Skip to main content

Is AI a Bubble? Follow the Funding, Not the Returns!

A question popped up in a WhatsApp group recently: What is the definition of an AI bubble? When everyone says AI is a bubble, do they mean it’s a scam? A temporary fad? Or is it simply a massive concentration of funds and top talent? The answers—even from venture capitalists in the group—revolved around a familiar refrain: “A bubble is when returns don’t justify the investments.” That definition is completely upside down.

 

Returns failing to justify investments is the result of a bubble, not its cause. A market bubble is something fundamentally different. It occurs when the sum total of investments, the companies operating in a space, and the superstructure built to support it are based on something other than the core fundamentals of that particular business vertical.

 


 

 

When that equation goes awry, a bubble develops. Given the long gestation nature of industries like AI, evaluating them on immediate valuations and short-term returns is sub-optimal because the real returns sit far in the future. Whether an industry is in a bubble is determined by business basics: Product, Price, Place, Promotion, Growth Curve, Customer Lifecycle Stage, and the surrounding ecosystem.

 

The Anatomy of a Market Bubble

To understand whether AI—or any sunrise industry—is heading for a bust, we have to separate the core technology from the financial architecture propping it up. Here is how a bubble actually forms, step by step:

Phase 1: Technical Proof Without a Market

1.          A new tech field, a major shift in an established industry, or a blue-ocean category emerges.

2.          One or two companies make waves based on solid proof of concept.

3.          Others notice and rush in across a series of waves, each built on the technical proof established by the previous wave.

4.          At this stage, all proof is strictly confined to tech development, scientific research, and product options. There is zero customer focus anywhere, and funding comes from a limited pool of specialist investors.

Phase 2: General Capital Enters Before Customers Do

5.          These consecutive waves generate visible traction, catching the attention of the broader investment community.

6.          Capital requirements rise sharply for the operating companies.

7.          General investors step in to fund them. This is where the trouble begins—because no real market exists yet.

8.          As general market investors enter, financial risk spreads across the broader system through leverage and interconnections. The spectre of a bubble appears here: bubble risk is price running above fundamentals.

9.          Early innovators do not constitute a marketable segment by any business fundamental; there is simply no volume to justify a sustainable business.

10.      Developments continue on the tech, science, and product fronts while customer adoption naturally lags behind.

Phase 3: The Structural Mismatch

11.      Those technical developments spur even larger investment tranches. Bubble risks escalate because there is still no mature commercial market.

12.      Interest payments and operational costs continue to mount across both companies and investors.

13.      No supporting ecosystem exists yet—and no business can survive long-term without a complete ecosystem around its product or service.

14.      General market investors operate on short-term horizons, meaning short- to mid-term funds are now locked into long-gestation projects.

15.      The bubble takes a defined shape. Mounting operational costs combined with this fund-structure mismatch become the second-to-last straw.

Phase 4: Resolution or Rupture

16.      At this crossroads, the cycle can either continue or smoothly subside if ecosystem development and market adoption catch up. If they do, all is well.

17.      However, if unrelated developments in external markets trigger a liquidity upheaval and force investors to pull out, the system snaps. Because financial risk was spread across general markets, the bubble can no longer sustain itself—and it bursts.

Phase 5: The Aftermath

18.      Funds are pulled, development slows, and the weaker players fail.

19.      Assets and talent consolidate with the survivors, or move to other fields.

20.      Adoption resumes at lower prices if customers had already found traceable value. If not, it stalls.

 

So, Is AI a Scam, a Fad, or Just a Concentration of Funds?

Is it a scam? It doesn’t seem to be. A scam has a divergent blueprint, as the anatomy above shows: it rests on a fundamental premise that is fictional. AI’s premise is real, and the technology works. There are scams inside the AI wave, such as AI-washing, but that is the wrapper, not the core.

 

Is it temporary? Maybe, maybe not. Blockchain showed that a technology can survive while the hype does not. It depends on market adoption at real business scale, and that has not happened yet. Reports put adoption at roughly 18–33%, depending on what is counted, which against the scale of investment is nowhere near enough. At this investment scale, I would want to see 40–60% of companies paying and renewing annually soon enough to service the capital already committed.

 

Is it a concentration of funds and talent? For sure, and that part is beyond debate. But concentration alone doesn’t make a bubble. It becomes dangerous when funding is short-term and gestation is long, the mismatch in points 14 and 15 above.

 

Bubbles Aren’t Bad—Blindness to the Ecosystem Is

In a sunrise industry, bubbles by themselves are not inherently bad. They are often what fund the heavy lifting of early infrastructure. What makes them go bad is the structural mismatch between short-term capital and long-gestation business fundamentals.

 

More importantly, the collapse rarely begins in the core industry itself. Think of the 2008 Subprime crisis: bad mortgages alone would have meant a US housing correction. What made it a global rout was the financial structure the money flowed through. Similarly, when a tech bubble bursts, the financial structure funding the industry decides how far and how fast the damage spreads, usually after an unlinked external shock.

 

India has seen both halves of this. The infrastructure NPA cycle of the 2010s was long-gestation projects funded by shorter-term bank liabilities. The telecom shake-out kept the sector growing while heavily indebted players fell, and survivors like Jio and Airtel came out stronger.

In the real world, an unrelated external risk will always arise eventually. That is why tracking a bubble requires looking far beyond valuations, ROI, or the technology itself. To know where the real risk lies, you have to trace the financial spread, the maturity of the supporting ecosystem, and the movements of related industries across the entire business landscape.

