Picture courtesy: LinkedIn
During my recent participation in several technology conferences and meetups in Bangalore, I had the opportunity to engage in extensive discussions with India's leading tech executives, startup founders, and AI researchers. Away from the spotlight of keynote speeches, these tech leaders shared remarkably candid concerns about the AI investment frenzy. From CTOs of major IT services companies to founders of AI startups, the consensus was surprisingly sobering: while everyone acknowledged AI's transformative potential, most expressed deep skepticism about the current disconnect between massive investments and actual returns.
One senior executive from a prominent Indian IT firm put it bluntly: "We're implementing AI for our clients, but when we ask about ROI, the room goes silent." Another founder admitted their startup raised funding simply because they had 'AI' in their pitch deck, not because of a clear path to profitability. These conversations, combined with emerging data from global markets, paint a concerning picture of an industry racing toward what many privately fear could be a devastating correction. Here's what my discussions revealed about the current state of the AI bubble and why history appears to be rhyming once again with the speculative excess of the dot-com era.
The $400 billion question
The artificial intelligence industry stands at a critical juncture in 2025. Global corporate AI investments have exceeded $330 billion, with tech giants planning to spend approximately $400 billion on AI infrastructure this year alone. JP Morgan Asset Management notes that AI-related stocks have accounted for 75% of S&P 500 returns, 80% of earnings growth, and 90% of capital spending growth since ChatGPT launched in November 2022. These staggering figures represent more spending than the entire Apollo moon program, compressed into mere months rather than decades.
The math of this investment boom presents a sobering reality. For the current spending to make economic sense, AI revenues must grow from $20 billion to $2 trillion annually by 2030. This hundred-fold increase requirement has economists and market analysts drawing uncomfortable parallels to previous technology bubbles, particularly the dot-com crash of 2000.
The four pillars of a bubble
Economists Brent Goldfarb and David Kirsch identified four key signals that transform technological innovation into speculative bubbles. The AI sector currently exhibits all four characteristics with alarming clarity.
The fundamental uncertainty surrounding AI's business model represents the first warning sign. Despite massive investments, companies like OpenAI continue operating at substantial losses. OpenAI expects $13 billion in revenue in 2025, while Anthropic recently told Reuters that its "annual revenue run rate is approaching $7 billion”. Both companies are still losing billions of dollars a year. An MIT study revealed that 95% of companies adopting AI have seen zero measurable financial returns, with projects either stalling or failing to generate profits despite considerable effort.
The second signal manifests in the concentration of investment in pure AI plays. More than 1,300 AI startups now have valuations of over $100 million, with 498 AI "unicorns," or companies with valuations of $1 billion or more. In the first quarter of 2025, 71% of global venture capital funding flowed to AI-linked startups, creating a dangerous dependency on a single technological narrative.
The third component involves retail investor participation driven by fear of missing out. Nvidia became the world's most bought retail stock in 2024 and 2025, with U.S. retail investors alone pouring $30 billion into the company. Platforms like Robinhood and E*Trade prominently feature AI stocks in their trending lists, echoing the patterns of pets.com and webvan.com during the dot-com era.
The fourth signal revolves around the compelling narrative that AI will solve everything from climate change to cancer. Industry leaders like Sam Altman, Mark Zuckerberg, and Sundar Pichai consistently promote the message that AI represents a fundamental transformation of human civilization. This narrative creates an illusion that missing today's investment opportunity means eternal regret tomorrow.
The circular financing web
Perhaps the most concerning aspect of the current AI boom involves the intricate web of circular financing among major players. OpenAI is taking a 10% stake in AMD, while Nvidia is investing $100 billion in OpenAI; and OpenAI also counts Microsoft as one of its major shareholders, but Microsoft is also a major customer of AI cloud computing company CoreWeave, which is another company in which Nvidia holds a significant equity stake.
This circular pattern creates artificial demand and inflated valuations. Nvidia invests billions in OpenAI, which then uses that money to purchase Nvidia chips. OpenAI has signed a $300 billion deal with Oracle for data centers, and Oracle subsequently buys Nvidia chips to power those facilities. Microsoft accounts for nearly 20% of Nvidia's revenue while simultaneously being a major OpenAI shareholder.
The AI hyperscalers are using accounting tricks to depress their reported infrastructure spending, which has the effect of inflating their profits. The big AI firms are also shifting huge amounts of AI spending off their books into SPVs, or special purpose vehicles, that disguise the cost of the AI build-out. These financial engineering tactics mirror the collateralized debt obligations and subprime mortgage-backed securities that characterized the 2008 financial crisis.
The infrastructure arms race
The scale of current AI infrastructure spending defies historical precedent. Trillions are being poured into AI infrastructure, with projections for 2025 alone nearing $400 billion. Microsoft has projected capital spending of $125 billion in 2025, while Meta plans $70-72 billion for the same period with "notably larger" increases anticipated for 2026.
