AI bubble fears echo dot-com bust, but fundamentals differ

As the world navigates the frenzy of the Artificial Intelligence (AI) revolution, a haunting question looms over boardrooms, social media, and stock exchanges alike. Are we sleepwalking into another Dot-com crash?

The parallels are seductive. There are astronomical valuations, for instance, with privately held OpenAI eyeing a USD 500-plus-billion valuation—and a market heavily concentrated in a handful of technology darlings. The so-called “Magnificent Seven” stocks (Alphabet, Amazon, Apple, Meta Platforms, Microsoft, Nvidia, and Tesla) now account for roughly 37% of the S&P 500, a density of power that naturally invites anxiety.

Nvidia’s market capitalisation alone represents nearly 8% of the S&P 500, giving it the capacity to move the market almost single-handedly.

However, a closer look at the fundamentals suggests that while history may rhyme, it is not repeating itself. The fear of an imminent, systemic “AI bubble” bursting in a manner similar to the year 2000 appears misplaced.

The market may correct, and the hype will inevitably cool, but the structural foundations of the AI economy are fundamentally different from the house of cards built in the late 1990s. The reasons are worth examining.

Echoes of the Dot-com Era vs Today’s Reality

To understand why this cycle differs, it is essential to examine the velocity of this mania. Between 1995 and 2000, the Nasdaq Composite surged by nearly 400%. It was a period defined by euphoria around business models that did not yet exist.

Capital chased eyeballs rather than revenue. Start-ups with zero tangible assets and massive burn rates were listing on public markets, fuelled by retail FOMO (fear of missing out) and abundant cheap money.

By contrast, the current cycle presents a different picture. While technology stocks have rallied strongly, the growth curves—approximately 73% since November 2022, when ChatGPT was launched—are substantial but nowhere near the near-vertical 400% ascent of the dot-com era. More importantly, the character of capital deployment has shifted decisively.

In 2000, money flowed largely into marketing campaigns and web “portals”. Today, capital is being channelled into tangible, enduring assets. The AI boom is capital-intensive in a way the internet boom never was.

Hundreds of billions of dollars are being invested in data centres, specialised silicon, advanced cooling systems, land acquisition, and energy infrastructure. This represents a build-out phase of industrial proportions rather than a marketing-led spending spree. Even if AI software monetisation takes longer than anticipated, the physical infrastructure—the digital age’s equivalent of railroads—will retain long-term value.

Why Today’s AI Boom Has Adult Supervision

The most critical differentiator lies in who is investing. The dot-com bust was largely driven by fragile start-ups. The AI build-out, by contrast, is being financed by some of the most profitable, cash-rich corporations in history: Microsoft, Amazon, Meta, Oracle, and Alphabet.

These companies operate with fortress-like balance sheets and operating margins exceeding 30%. They are not burning venture capital in the hope of discovering a viable business model; they are reinvesting profits generated from entrenched positions in cloud services, search, social platforms, and e-commerce to secure the next technological frontier.

When Microsoft invests in OpenAI or builds hyperscale data centres, it deploys cash generated from high-margin Azure and Office businesses. This creates a structural floor under the market that did not exist in 2000.

Financial Signals: Valuations, Yields, and Inflation

Beyond qualitative differences, financial indicators point towards rationality rather than excess. Recent analyses, including perspectives echoed in mainstream US media on investor sentiment heading into 2026, highlight two metrics in particular: excess CAPE yield and breakeven inflation rates.

The excess CAPE yield (cyclically adjusted price-to-earnings yield) measures the relative attractiveness of equities compared with risk-free government bonds. In 2000, this metric had turned sharply negative; equities were so expensive relative to bonds that investors were effectively paying to assume risk.

Today, despite elevated price-to-earnings ratios, strong corporate earnings, and stabilising bond yields, the excess CAPE yield for the 10-year S&P 500 remains positive, or at least within investable territory. Equities may be expensive, but they are not irrationally priced compared with the alternative of holding bonds.

Similarly, the breakeven inflation rate—the spread between nominal US Treasury yields and inflation-protected securities—indicates that inflation expectations remain anchored.

Unlike the volatile post-pandemic inflation spikes, expectations have settled. This stability gives the US Federal Reserve and other central banks room to manoeuvre, increasing the likelihood of a “soft landing” rather than a disorderly crash.

Key Risks: Market Concentration and Circular Demand

This does not imply an absence of risk. Two specific red flags merit attention. The first is concentration risk. With 37% of the S&P 500 tied to just seven companies, any regulatory shock or earnings disappointment from a single heavyweight—such as Nvidia or Microsoft—can ripple across the entire market. Investors are, in effect, crowded into the same trade.

The second concern is the emergence of a “circular AI economy”. At present, a significant portion of AI-related revenue growth stems from companies selling services to one another, or hyperscalers selling cloud compute to start-ups in which they themselves have invested.

This circularity can create the illusion of sustainable revenue growth without broad-based end-user adoption. If enterprises fail to realise meaningful productivity gains from tools such as Copilot or models from ChatGPT, Gemini, or Anthropic within a reasonable timeframe, spending could slow, potentially triggering an “AI winter” or a sharp market correction.

India’s Strategic Moment in the AI Build-Out

For India, the distinction between a speculative bubble and a structural build-out is critical. As the global narrative shifts from hype to implementation, India’s role is evolving.

The country is no longer positioned merely as a back office. With initiatives such as the IndiaAI Mission and a deep talent pool in science, technology, engineering, and mathematics, India is increasingly well placed to become a factory floor for the AI era.

The global emphasis on tangible infrastructure—spanning semiconductors and large-scale data centre investments, including recent announcements by Alphabet and Microsoft—aligns closely with India’s own AI ambitions and industrial priorities.

Expect Volatility, Not a Systemic Collapse

Market volatility should be expected. A correction of 10% or even 20% is plausible if earnings disappoint or valuations race ahead of fundamentals. However, a systemic collapse appears unlikely.

The AI revolution rests on the balance sheets of corporate giants and the construction of real-world infrastructure. It is driven by tangible capital expenditure, underpinned by broadly rational—if optimistic—financial metrics, and powered by a technology that promises genuine productivity gains.

Vigilance remains essential, but panic is unwarranted. The market is more likely to enter a phase of digesting elevated valuations—through consolidation, plateaus, or modest pullbacks—than to experience a catastrophic burst. The dot-com era was a story of fiction; the AI era, by contrast, is shaping up as an expensive work of non-fiction.

The author is a consultant and former Partner at KPMG.
Views are personal.