The Great AI Paradox: Everyone Says It’s a Bubble, No One’s Slowing Down

The chorus of warnings comes with a curious twist: none of them are pulling back from the AI revolution.

The Wall Street warnings diverge from typical AI bubble hysteria: the same institutions sounding alarms are also identifying reasons for optimism.

(Photo: The Quint)

When Silicon Valley's visionaries and Wall Street's gatekeepers simultaneously acknowledge a bubble, everyone should listen.

Jeff Bezos, the founder of Amazon, recently described the Artificial Intelligence (AI) boom as an "industrial bubble" at the Italian Tech Week, though he insists the technology is "real" and will bring "gigantic" societal benefits. Open AI founder Sam Altman has been even more candid, acknowledging the "insane" valuations and admitting that investors are "overexcited about AI," warning that people will "get very burnt."

The AI Bubble Paradox

Yet, the chorus of warnings—from Bezos and Altman in tech to David Solomon at Goldman Sachs and Jamie Dimon at JP Morgan—comes with a curious twist: none of them are pulling back.

This paradox reveals something profound about the nature of the AI transformation unfolding before us, and why this bubble might be fundamentally different from those that preceded it particularly the dot-com crash.

The concerns are mounting from multiple fronts. Goldman Sachs CEO David Solomon warned at the same Italian Tech Week that AI's rapid acceleration is driving capital formation that could push valuations beyond sustainable fundamentals, while Solomon noted that "there will be a lot of capital that was deployed that didn't deliver returns" and emphasised "it's not different this time".

JP Morgan's Jamie Dimon described himself as "far more worried than others" about an AI-driven market boom that could mirror the dot-com crash. Morgan Stanley's top analyst Lisa Shalett expressed being "very concerned" about AI's grip on markets, noting the prominence of private equity and debt capital "tends to produce bubbles, because it may be unspoken-for capacity".

The numbers justify their alarm. Microsoft plans to spend $80 billion on AI data centers this fiscal year, while Meta projects up to $72 billion in AI infrastructure investments. Just five AI hyperscalers are projected to spend over $1 trillion collectively by 2027. The concentration is staggering: AI companies have accounted for 75 percent of S&P 500 gains, 80 percent of profits, and a shocking 90 percent of capital expenditures.

Yet, here's where the Wall Street warnings diverge from typical bubble hysteria: the same institutions sounding alarms are also identifying reasons for optimism.

Goldman Sachs noted that factors including lower levels of IPOs, scrutiny of AI returns, and valuations below late 1990s peaks argue against a current bubble scenario. JP Morgan observed that strong financials and measured capex show no signs of an AI bubble, with valuations remaining contained and less than 10 percent of US companies actively using AI to produce goods or services. The runway, they suggest, remains long.

This is where the dot-com comparison breaks down fundamentally. The late 1990s internet bubble was characterised by companies with no revenue, no viable business models, and often no real product beyond a website. Pets.com burned through $300 million selling pet supplies at a loss. Webvan promised grocery delivery without understanding unit economics.

The internet was real, but most companies riding the wave were not.

AI, conversely, is already delivering measurable value. Large language models are writing production code, analyzing complex legal documents, and assisting in medical diagnoses. Computer vision systems are improving manufacturing quality control. The technology works—the question is whether returns will justify the investment scale and timeline.

Industrial vs Financial Bubble

Bezos distinguished between what he called an "industrial bubble" versus financial bubbles, and this matters enormously. He compared the situation to the 1990s biotech bubble, noting that while investors lost money, society gained lifesaving drugs. Industrial bubbles leave behind infrastructure, knowledge, and capabilities that outlast financial wreckage. Britain's 1840s railroad bubble bankrupted countless investors but built a transportation network. The late-1990s fiber optic bubble destroyed billions in shareholder value but laid the physical backbone for today's internet.

Today's AI spending is similarly building durable infrastructure. Data centers, chip fabrication facilities, and research capabilities don't vanish when valuations correct. The intellectual capital—trained researchers, refined algorithms, accumulated datasets—persists. As Altman acknowledged, "People will overinvest and lose money, and underinvest and lose a lot of revenue," but the technology itself will deliver over the longer arc.

The skepticism centers on monetisation and timing. Many AI companies sport valuations requiring them to capture enormous market shares in industries they've barely penetrated.

JP Morgan Global Research found that AI hasn't materially changed business practices across non-tech industries yet, with less than 10 percent of firms actively using AI as of mid-2025. The gap between demonstration and deployment, between pilot projects and enterprise-wide adoption, remains significant. Some investors will indeed lose fortunes.

Yet the capital intensity appears more rational than critics suggest. The winner-take-all dynamics of AI—where model training costs and compute requirements create enormous barriers to entry—favour well-capitalised players making calculated bets that controlling AI infrastructure will prove as lucrative as controlling cloud infrastructure has been for Amazon, Microsoft, and Google.

India and the AI Revolution

For India, the implications are profound. The country's IT services industry faces disruption as AI threatens to automate routine coding and back-office tasks, employing millions. Yet India's technical talent pool, thriving startup ecosystem, and massive domestic market position it as a potential AI winner.

Indian companies are integrating AI into customer service, agriculture, healthcare, and financial services. The challenge lies in moving from AI implementation to AI innovation before the window closes.

India's digital public infrastructure and diverse datasets provide unique advantages for developing AI applications for emerging markets. If Indian companies can solve problems specific to developing economies—multilingual interfaces, low-bandwidth optimization, cash-based systems—they may develop exportable solutions that work across the Global South, creating an AI corridor that bypasses Western dominance.

The AI bubble debate ultimately misses the nuance.

Yes, David Solomon warns that markets are due a drawdown after years of AI-propelled highs. Yes, valuations will correct. Yes, high-profile failures await. But beneath the froth lies genuine technological transformation. The financial industry's warnings aren't about AI's validity—they're about timing, concentration, and the inevitable pain when reality meets hype.

If history offers guidance, the long-term gains will justify short-term excess. The first wave often gets cut down, but they take the beach. The question isn't whether AI is a bubble—even the bulls acknowledge that. The question is whether you're building lasting infrastructure or chasing momentum. That distinction will separate the Amazons from the pets.coms of this era.

(Subimal Bhattacharjee is a Visiting Fellow at Ostrom Workshop, Indiana University Bloomington, USA, and a cybersecurity specialist. This is an opinion piece. The views expressed above are the author’s own. The Quint neither endorses nor is responsible for them.)