In the second quarter of 2025, capital expenditures by the five U.S. tech giants, Amazon, Google, Microsoft, Meta, and Apple, soared to $92.17 billion, a 66.67% jump compared to the same quarter last year. This spending spree has created what economists call a "GPU glut," where the supply of computing capacity may soon exceed actual demand.
The infrastructure boom extends beyond traditional tech companies. More than 100 non-tech global companies noted data centres on quarterly calls, including Honeywell, turbine maker GE Vernova, and heavy equipment maker Caterpillar. Goldman Sachs estimates global AI-related infrastructure spending could reach $3 trillion to $4 trillion by 2030.
Market concentration and systemic risk
The concentration of market value in AI-related companies has reached levels not seen since the dot-com era. The "Magnificent Seven" tech companies now account for nearly a third of the S&P 500's total market valuation. Just seven stocks — Alphabet, Amazon, Apple, Meta, Microsoft, Nvidia, and Tesla — have accounted for 55% of the S&P 500's gains since the end of 2022.
This concentration creates systemic risk. When so few companies drive market performance, any disappointment in their earnings or AI initiatives could trigger widespread market disruption. The Bank of England has warned that equity market valuations appear stretched, particularly for AI-focused technology companies, leaving markets exposed should expectations around AI's impact become less optimistic.
The Revenue-investment gap
The disconnect between AI spending and actual returns continues to widen. Meta increased capital spending by more than 100%, but its advertising revenue rose just 21.5% last quarter, and not all of that was due to AI. Despite technical achievements, modern AI models have yet to demonstrate their ability to generate sustained, significant revenue that justifies the massive infrastructure investments.
In August, investors were rattled after researchers at the Massachusetts Institute of Technology found that 95% of organizations saw zero return on their investment in AI initiatives. Harvard and Stanford researchers found that employees are using AI to create "workslop," which they define as AI-generated work content that masquerades as good work, but lacks the substance to meaningfully advance a given task.
The depreciation problem
Unlike traditional infrastructure investments, AI assets face rapid obsolescence. The useful life of AI chips is shrinking to five years or less, forcing companies to write down assets faster and replace them sooner. This accelerated depreciation cycle means companies need to generate returns much more quickly than with traditional capital investments, adding pressure to an already strained financial model.
GPUs become obsolete as new models demand more computing power. Data centres require constant upgrades. Energy costs continue escalating. The countdown clock for AI companies needing major capital infusions isn't just ticking—it's accelerating.
Historical parallels and divergences
The comparison with the dot-com bubble reveals both similarities and crucial differences. During the late 1990s, the NASDAQ Composite Index rose 500% in five years before crashing 80% between 2000 and 2002, wiping out approximately $5 trillion in market capitalization. Companies like Amazon saw their stock prices fall 90%, while many others simply ceased to exist.
However, today's AI leaders differ from their dot-com predecessors in important ways. Unlike the 1990s tech bubble that featured soaring stocks from unprofitable early-stage companies, strong mega-cap company earnings are driving this year's rally. The Magnificent Seven possess substantial revenue streams and large cash reserves, providing more resilience than the speculative startups of the dot-com era.
Yet the scale of investment dwarfs previous bubbles. The dot-com bubble saw telecom companies spend $121 billion annually at its 2000 peak. Today's AI spending levels dwarf that—and it's compressed into a shorter timeframe. AI-related capital expenditure now accounts for an estimated 1.1% of GDP growth, making it a critical component of overall economic expansion.
The China factor
Competition from China adds another layer of complexity to the AI bubble dynamics. Chinese companies are flooding the market with competitive, low-cost AI models that risk undercutting Silicon Valley on price in certain markets. While U.S. firms generally maintain technological leadership, the pricing pressure from Chinese alternatives makes it harder for Western companies to recoup their massive infrastructure investments.
The geopolitical dimension drives additional spending as nations race to establish AI supremacy. This competitive dynamic locks companies into an arms race where slowing investment means risking obsolescence should rivals achieve breakthroughs. Each major player feels compelled to maintain aggressive spending regardless of immediate returns.
Potential trigger points
Several factors could precipitate an AI bubble burst in 2025 or 2026:
Rising interest rates historically trigger reassessment of speculative investments. The Federal Reserve raised interest rates multiple times throughout 1999 and 2000. The federal funds rate climbed from around 4.7% in early 1999 to 6.5% by May 2000, making speculative investments less attractive as investors could earn higher returns from safer bonds. Any significant rate increases could similarly impact AI valuations.
The resolution of GPU scarcity could fundamentally alter market dynamics. As supply catches up with demand, Nvidia's pricing power may diminish. The company currently commands premiums of up to 300% over competitors for its chips. When this pricing advantage erodes, it could trigger a reassessment of the entire AI value chain.
For this not to be a stock market bubble, the spending and revenue gap for AI companies must close soon. The only realistic scenario where this can happen is if OpenAI, Anthropic, and Google get AI to replace substantial portions of white-collar work. Any significant delay in achieving this automation could prompt investor flight.
