Search through thousands of pages on Integral Theory
TRANSLATE THIS ARTICLE
Integral World: Exploring Theories of Everything
An independent forum for a critical discussion of the integral philosophy of Ken Wilber
Ken Wilber: Thought as Passion, SUNY 2003Frank Visser, graduated as a psychologist of culture and religion, founded IntegralWorld in 1997. He worked as production manager for various publishing houses and as service manager for various internet companies and lives in Amsterdam. Books: Ken Wilber: Thought as Passion (SUNY, 2003), and The Corona Conspiracy: Combatting Disinformation about the Coronavirus (Kindle, 2020).

SEE MORE ESSAYS WRITTEN BY FRANK VISSER

NOTE: This essay contains AI-generated content
Check out my other conversations with ChatGPT

The AI Bubble

When Technological Revolution Meets Financial Mania

Frank Visser / ChatGPT

The AI Bubble: When Technological Revolution Meets Financial Mania

Artificial intelligence may well be the most consequential technological development of the twenty-first century. Generative AI has already transformed how people write, code, research, communicate, create images, and process information. Its potential applications in science, medicine, education, industry, and administration are enormous. Dismissing AI as merely another technological fad would be a serious mistake.

But there is another possibility that deserves equal attention: the technology may be revolutionary while the business expectations built around it are wildly exaggerated.

This distinction is essential to understanding the current AI boom. The question is not whether artificial intelligence will change society. It almost certainly will. The question is whether the enormous investments, soaring valuations, extravagant predictions, and competitive ambitions surrounding AI can possibly be justified by the economic returns that will eventually materialize.

We may be witnessing two developments simultaneously: a genuine technological revolution and a speculative financial bubble.

History suggests that these two things are not mutually exclusive. Indeed, they often occur together.

1. The Historical Pattern: Every Revolution Has Its Bubble

The history of technological innovation is also a history of financial speculation.

The railway boom of the nineteenth century, the electrification of industry, the automobile revolution, the dot-com boom of the 1990s, and the more recent cryptocurrency craze all demonstrate a recurring pattern. A transformative technology appears, investors recognize its potential, capital floods into the sector, and expectations begin to outrun realistic economic calculations.

The internet provides the most obvious comparison.

During the late 1990s, investors became convinced that the internet would fundamentally change commerce, communication, and business organization. They were right. But many companies established during the dot-com boom had no viable business models, no sustainable revenues, and no realistic prospects of profitability.

When the bubble burst in 2000-2001, enormous amounts of capital disappeared. Thousands of companies failed, and many investors suffered substantial losses.

Yet the internet did not disappear. On the contrary, the infrastructure built during the boom helped make the subsequent digital economy possible. Companies such as Amazon survived and eventually became enormously successful.

This historical lesson is particularly relevant to AI.

A financial bubble does not necessarily mean that the underlying technology is worthless. It means that expectations about its commercial potential, the speed of its adoption, or the profits it will generate have become detached from economic reality.

The railway mania built tracks that proved useful long after speculative railway companies had collapsed. The dot-com bubble financed fiber-optic networks that became essential infrastructure. Similarly, today's AI investment boom may leave behind data centers, semiconductor capacity, software, and research that will benefit society for decades.

But that does not mean today's investors will necessarily profit from their investments.

2. The Extraordinary Scale of AI Investment

The most striking feature of the current AI boom is the sheer amount of money involved.

Technology companies are spending unprecedented sums on data centers, specialized processors, electricity infrastructure, cooling systems, and computing capacity. The largest American technology companies are competing to establish themselves as indispensable providers of AI infrastructure.

According to the European Central Bank's November 2026 Financial Stability Review, cumulative capital expenditure by major American technology companies could reach approximately $5.5 trillion by 2030. The ECB also reports that free cash flows turned negative for most US hyperscalers during 2026 because AI infrastructure investment exceeded operating cash flows.

These are extraordinary numbers.

The investment is not confined to a handful of AI startups. It involves some of the richest corporations in history: Microsoft, Alphabet, Amazon, Meta, Oracle, and Nvidia, among others.

Their financial strength provides an important distinction from the dot-com era. These companies have established businesses, substantial revenues, existing customers, and profitable operations outside AI.