 

Where AI Stands Today

A bubble has two sides, value and finance, and they have to be read separately. The anatomy above was written for any sunrise industry, and AI has already moved past points 4, 7 and 9: it has customers and revenue. The market exists, but it is sub-scale against the capital committed.

 

On the value side, genuine value is being created, so this is not a case of value being outstripped by funding. But it is not yet enough to cover the cost of delivering it. That may be a technology weakness, since costs have to come down, and partly an industry that has yet to mature, with capacity being built ahead of demand. Press-reported figures put the two leading labs’ combined annualised revenue at around $105 billion, against hyperscaler capex guided at roughly $730–775 billion for 2026. Whether the value can justify current prices at scale is still open.

 

On the financial side, circular financial commitments exist among the companies, and pension funds, foreign portfolio investors and hedge funds are moving in. Contagion needs wide exposure across global financial structures, and so far that exposure is limited and concentrated. But size alone is the wrong yardstick; leverage and interconnection are what made subprime global. Morgan Stanley forecasts about $570 billion of AI-related debt issuance in 2026, the BIS reports that direct-lending funds now hold about 15% of their portfolios in AI and IT, and AI-linked stocks are roughly 45–49% of the S&P 500, depending on the definition. The IMF says banks would remain adequately capitalised in an AI-led correction, so the risk sits mainly in non-bank channels.

 

So: Bubble risk = price above fundamentals. Bubble Crisis or Burst = bubble risk plus fragile funding. AI may be showing the first, and the second is what I’m watching. It is too early to call a bubble, but not too early to see the risk. The aftermath of a burst matters too. Withdrawn funds slow development, and talent tends to concentrate in the survivors.

 

What to Watch

It takes the removal of but one brick, the keystone, to bring a house down. These are the bricks I would track:

•             Who holds the debt (banks, insurers, retail, hedge funds) and how concentrated it is

•             Debt maturity against asset life: GPUs last perhaps 4–6 years, while financing and data-centre commitments run longer

•             How much revenue is circular, with investors funding customers who buy from them

•             Revenue against capex and debt service

•             Paying, renewing enterprise customers on an annual recurring basis, not pilots or free tiers

•             Economic and geopolitical shocks that could make funds retreat to home markets.

 

Conclusion

AI is not a scam, and it is too early to call it a bubble. Real value is being created, and the risk lies in how the money is structured, not in the technology. Bubble risk means price above fundamentals; a crisis needs fragile funding on top. Today, exposure is real but concentrated. That can change if leverage and interconnection widen, or if an unrelated shock forces funds to withdraw. So watch the financing, not the headlines: who holds the debt, how long it runs, and whether paying customers arrive in time. Bubbles are not bad; blindness to the ecosystem is.

Comments

Popular posts from this blog

Tarkeshwar Mahadev : Pune Hidden Gems

What do you do when you have something good, something that is praiseworthy, and something that can be an attraction? Answer, if you are in Pune – keep silent about it, tell no one. This is seemingly exaggerated – perhaps it is exaggerated; but I am flabbergasted by a series of unbelievable locations that I have visited in Pune City – within main Pune City, mind you . These are not well known – at least not one single localite informed me, even on asking . At least those I talked. If I didn’t talk to the right people, perhaps I am in the wrong. But – if you expand your vision to TV, Cinema, Popular opinion, hotels – the situation above gets proof. I earlier visited Pune on a family holiday, stayed in a good Hotel. Not one Hotel informed me of these; not one person – Taxi, Tour Guide – even mentioned these . Thus, it seems to me that Puneites don’t realise how lovely a city they have, how mesmerizing are its many, many tourist-worthy places, how rich and unspoiled,...

Book Review - Stay Hungry Stay Foolish by Rashmi Bansal

This is the first Rashmi Bansal book I am reading, and I have to admit that her writing makes for an interesting read. Stay Hungry Stay Foolish is a very interesting and diverse collection of stories of Entrepreneurs – I wont say Case Studies, as the focus is on the individual rather than the company, the Brand or the strategy employed. And this is just about the only negative I can spot in an otherwise near-flawless book on Business in India. A worthy addition to any regular reader on Business, rated 4 stars. THE BOOK The book’s collection of Entrepreneurs is subdivided into 3 interesting heads : The Believers, The Opportunists, The Alternate Visionists. The first – the Believers – consists of examples of Entrepreneurs who went into Business straight after their MBA: the Opportunists went into Business when opportunity presented itself; and the third group – those with creative intent or social causes at heart. This classification makes it interesting, to say the least...

Drawing Lessons From The Rehn Model of Sweden

In the previous article, the first of this Risk Mitigation and Economic Planning Series I am writing, I made a strong play in favour of anticipating risks and actively planning for their mitigation. In this relatively short article today, I look at one model that was developed in the 1950s – The Rehn Plan, which was implemented in Sweden in the 50s to the 70s. I do not recommend this for India; this example is used to underscore and highlight how proper planning needs to proceed, and how we need to anticipate and plan rather than be reactionary; and how we need to develop solutions as per our own internal needs. In the normal scheme of things in the modern capitalistic context, readjustment of employment is considered a natural outcome, and that the market for labour will adjust accordingly. This is a very hard hearted approach, leaving the readjustment to market forces, which can result in loss of productive resources as well. The assumption   is that the individual...