AI Stock Euphoria Now Outruns Japan’s Bubble, Ikigai Warns

Indian fund manager Pankaj Tibrewal has placed artificial intelligence stocks in what his firm calls a “zone of visible excess,” arguing that a decade of Nasdaq gains has now outrun Japan’s 1980s bubble, the 1920s US bull market and the late-1990s technology boom. In its latest quarterly newsletter, Mumbai-based Ikigai Asset Manager stops short of calling a market top and says AI itself will keep transforming businesses. The concern it raises is the price investors are paying for that promise.

Tibrewal is founder and chief investment officer of IKIGAI Asset Manager, a portfolio management firm he set up after years on the buyside. The argument Ikigai has now published is not a forecast of when, or whether, stocks will fall. It is a diagnostic on how many indicators of excess are flashing at once, and what that combination has historically implied.

The Diagnosis From Ikigai

Ikigai’s case rests on several gauges simultaneously moving toward caution. The firm frames them as independent signals pointing the same way, not as one alarming data point. “There is little doubt AI will continue transforming businesses over the next decade,” the newsletter says. “The question is different: How much of that future is already reflected in today’s prices?”

The fund’s argument is not that AI will fail. It is that investor expectations have moved ahead of the underlying economics, with valuations, leverage, concentration and capital expenditure all approaching levels historically tied to euphorias that ended badly. Ikigai does not name a date for a correction. It says discipline becomes more important when optimism becomes abundant.

A Decade That Outran Three Famous Bubbles

The headline number is the Nasdaq 100. Over the past ten years, the index has delivered cumulative returns of more than 640%, a figure Ikigai says surpasses the gains recorded during Japan’s equity bubble of the 1980s, the roaring US bull market of the 1920s and the technology boom of the late 1990s. Past returns are not a forecast, and history does not always repeat. The fund’s point is that the cumulative gain itself is rare, and rare moves have ended with rare consequences.

Other gauges sit close to historical extremes. The S&P 500 trades at a cyclically adjusted price-to-earnings (CAPE) multiple above 35 times, a level the firm says has been seen only a handful of times in the past century and that sits close to the dot-com peak. Margin debt extended by US brokers has crossed a record $1.3 trillion, and retail borrowing has accelerated sharply alongside it. Each indicator on its own could be dismissed. Stacked together, Ikigai argues, they describe a market where the cost of being wrong has grown.

A third gauge, the ratio of capital expenditure to operating cash flow at the four largest hyperscalers, has more than doubled in three years. In 2023 those companies spent 41% of operating cash flow on capex. By 2026 the firm expects that figure to reach almost 92%. The four largest hyperscalers are Alphabet, Amazon, Meta and Microsoft, and the denominator in that ratio is the cash those companies generate, not their revenue.

The fund also reports that the Magnificent Seven have recently begun underperforming both the broader S&P 500 and the Russell 2000, an early sign of rotation that Ikigai reads as investors slowly moving beyond the narrow AI leadership that has dominated markets in recent years.

• Nasdaq 100 ten-year return: more than 640%
• S&P 500 CAPE: above 35 times
• US broker margin debt: a record $1.3 trillion
• Top 10 weight in the S&P 500: nearly 40%
• Hyperscaler capex as share of operating cash flow: ~92% by 2026, up from 41% in 2023

The $1 Trillion Question

The single largest number in Ikigai’s note is the AI capex bill. Alphabet, Amazon, Meta and Microsoft together are estimated to spend around $700 billion on capital expenditure this year and more than $800 billion next year. Adding Oracle, OpenAI, Anthropic and other AI infrastructure providers lifts expected AI investment to more than $1 trillion in the year ahead, roughly equivalent to 3% of US GDP and to about one-third of the total pre-tax profits of all US non-financial companies.

Within that, the memory piece alone is now absorbing roughly 28% of operating cash flow at the largest hyperscalers this year. The bill has to be financed, and the source of that financing matters as much as the size of the bill. Ikigai’s worry is not that the spending is happening. It is what happens when investors start asking whether the returns will justify it.

A sudden realisation that hyperscalers, OpenAI and Anthropic may not earn sufficient returns on AI-related investments could trigger an unwillingness to continue funding these investments.

The mechanism Ikigai describes is feedback. If a hyperscaler pauses a data-centre build, the supplier takes the hit. If a supplier’s revenue is contingent on a hyperscaler’s build, the supplier’s lenders start asking questions. What looks like a single capex line on one company’s balance sheet is, by Ikigai’s count, a chain that already runs through balance sheets across the AI supply chain. The same newsletter notes that financing arrangements between suppliers and customers could amplify any reassessment.

Ikigai is also asking a separate question about the same trillion dollars: how much of it is being financed by the customers who will eventually need to monetise the AI compute they are buying. The supplier-buyer loop is the part Tibrewal has previously flagged in interviews on the relationship between hyperscalers and AI labs, and it is the part most exposed if revenue at the AI layer disappoints.