Nevertheless, the scale of their commitments raises a fundamental question: how much future revenue will be required to justify these investments?

The answer depends on whether AI becomes a sufficiently profitable business to support the enormous infrastructure being constructed.

A data center is not valuable simply because it contains advanced processors. It must generate enough revenue over its useful lifetime to cover construction, electricity, maintenance, financing, and the rapid obsolescence of its equipment.

And that is where the financial risks begin.

3. The Circular Financing Problem

One particularly worrying development is the increasingly complicated financial relationships between companies in the AI ecosystem.

AI laboratories need computing capacity. Cloud providers build data centers and supply that capacity. Semiconductor manufacturers sell processors to the infrastructure providers. Investment funds finance new companies. Meanwhile, large technology companies invest in AI laboratories that subsequently purchase computing services from them.

This creates an ecosystem of mutual financial dependencies.

The European Central Bank has specifically warned about circular financing arrangements in which companies finance purchases or investments that generate revenue elsewhere in the same ecosystem. Such arrangements can make reported demand and expected future revenues appear stronger than the underlying independent demand actually is.

This does not mean that every transaction is artificial or fraudulent. The companies involved may be purchasing genuinely useful services and building genuinely productive infrastructure.

But the arrangement creates a potential accounting and economic distortion.

Imagine a semiconductor manufacturer financing a company that buys its processors, which are then installed in a data center operated by another company that receives investment from the original semiconductor manufacturer.

The resulting economic activity may be real, but the apparent independence of the demand is questionable.

If the entire ecosystem depends on a continuous flow of investment, a slowdown in one part can affect all the others.

This is precisely the kind of interconnectedness that can turn an investment boom into a systemic financial problem.

4. The Valuation Problem: How Much Is AI Really Worth?

Another warning sign is the enormous valuation attached to companies whose long-term profitability remains uncertain.

The November 18, 2026, Anthropic IPO provides a particularly striking example. Reuters reported that the company raised approximately $81.3 billion in its public offering, with a fully diluted valuation of around $1.5 trillion. Its annualized revenue run rate had reached approximately $81 billion, but it was still recording an operating loss and had committed enormous sums to future computing capacity.

These figures illustrate both sides of the AI phenomenon.

On the one hand, the revenue growth of leading AI companies is genuinely remarkable. This is not simply a collection of speculative startups without customers.

On the other hand, investors are paying enormous amounts for expectations of future growth, technological dominance, and profitability.

The crucial question is whether those expectations can be fulfilled.

A company can have rapidly growing revenue and still be a poor investment if its valuation already assumes an even more spectacular future.

The distinction between a successful business and a successful investment is often overlooked during speculative booms.

Nvidia, for example, has become one of the central beneficiaries of the AI infrastructure race. Its processors are essential to much of the current AI ecosystem. But even an exceptionally successful company can become overvalued if investors extrapolate its present growth indefinitely.

There is no law of economics stating that the most important technology companies must also be the most profitable investments at every point in their development.

5. The Productivity Paradox: Where Are the Economic Returns?

Perhaps the most important question is whether AI will deliver the enormous productivity improvements promised by its advocates.

The argument is straightforward. If AI can automate substantial portions of intellectual work, companies should become more productive. Employees should be able to accomplish more in less time. Administrative costs should decline, research should accelerate, and new products and services should emerge.

There is already evidence that AI can improve productivity in particular tasks.

It can assist programmers, summarize documents, translate languages, generate marketing material, analyze data, and support scientific research. In some occupations, the benefits are substantial.

But there is a difference between improving individual tasks and transforming the productivity of an entire economy.

The Stanford AI Index 2026 reports that 88% of surveyed organizations used AI in at least one business function, while 70% reported using generative AI in at least one function. Yet the deployment of AI agents remained relatively limited across most business functions.

This distinction matters.

Experimentation and adoption are not the same as large-scale economic transformation. A company can introduce AI tools throughout its organization without necessarily generating substantial additional profits.

There are several reasons.

First, AI systems require human supervision. Their outputs can be incorrect, misleading, or incomplete. Second, companies need to redesign workflows, train employees, integrate software, and address security and legal concerns. Third, the costs of using advanced AI systems can be substantial.

