Is AI the new Tulip Mania?

Is AI the new Tulip Mania?

In the 1630s, the Dutch Republic experienced what is often described as the first recorded speculative bubble: Tulip Mania. Prices for rare tulip bulbs soared to extraordinary levels before collapsing abruptly in 1637.

Nearly four centuries later, massive capital flows into AI chips, data centers, and foundation models raise a familiar concern: Is AI our tulip moment?

Tulips were not useless. They were beautiful, desirable, and culturally significant. Their value stemmed from rarity and difficulty of reproduction; scarcity became a signal of importance. A simple belief system took hold: rarity meant value, rising prices confirmed significance, and ever-larger contracts promised future wealth. The logic was circular, but powerful.

When the fever broke, little remained beyond financial losses and a cautionary tale. This is the defining characteristic of a pure bubble: capital concentrates around an asset whose price reflects belief more than structural contribution to societal transformation.

Today’s AI expansion revolves around compute infrastructure (GPUs, hyperscale data centers, foundation models) and the pursuit of scale within an intensely competitive global ecosystem. Companies such as NVIDIA have experienced extraordinary growth, while governments increasingly frame AI as critical infrastructure.

Unlike tulips, AI systems are embedded in software development, customer service, drug discovery, logistics, finance, and education. Their applications are real. The risk is not that AI lacks potential, but that we may be anchoring its promise to the wrong variable: scale.

That is, the implicit doctrine of the current AI race follows a familiar ‘tulip’ pattern: the largest model will win, the most compute will dominate, scale equals intelligence. This is not an engineering law; it is a belief system. We have shifted from observing that scale can improve performance to assuming that scale itself constitutes the breakthrough.

The consequences are visible: concentration of power in a small number of actors, escalating environmental demands, education systems prioritizing prompt literacy over deep competence, labor strategies oriented towards replacement rather than augmentation, and governance that lags behind infrastructure expansion. Infrastructure accumulation is mistaken for intellectual progress.

Moreover, even where measurable task-level gains are evident, these do not automatically translate into structural productivity growth, institutional innovation, democratic resilience, or expanded human capabilities. If scale were synonymous with transformation, we would expect to see sustained acceleration in economy-wide productivity. Yet national statistics show no clear AI-driven rise in GDP per hour worked. Explanations abound: diffusion takes time; benefits concentrate in specific sectors; measurement tools miss qualitative improvements; returns accrue mainly to capital rather than labor. These may be valid. But they echo earlier episodes, including the productivity paradox of the early IT era. When gains are captured narrowly within specific geographies, sectors, or financial systems, valuations can inflate without systemic advancement. Mania generates transactions; transformation requires institutional redesign.

The more subtle bubble may be forming in education and labor. When educational systems overemphasize short-term tool proficiency while underinvesting in critical thinking, governance, ethics, and socio-technical understanding, they align human development with the hype cycle rather than long-term societal needs. When labor markets prioritize substitution over augmentation and fail to build serious reskilling pathways, inequality deepens and institutional resilience weakens.

Tulips were beautiful. AI models are powerful. Both can attract valuation based on expectations of future dominance secured through size. The danger is not that AI is empty. The danger is conflating bigness with intelligence and centralization with progress.

Tulip Mania collapsed when belief shifted. AI will not disappear, but belief can migrate from exuberance to disillusionment with equal speed. If today’s investments continue to privilege scale over institutional redesign, dominance over distribution, and infrastructure over human capability, the reckoning will not be technological failure. It will be social and economic distortion: concentrated power, widened inequality, brittle institutions, and public trust eroded by unmet promises.

The choice is not between hype and rejection. It is between embedding AI within democratic, human-centered socio-technical systems, or allowing a scale-driven paradigm to define progress on its own terms. The first demands restraint, governance, and deliberate institutional change. The second accelerates until correction is imposed from the outside. Bubbles do not end because technologies vanish. They end because societies refuse to sustain the imbalance.