Concentration Where It Counts

Concentration is another pillar of the case. The top 10 companies now account for nearly 40% of the S&P 500, an unusual share of the index that places the burden of US equity returns on a small group of names. Outside the US, the picture sharpens. Taiwan Semiconductor Manufacturing Co. represents 58% of Taiwan’s benchmark, and Samsung Electronics and SK Hynix together account for more than half of South Korea’s technology leadership.

For context, two of those Asian suppliers have already been at the centre of sharp 2026 market moves. The KOSPI circuit breaker episode with Samsung and SK Hynix showed how a single shock in Korean tech can spill across the region. And Japan’s parallel AI push, including Japan’s trillion-yen national AI model plan, is now layering another large capital commitment onto a region whose largest equity benchmarks already trade as concentrated proxies for AI demand.

The implication Ikigai draws is not that any of these names will fall. It is that the broader indices no longer diversify away the AI trade. A re-rating in a handful of US stocks moves the S&P 500, and a re-rating in a handful of Asian suppliers moves the AI supply chain.

The First Cracks in AI Leadership

The rotation the fund points to is visible but early. The Magnificent Seven have begun underperforming both the broader S&P 500 and the Russell 2000, an inversion of the leadership that defined 2024 and most of 2025. A separate 2026 rotation narrative tracks the same shift, with Magnificent Seven stocks in the red for 2026 as collective momentum has stalled and small-caps and rate-sensitive cyclicals picking up the lead.

Ikigai does not present this rotation as a verdict on AI. The fund reads it as investors slowly moving beyond narrow AI leadership, which can coexist with continued AI investment by the largest companies. The distinction matters: rotation in the equity market is not the same as retrenchment in capital spending, and the hyperscaler capex number has continued to climb even as leadership has broadened.

Ikigai’s note lands at a moment when several AI-related catalysts are competing for attention. Hyperscalers are guiding to higher capex, AI start-ups are signing long-dated compute contracts, and several Asian chipmakers are reporting record quarters. Tibrewal’s argument is that each of these facts is being received as further confirmation of a trend that may already be priced in, which is the asymmetry he wants investors to notice.

What Tibrewal Is Not Saying

The framing in Ikigai’s newsletter is careful. Tibrewal is not calling a top, is not naming a date for a correction, and has elsewhere described himself as 99.5% invested through the April 2026 selloff. A separate Moneycontrol interview reports that he deployed capital at that level in the first week of April 2026, calling it only the second time he had run that concentrated. He has also pushed back on fears that AI will disrupt Indian IT companies, on his own firm’s published media appearances, arguing that the terminal-value debate is overdone.

The newsletter’s conclusion is that discipline matters more when optimism is abundant, not that investors should sell. The fund calls AI a transformational technology and says the risks it catalogues are conditional on investors beginning to question whether the unprecedented investments can generate adequate returns. Tibrewal’s prior comments on the rotation toward non-US markets, resources and India suggest that his prescription, when he gives one, is to reposition rather than retreat.

What Ikigai has now published is best read as a stress test, not a forecast. The indicators it lists are real and dated. The interpretation it offers is that more of them are flashing at once than at any point since the late 1990s, and that the gap between the AI story and the AI price has narrowed enough to matter.

Frequently Asked Questions

What is Ikigai Asset Manager saying about AI stocks?

Ikigai’s quarterly newsletter argues that AI investor enthusiasm has reached a “zone of visible excess,” with several indicators simultaneously flashing caution. The firm does not call a market top.

Which indicators does the fund point to?

Nasdaq 100 returns of more than 640% over the past decade, an S&P 500 CAPE above 35 times, US broker margin debt above $1.3 trillion, top-10 S&P 500 concentration near 40%, and hyperscaler capex expected at about 92% of operating cash flow by 2026.

How much is being spent on AI infrastructure?

Alphabet, Amazon, Meta and Microsoft are estimated to spend around $700 billion on capex this year and more than $800 billion next year. Including Oracle, OpenAI, Anthropic and other AI providers takes the figure past $1 trillion next year, equivalent to roughly 3% of US GDP.

Is Tibrewal calling a crash?

No. The fund says none of the indicators necessarily signal an imminent correction and stresses that AI itself will keep transforming businesses. The risk it names is investors starting to question returns on AI investment.

What does Tibrewal say investors should do?

The newsletter says discipline becomes more important when optimism becomes abundant. Tibrewal is not recommending selling; he is asking how much of AI’s future is already in today’s prices.

Disclaimer: This article discusses market commentary from a single fund manager and is for informational purposes only. It is not investment advice. Markets carry risk, and figures cited are accurate as of publication on July 10, 2026. Consult a qualified financial professional before making investment decisions.