AI spending is soaring, but revenue is lagging: Here's why Indian investors should pay attention

Artificial intelligence (AI) has become the biggest driver of global stock markets, with companies spending hundreds of billions of dollars to build AI infrastructure. While investor excitement remains high, experts believe the economics behind the AI boom are far less convincing than the market suggests.

Karan Aggarwal, Co-Founder & CIO at Ametra PMS, says the gap between AI spending and the revenue it currently generates raises important questions for investors, especially those investing in global equity funds with significant exposure to US technology stocks.

AI spending is racing ahead of revenue

“The gap between approximately $725 billion of AI capital expenditure and only $50-60 billion of current AI revenue matters because it translates into less than $1 of revenue earned for every $10 invested annually. Importantly, this is revenue, not profit,” Aggarwal mentioned.

In simple terms, companies are investing heavily today in the hope that AI will become highly profitable years later. Whether those returns eventually materialise remains uncertain.

Cash flows are coming under pressure

The AI race is also becoming expensive for the companies leading it. Aggarwal points out that most large hyperscalers are using nearly all of their free cash flow to fund AI expansion, while some are borrowing money or raising fresh equity to continue investing.

“At present, AI capex is accounting for 94% of free cash flow of hyperscalers, and this number is projected to rise to 157% by 2030. Hyperscalers would become debt-laden companies burning cash into unknown and unproven growth avenues,” he said.

For investors, this means today's earnings may increasingly be sacrificed in pursuit of uncertain long-term gains.

Adoption is growing, but monetisation remains weak

AI is being adopted across industries, but generating meaningful revenue from it has proved far more difficult.

Citing an MIT NANDA study reported by Fortune, he said 95% of generative AI pilots fail to reach production. The consumer story is equally bleak. According to NPR, only 3% of consumers actually pay for AI services.

He also pointed to Sequoia Capital's estimates showing the AI revenue shortfall widening from $125 billion in 2024 to nearly $600-700 billion today, highlighting how infrastructure spending continues to outpace demand.

A lesson from the dot-com era

Aggarwal believes the current environment bears similarities to the internet boom of the late 1990s.

“It seems very similar to the internet bubble of the early 2000s, where the internet saw widespread adoption but nearly 80% of internet firms went out of business,” he said. Even the internet boom took nearly a decade before companies began generating meaningful returns.

He also warns that AI remains a “winner-take-all” industry. It would take one low-cost and equally efficient model from China to put trillions of AI capex going into useless junk.

The AI rally carries risks for Indian investors

Many Indian investors gain overseas exposure through global mutual funds, ETFs or international feeder funds. The challenge, Aggarwal says, is that these portfolios are increasingly concentrated in a handful of large US technology companies.

He notes that the top 10 companies account for more than 40% of the S&P 500's market capitalisation, compared with a historical average of 20-25%, while most global equity funds have nearly 70% exposure to the US.

Several emerging-market funds also have around 45% exposure to Taiwan and South Korea, with three companies accounting for almost 32% of the portfolio weight.

Aggarwal also believes US equity valuations leave little room for disappointment. The US market-cap-to-GDP ratio is above 220, far above its historical comfort range of 80-120, while the S&P 500's price-to-earnings ratio is close to 30, levels last seen during the dot-com era.

“With the global equity boom revolving around unrealistic revenue expectations around AI adoption and the largest capex cycle in human history, there is very little margin of error. Any setback in expectations can trigger a 2000 redux,” he said.

Aggarwal noted that “Going by the precedent of big crashes in US markets in the last 25 years – be it 2000 or 2008, Indian markets can also see sharp valuation compression on account of bearish global cues, which indicate that Indian investors must remain cautious against negative global trends.”

Key takeaways for investors

“When it comes to global investing, it is time for a more nuanced approach towards geographic diversification, with a focus on value plays across a wide universe of emerging and developed markets,” Aggarwal said.

He added that AI-led euphoria has made global markets increasingly concentrated, making broad-based index investing alone a less effective strategy for global diversification.