The productivity gains may therefore be slower, more uneven, and more expensive to realize than enthusiasts anticipate.

This is reminiscent of the early computer revolution. Computers were introduced into businesses long before their full productivity benefits became apparent. Organizational transformation took time.

AI may follow the same pattern. But investors who expect immediate and exponential productivity gains may be disappointed.

6. The Myth of Infinite Scalability

One of the recurring assumptions in AI enthusiasm is that the technology will become increasingly powerful while simultaneously becoming cheaper and more accessible.

There is some truth to this. Hardware improvements, software optimization, smaller models, and more efficient algorithms can substantially reduce the cost of particular AI tasks.

However, advanced AI also requires enormous computing resources.

Training frontier models requires specialized chips, sophisticated data centers, and vast quantities of electricity. Operating popular models at scale creates additional costs every time users submit requests.

There is no guarantee that these costs will decline faster than demand increases.

Indeed, the more successful AI becomes, the more computing capacity may be required.

This creates an interesting economic paradox. AI companies may attract hundreds of millions of users and generate impressive revenues while still facing enormous infrastructure expenses.

The AI industry is therefore not necessarily the lightweight, software-based business model that investors have traditionally associated with Silicon Valley. It increasingly resembles heavy industry, with enormous fixed investments and substantial energy requirements.

The electricity question is particularly important. Data centers compete for access to power, water, land, and grid capacity. Their expansion can require new power plants, transmission lines, and other infrastructure.

If the economic returns fail to justify these investments, the consequences will extend beyond the technology companies themselves.

7. AI and the Labor Market: Productivity for Whom?

There is another dimension of the AI boom that receives less attention in financial discussions: the distribution of its economic benefits.

AI enthusiasts frequently present automation as an opportunity to liberate humanity from repetitive and intellectually tedious work.

That is an attractive prospect. But the transition may be considerably less comfortable.

If AI systems can perform tasks previously carried out by employees, businesses have an incentive to reduce labor costs. This could affect administrative workers, customer-service representatives, translators, junior programmers, researchers, designers, and many other professional groups.

The Stanford AI Index 2026 reports early evidence of employment pressure among younger software developers in exposed occupations. It also finds that approximately one-third of surveyed organizations expect AI to reduce their workforce in the coming year.

These developments should not be exaggerated into claims that mass technological unemployment has already arrived. The overall labor-market effects remain uneven, and new occupations may emerge.

Nevertheless, there is a fundamental social question.

If AI dramatically increases productivity, who receives the benefits?

Will they be distributed through higher wages, shorter working hours, cheaper goods, and better public services? Or will they primarily accrue to the owners of computing infrastructure, intellectual property, and financial capital?

A technological revolution can increase aggregate wealth while simultaneously increasing economic inequality.

The AI boom could therefore produce a paradox: extraordinary technological abundance alongside growing economic insecurity for substantial sections of the population.

8. The Artificial General Intelligence Factor

The financial enthusiasm surrounding AI is not driven exclusively by present-day commercial applications.

It is also fueled by expectations of Artificial General Intelligence (AGI), sometimes described as human-level or beyond-human intelligence.

The prospect of machines capable of performing virtually all intellectual tasks has enormous implications. If realized, it could transform scientific research, engineering, medicine, education, and the organization of economic life.

Some technology executives and investors appear to regard AGI as an almost inevitable destination of current developments.

But this assumption requires scrutiny.

Today's large language models are impressive systems. They can generate sophisticated texts, write computer code, answer questions, and assist with complex intellectual tasks. Yet they also exhibit significant limitations, including hallucinations, unreliable reasoning, inconsistent performance, and difficulties with autonomous long-term planning.

The distance between increasingly capable AI systems and genuinely general, reliable, autonomous intelligence remains uncertain.

There is also a danger of circular reasoning:

  • AI systems become increasingly capable.
  • Their capabilities are interpreted as evidence that AGI is approaching.
  • Expectations of AGI increase investment and company valuations.
  • The resulting investment is presented as evidence that the future of AGI is economically inevitable.

But investment enthusiasm is not evidence of technological inevitability.

