The case against an AI bubble: the strongest bull arguments

What is the strongest bull argument against an AI bubble?

The first bull argument is demand. The EQOS Asia AI study projects productivity gains of +40% (EQOS Asia AI study, April 2026). A gain of that size gives buyers a reason to keep paying, and repeat payment is what turns spending into revenue. The bull argues that a technology with a measurable benefit finds its customers, even when early projects take longer than the forecast.

A fair bull states the test: the gain must show up in the accounts of the buyers. Look for lower cost per task, shorter cycle times, or higher output per employee, measured before and after deployment. When buyers report these gains, spending rests on earnings, which is the mark of a boom. Record the source and the date of each report you collect.

How does the scale of AI change argue against a bubble?

The second argument concerns scale. The study counts 300 million jobs disrupted by AI, and it cites McKinsey’s estimate that 30% of hours worked could be automated by 2030 (EQOS Asia AI study, April 2026). Change on that scale reaches every sector, so the market for AI tools spans the whole economy. The bull reads breadth as a reason for demand to outlast any single investment cycle.

Breadth also spreads risk. A technology used in many industries depends less on any one customer group. The study adds a condition: fewer than 15% of companies have a retraining plan for the people affected. The bull sees a large service market in training and change management, and counts it as extra demand around the tools themselves.

Why do bulls treat AI project failures as proof against an AI bubble?

The third argument reframes failure. MIT 2025 found 95% of generative-AI projects failing to reach production, and the bull reads this as the early phase of a new tool. The study cites Bain on a wider pattern: $2.3 trillion of digital-transformation projects fail each year across the world. Failure at this rate belongs to digital change in general, and AI joins an old pattern.

Learning curves bend. Teams that fail once learn which data, which use case, and which controls matter, and the next project starts from that knowledge. The bull therefore tracks the share of pilots that reach production each year, rather than the first-year failure rate. If that share rises, the learning is real and the revenue follows. If it stays flat, the bear gains ground.

What does the structure of the AI market say against a bubble?

The fourth argument concerns structure. The study finds every dimension of the enterprise AI market above the critical line of 0.3. Energy scores 0.588, the highest of the 9, and clarity scores 0.487. The overall φ(S) of 0.337 sits in the fragile zone, which the study defines as a system that functions. The bull stresses that functioning is the first condition of a boom, and a fragile market can still strengthen.

Fragile means open to improvement as well as to shocks. The study sets 0.5 as the viability threshold for a successful AI deployment, and the distance from 0.337 to 0.5 is a work list: data quality, explainability, strategy alignment, and retraining. Each item sits within the control of a company. A market made of companies that act on the list moves upward without waiting for outside help.

How do fixable causes weaken the case for an AI bubble?

The fifth argument concerns correction. The study lists 4 actions: measure data quality before deploying, require explainability, audit strategy against field reality, and fund retraining. Applied together, they lift φ(S) from 0.337 to 0.401 and lower the predicted failure rate from 94% to 82% (EQOS Asia AI study, April 2026). The bull reads this as proof that the failure rate responds to management choices.

The study values the gain at $80 billion of recoverable value per year for the world market, about 40% of the $200B invested. A bull treats this figure as the prize for learning faster. You can run your own version on the Λ simulator, free and anonymous, and compare your organisation with the market score. A company above the market score sits on the bull side of the argument.

What observations would weaken the case against an AI bubble?

A bull case earns trust when it names its own disproof. Four observations would weaken it: gains that stay out of the buyers’ accounts, pilots that stop reaching production, a shock that pushes communication below 0.3, and funding that tightens before revenue arrives. The study names the shocks that matter: a regulation, an incident, or a recession (EQOS Asia AI study, April 2026).

The AI Act adds a known date. The study lists AI Act compliance on December 2, 2027, and the bull sees a fixed deadline as a planning aid for buyers. The bear sees a cost. Both readings fit the same fact, so you track the effect on spending after the date. Mark each pillar green or red every quarter and compare the result with the bear side.

How can you use the bull case against an AI bubble in a decision?

Use the bull case as an opportunity test, not as a promise. Ask which AI use case in your business has a measurable gain, which data feeds it, and who owns the result. Fund the use cases that answer all 3, and review each one at a fixed date. This turns a hopeful argument into a short list of staged projects.

EQOS Asia scores a single decision with the same method. If your decision involves AI spending in Singapore, ASEAN, or Hong Kong, a 30-minute pre-diagnostic returns a score and a ranked set of options, and the method page shows how the score is built. This page is general information, not investment advice.

Frequently asked questions

About the author

Stéphane Jambu

President, EQOS Forecast Asia

Stéphane Jambu leads EQOS Forecast Asia and works as a consultant in e-commerce, search visibility, and AI. LinkedIn profile