THE GREAT AI GAMBLERS 

October 8, 2025

How the Indian Prime Minister is Steering us through the maze

Parneet Sachdev

The AI investment supernova is underway. The race is on to pour in infrastructure, chips and models that might someday realize the dream of artificial general intelligence (AGI). Yet amid the euphoria lies a question that few want to ask: what if this bet fails to deliver? What are the odds of a convulsive crash?

HOW BIG THE BET REALLY IS

The scale of AI investment is already titanic. In 2024 alone, corporate spending on AI projects globally hit USD 252.3 billion, surging nearly 45 percent year over year. Meanwhile, generative AI startups alone pulled in USD 33.9 billion in private funding(Stanford HAI).

But beyond projects and software, the real capital war is being waged in compute infrastructure. McKinsey estimates that by 2030, the world will need USD 6.7 trillion in new compute and data-center capital, with USD 5.2 trillion of that dedicated to AI workloads.Other estimates place the required investment even higher — up to $7 trillion by 2030(TechBlog).

Already, a venture named Stargate—backed by OpenAI, Oracle, SoftBank and others—aims to mobilize roughly USD 500 billion over coming years to scale AI infrastructure.³ Major cloud and chip firms are structuring multibillion-dollar deals around AI so intimately that their fate hinges on the success of the models they host(Stargate LLC).

In 2024, corporate AI investment globally reached $252.3 billion — up nearly 45 percent year over year. Meanwhile, generative AI alone pulled in $33.9 billion in private investment during the same year(FTI Consulting).As per the Guardian, in September 2025, Nvidia announced it would invest $100 billion into OpenAI, effectively tying its own fortunes to that of its AI partner.

Some analysts estimate that an AGI breakthrough could spark annual growth rates of 15–20 percent or more across global productivity. You simply need to be the firm that “owns the keys” to reap near-monopolistic returns.

The logic goes; when the risks are existential, timid spending loses. If you don’t build a gargantuan model now, someone else will, and you’ll be left behind.

THE FLIP SIDE AND THE PAIN

But the flip side of that logic is terrifying. Data centers are capital-intensive. A recent Bain & Company report argues that by 2030 the AI sector will need $2 trillion in recurring revenue just to cover the cost of compute scaling — yet even the most optimistic scenarios leave an $800 billion shortfall. Similarly, Deutsche Bank warns of an $800 billion gap in infrastructure funding as AI investments race ahead of monetization ability(datacentremagazine).

If utilization falls short, billions in data center assets may become white elephants generating high maintenance costs with shrinking margins.Moody’s notes that the data center boom is driving default and demand uncertainty.

AI is voracious on energy. In 2024, global data centers consumed about 415 terawatt-hours (TWh) of electricity, roughly 1.5 percent of global demand. By 2030, that figure is projected to more than double to 945 TWh, mostly driven by AI workloads(IEA Energy and AI report). The IEA forecasts data center electricity demand will grow at roughly 15 percent per year from 2024 to 2030—more than four times faster than overall electricity demand.

Goldman Sachs projects that data center power demand will increase by 165 percent by the end of the decade compared to 2023 levels. A U.S. Department of Energy report noted that data centers accounted for 4.4 percent of U.S. electricity in 2023—and may rise to 6.7–12 percent by 2028.

Cooling, power delivery, land permitting, and transmission sit at the intersection of compute ambition and physical constraints. In some regions, utility upgrades or permitting delays may force data centers to idle or relocate.

THE RETURNS ON AI AND HISTORICAL LESSONS

The most damning evidence may come from the results of near-term AI deployments. According to a study reported by major media, 95 percent of companies have seen zero return on their generative AI investments(TOI).

Not all AI systems outperform humans and traditional automation.Some AI systems do not adapt well to context shifts, and have limits to generalization. A model may beat human benchmarks in controlled settings but fail in messy real-world deployment.If enough major AI projects underperform, the cascade could be brutal: loss of investor capital, infrastructure, layoffs, and even contagion to vast sectors.

The pattern—hyper-investment, speculative fervor, and eventual collapse or consolidation—is not new. The dot-com boom saw massive capital flowing into network infrastructure and content plays before many collapsed. The cleantech bubble of the late 2000s similarly featured overinvestment in solar firms before consolidation.

Yet the AI wave is different in scale and centrality: this time, the bet is not just on a new product class, but on the core engine of future productivity. The halo effect of AI also pulls in secondary sectors — electric utilities, real estate, chip frontiers, cooling systems — meaning the losses, if misallocated, may ripple widely.In such a climate, if you hesitate in investing, someone else will build a moated lead. However, one fact stares us on our faces as per the Economist

“Even if the technology succeeds, plenty of people will lose their shirts….

We are in the early innings of a war for compute, capital, and talent.”

Throughout history, speculative bubbles in technology sectors yield spectacular gains—and brutal retrenchments. The dot-com boom of the late 1990s offers a sobering analogue.Between 1995 and March 2000, the Nasdaq composite rose nearly 600 percent. Within two years post-peak, it sank 78 percent, erasing most gains.Many high-profile startups failed outright: Pets.com, Boo.com and Webvan declared bankruptcy.Others—like Cisco and Amazon—survived but saw stock values fall 80–90 percent.

In total, the tech collapse wiped off USD 5 trillion in market capitalization by late 2002 from the peak.

Unlike the dot-com era, the AI stakes are woven more deeply into infrastructure, energy, real estate and national strategy—and that only magnifies the systemic risk.

HOW THE INDIAN PRIME MINISTER IS STEERING THE COUNTRY THROUGH THIS MAZE

India stands at a uniquely paradoxical crossroads in the global AI boom. On one hand, it is the world’s fastest-growing digital talent base and the second-largest pool of AI and data engineers after the United States. Major Indian IT firms—Infosys, TCS, Wipro, HCL—are committing billions to AI platforms and cloud infrastructure, while startups in Bengaluru, Hyderabad and Gurgaon attracted more than $5 billion in AI-related funding between 2022 and 2024. Yet India is constrained by patchy energy availability, inadequate data-center capacity, and a still-evolving regulatory framework. Power demand from AI data centers is projected to grow over threefold by 2030, but India’s grids already struggle with peak-load shortages and transmission losses of 15–20 percent in several states. Unlike the U.S. and China, India does not have the capital depth to absorb large-scale AI write-offs if the boom turns.

India’s tech giants rely heavily on service-based revenue from Western clients rather than proprietary AI models; that makes them very vulnerable to tariffs and barriers. In essence, India is racing to plug into the AI value chain—through chips, cloud zones, and talent exports—while lacking the trillion-dollar buffers that cushion the U.S. or China. If the global AI frenzy yields strong returns, India could emerge as a crucial beneficiary of outsourced development. But if the bubble deflates, it will feel the shock through collapsing demand for its IT services, stalled data-center investments, and intensified competition for capital at home.

The infusion of capital into AI today is nothing short of historic given the promise of AGI, transformational productivity, and limitless frontiers.

In the coming decade, whether AI becomes the defining engine of a new era—or a cautionary tale of overreach—will depend on which side of that mirror investors, companies, and nations fall.

(Views expressed are the author’s own).

Parneet Sachdev, IRS is the Chairman of Real Estate Regulatory Authority and a leading author.