Economic warfare is also affected, such as the ongoing year-long trade war.

Tariffs are tools to protect domestic industries in resource-scarce worlds from cost competition. But if AI collapses production costs everywhere, why do we still need tariffs? In a high-abundance environment, protectionism becomes economically inefficient.

History shows that periods of technological acceleration tend to reduce global conflicts in the long run. Post-WWII industrial expansion decreased the incentives for major powers to confront directly.

AI-driven abundance could accelerate this trend. More efficient energy management, resilient supply chains, and localized production through automation make nations less vulnerable. As economic security increases, geopolitical aggression becomes less rational.

The most optimistic AI outcome is not just higher productivity or stock indices but a world where economic growth is no longer a zero-sum game.

Conclusion: What if the world doesn’t end?

AI amplifies outcomes. If institutions cannot adapt, it can magnify vulnerabilities; if productivity outpaces destruction, it can amplify prosperity.

Anthropic’s “disassemblies” signal workflow re-pricing and the decreasing cost of cognitive labor—a clear transformation.

But transformation does not mean collapse, just as every major technological revolution initially appears destabilizing.

The most underestimated possibility today is not utopia but abundance. AI may compress rents, reduce friction, and reshape labor markets, but it could also bring the greatest real productivity expansion in history.

The difference between a “Global Intelligence Crisis” and a “Global Intelligence Prosperity” lies not in capability but in adaptation.

And this world always finds ways to adapt.

Finally, those who can remain objective and follow processes during current turbulence are entering the best trading environment in history.