Nor does the achievement of AGI automatically guarantee that the companies developing it will capture sufficient economic value to justify their valuations.

The financial markets may be pricing in a future that remains technologically, economically, and socially uncertain.

9. The AI Bubble as an Ideological Phenomenon

The AI boom is not merely an economic development. It is also an ideological phenomenon.

We are witnessing the emergence of a powerful narrative in which artificial intelligence is presented as the solution to almost every major human problem.

It will revolutionize medicine, accelerate scientific discovery, improve education, eliminate bureaucracy, increase economic growth, and perhaps even solve the mysteries of consciousness.

Some of these expectations are reasonable. Others are highly speculative.

The language surrounding AI increasingly resembles the rhetoric of earlier technological utopias. We are promised an imminent transformation of humanity, sometimes accompanied by predictions of abundance, radically extended lifespans, and the disappearance of conventional economic constraints.

This narrative has commercial value. It attracts investors, customers, talented employees, government subsidies, and political attention.

But it can also become self-reinforcing.

When technological developments are interpreted through an almost religious framework of inevitable progress, criticism becomes difficult. Skeptics are portrayed as technological Luddites, while genuine questions about costs, reliability, social consequences, and financial sustainability are dismissed as temporary obstacles.

There is an important distinction between technological optimism and technological messianism.

The former recognizes possibilities while demanding evidence. The latter treats the promised future as sufficiently certain to justify almost any present investment.

We should be particularly cautious when the same executives who stand to benefit financially from the AI revolution are also its most enthusiastic prophets.

10. What Could Trigger an AI Bubble Collapse?

Financial bubbles rarely burst simply because somebody announces that valuations are too high. They usually collapse when an event exposes the gap between expectations and economic reality.

Several developments could trigger such a correction.

Disappointing financial returns. If AI companies fail to generate sufficient profits to justify their infrastructure investments, shareholders may begin questioning the entire business model.

Slower technological progress. If improvements in AI capabilities become increasingly expensive or incremental, the expectation of continuously accelerating development could weaken.

Intensifying competition. Open-source models, smaller specialized systems, and cheaper alternatives could undermine the pricing power of the largest AI providers. The more accessible AI becomes, the harder it may be for individual companies to maintain extraordinary profit margins.

Infrastructure overcapacity. If companies build more data centers than the market ultimately needs, computing prices could fall, leaving investors with expensive facilities that generate inadequate returns.

Rising financing costs. As AI infrastructure increasingly depends on borrowing, higher interest rates or declining investor confidence could make projects more difficult to finance.

These risks are not merely theoretical. In its November 2026 Financial Stability Review, the ECB explicitly warns that the AI investment boom could produce overcapacity, capital misallocation, and credit losses. It also draws parallels with the dot-com bubble, while emphasizing the additional risks created by debt financing and opaque financial arrangements.

The Bank for International Settlements has likewise warned that the scale of the AI investment boom and the expectations surrounding its economic returns resemble previous episodes of financial exuberance. Bank for International Settlements

A correction could take several forms. AI startup valuations might collapse, infrastructure companies could default, technology shares could decline sharply, and investment projects might be cancelled.

But a financial correction would not necessarily mean the end of AI development.

Indeed, it might produce a healthier industry by forcing companies to concentrate on economically useful applications rather than speculative expansion.

11. Who Will Be the Winners and Losers?

An important lesson from previous technological bubbles is that the eventual winners are not always the companies that initially attracted the most attention.

During the dot-com boom, numerous internet companies disappeared. Yet the underlying infrastructure and technology continued to develop. Some of the companies that survived became dominant economic forces.

The same could happen with AI.

The ultimate beneficiaries might not be the companies currently commanding the highest valuations. They could include established businesses that successfully integrate AI into their operations, specialized software developers, scientific research organizations, healthcare providers, and smaller companies offering efficient and affordable AI applications.

The companies manufacturing chips and building infrastructure may also benefit, although their fortunes depend on sustained demand and the profitability of their customers.

There is another possibility: AI capabilities may become increasingly commoditized.

If powerful models can be developed and operated at substantially lower costs, the competitive advantage of today's leading providers could diminish. The economic value might then shift from developing foundational models to applying them effectively.