Global eionomic Implications
An AI bubble burst would have far-reaching consequences beyond Silicon Valley. AI infrastructure spending is now such a large component of GDP growth that its collapse would ripple through the broader economy, plunging us into a recession. The interconnected nature of modern financial markets means that a significant correction in AI stocks could trigger global market volatility.
The impact would extend to employment markets in complex ways. While an AI crash might slow the replacement of human workers in some sectors, companies have already committed to automation strategies. Low-end white-collar jobs in data entry, customer support, and junior coding would likely continue facing pressure regardless of AI company valuations.
Credit markets face particular vulnerability. Many private credit funds and banks have begun offering loans against future AI startup earnings. A valuation collapse could trigger a default chain reaction reminiscent of the 2008 financial crisis, though with machine learning models rather than mortgages at the center.
The India factor
India's position in the global AI ecosystem presents both risks and opportunities. The country's IT services giants—TCS, Infosys, Wipro, and HCL—implement AI solutions for global clients, making them vulnerable to any downturn in U.S. and European tech spending. By mid-2025, renewal rates for AI consulting projects at major Indian IT firms have already begun slowing.
Indian AI startups raised nearly half a billion dollars in the first seven months of 2025, the highest amount in five years. However, global investors like Tiger Global and SoftBank are already reducing their exposure to India's generative AI deals, signaling growing caution about the sector's sustainability.
The employment impact could be severe. India's IT and BPO sector directly employs 5-6 million people, many in roles targeted for AI automation. An AI crash wouldn't save these jobs; automation has become a permanent business imperative regardless of AI company valuations.
However, India might benefit from an AI crash in unexpected ways. Currently four to five years behind in AI adoption compared to the U.S., India could have a valuable observation window. Technologies that prove genuinely valuable after a market correction—in healthcare, agriculture, logistics, and education—could be adopted at lower costs. The dot-com crash preceded India's IT outsourcing boom; a similar pattern might emerge from an AI correction.
Survival strategies
Not all AI investments are equally vulnerable. Companies with genuine technological moats and diversified revenue streams stand better chances of weathering a correction. Despite 1,500% gains over the last three years, market leader Nvidia has a P/E ratio of 52. While that is above the S&P 500's average of 31, it hardly represents bubble territory. Companies like Alphabet and Meta maintain P/E ratios below 30, suggesting more reasonable valuations.
The key differentiator will be companies that can demonstrate clear paths to profitability rather than relying on narrative alone. Those providing essential infrastructure—cloud services, semiconductors, networking equipment—may prove more resilient than pure-play AI startups with no clear monetization strategy.
The accountability moment
The AI boom has reached what marks a critical inflection point, where the initial euphoria surrounding AI's potential is giving way to a more pragmatic assessment of its financial viability. Investors increasingly demand tangible returns rather than promises of future transformation.
Recent earnings seasons have highlighted this shift. Meta's stock plummeted 11-13% despite beating revenue expectations, purely due to concerns about its aggressive AI spending plans. The market no longer accepts blank checks for AI development without clear monetization strategies.
Beyond the bubble
History suggests that technological revolutions follow predictable patterns: initial breakthrough, speculative boom, crash, then eventual genuine transformation. The internet did change the world, just not as quickly as or in the ways early investors imagined. Railways did transform commerce and society, but not before multiple boom-bust cycles destroyed fortunes.
AI likely follows a similar trajectory. The technology's transformative potential remains real, but current valuations and investment levels assume an immediacy and scale of impact that history suggests is unrealistic. Like the 19th-century rail tracks and the 20th-century broadband Internet build-out, AI will first rise first, then crash, and eventually transform the world.
The question isn't whether AI will transform the economy—most experts agree it eventually will. The question is whether current infrastructure investments can generate returns before investor patience expires. With companies needing to generate $2 trillion in annual revenue by 2030 to justify today's spending, the math of sustainability becomes increasingly challenging.
The reckoning approaches
The AI bubble exhibits all the classic signs of unsustainable speculation: circular financing, extreme valuations, massive infrastructure overbuilding, and a widening gap between investment and returns. While AI technology will undoubtedly play a crucial role in the future economy, current market dynamics suggest a painful correction lies ahead.
The scale of this potential correction dwarfs previous technology bubbles. With AI spending approaching 2% of global GDP and market concentration at historic levels, a burst would send shockwaves through the entire global economy. Unlike the dot-com crash, which primarily affected technology investors, an AI correction would impact employment, credit markets, and economic growth across all sectors.
Yet within this cautionary tale lies opportunity. Companies and countries that prepare for the correction, maintain financial discipline, and focus on genuine value creation rather than speculative narratives will emerge stronger. The technologies that survive the crash will form the foundation for the next phase of economic development, just as Amazon and Google emerged from the dot-com rubble to dominate the digital economy.
The AI revolution will continue, but first must come the reckoning. Those who recognize the signs and prepare accordingly will be best positioned to capitalize on both the correction and the eventual recovery. History doesn't repeat, but it certainly rhymes—and the rhyme scheme of technology bubbles remains remarkably consistent across centuries.