This would be a familiar pattern in technological history. The introduction of a revolutionary technology initially creates enormous profits for its pioneers, but subsequent competition gradually distributes its benefits more widely.

The internet eventually became a general-purpose infrastructure. AI may follow a similar trajectory.

The technology could become more important to society even as the financial fortunes of some of its current champions deteriorate.

12. The Environmental and Social Bill

There is also a broader cost-benefit question.

The AI investment boom is often presented as though technological progress automatically produces social progress. But the construction and operation of enormous data centers have material consequences.

They require land, construction materials, electricity, cooling systems, and water. Their environmental footprint depends partly on the energy sources used and the efficiency of their operations.

There are also questions about the concentration of technological power.

A relatively small number of corporations control much of the infrastructure, computing capacity, and intellectual property underlying advanced AI systems. This concentration creates dependencies for businesses, governments, educational institutions, and individual users.

The political implications are considerable.

If AI becomes essential infrastructure, access to it will increasingly become a question of economic and geopolitical power. Countries that depend on foreign providers may find themselves vulnerable to commercial decisions, export restrictions, and technological dependencies.

Consequently, the AI bubble debate should not be restricted to stock-market valuations.

We should also ask whether the enormous resources being committed to AI are producing benefits commensurate with their economic, environmental, and social costs.

A technology can be commercially successful without necessarily delivering the greatest possible public benefit.

13. The More Fundamental Question: What Is AI Actually Worth?

We should distinguish three questions that are frequently conflated.

First: Is artificial intelligence technologically important?

Almost certainly. Generative AI is already changing intellectual work, and further advances are likely.

Second: Will AI generate substantial economic value?

Very probably. Its applications across industries are too extensive to dismiss. But the timing, distribution, and magnitude of the returns remain uncertain.

Third: Are current investments and valuations economically justified?

That is the genuinely open question.

The answer cannot be derived from technological enthusiasm alone.

It requires evidence about productivity improvements, customer willingness to pay, operating costs, profit margins, infrastructure utilization, competition, and the sustainability of financing.

The central problem is that investors may be extrapolating from the first two propositions to an unjustified conclusion about the third.

Just because a technology will transform society does not mean that every company investing in it will succeed.

Just because AI becomes indispensable does not mean that today's AI leaders will maintain their dominance.

And just because the future is technologically extraordinary does not mean that investors should pay extraordinary prices for it today.

Conclusion: A Real Revolution Inside a Potential Financial Bubble

The most sensible position on the AI boom is neither technological pessimism nor uncritical enthusiasm.

AI is real. Its achievements are real. Its economic potential is real. But the financial expectations surrounding it deserve much greater scrutiny.

The historical parallels with the railway mania and the dot-com bubble are instructive. In both cases, a genuinely transformative technology became the object of excessive speculation. Investors confused the importance of the innovation with the financial prospects of the companies associated with it.

Something similar may be happening today.

The extraordinary investments in AI infrastructure, the increasingly complicated financing arrangements, the enormous valuations of AI companies, and the expectation of almost unlimited future productivity gains all warrant caution.

The November 2026 warnings from the ECB and BIS make it particularly difficult to dismiss the bubble hypothesis as merely the pessimism of technological skeptics.

But we should avoid the opposite mistake of assuming that a financial correction would invalidate the technology itself.

The internet survived the dot-com crash. The railway system survived railway speculation. AI will almost certainly survive any correction in its financial markets.

Indeed, the eventual beneficiaries might be those companies and institutions that emerge after the speculative excess has been removed.

There is a final irony here.

Artificial intelligence is supposed to help humanity become more rational, more informed, and better equipped to distinguish evidence from illusion. Yet the investment boom surrounding it exhibits some of the oldest irrationalities in human economic behavior: herd mentality, fear of missing out, exaggerated expectations, and the belief that this time things will be different.

Perhaps the greatest test of artificial intelligence is not whether machines can become smarter than humans, but whether the humans investing in them can remain sufficiently rational to recognize a bubble when they are inside one.

The technology may be revolutionary. The financial expectations may be delusional. And both propositions can be true at the same time.


PLEASE NOTE: Comments containing links are not allowed, to avoid spam.


Comment Form is loading comments...