Aswath Damodaran argues that Nvidia and OpenAI, while dominant today, mirror the Cisco and Sun Microsystems of the dot-com era — infrastructure plays that capture early attention but not the enduring value. According to his thesis reported by Livemint, the largest AI fortune will accrue to a consumer-facing application layer company that likely has not been built yet, echoing how Amazon and Google eclipsed the firms that sold them bandwidth and servers.

Here is a number that should keep every Nvidia bull awake at 2 a.m.: Cisco Systems, the company that literally built the plumbing of the internet, peaked at a market capitalisation of roughly $555 billion in March 2000. A quarter-century later, it trades below that peak. Amazon, the company that merely used those pipes, is worth north of $2 trillion. The infrastructure king made the revolution possible; the application queen kept the receipts.

Aswath Damodaran — the NYU Stern professor whose valuation models are treated as scripture from Dalal Street to Wall Street — is now making the sharpest version of this argument for the AI era. According to his thesis, as reported by Livemint, Nvidia and OpenAI are not the 'next Amazon.' They are the next Cisco and Sun Microsystems: brilliant, indispensable, and ultimately not the address where the largest pile of AI wealth will reside.

The claim sounds almost heretical when Nvidia is trading at a market cap that dwarfs the GDP of most nations and OpenAI is reportedly valued above $150 billion. But Damodaran's logic is not about whether these companies are good — it is about where the permanent, compounding, generational wealth settles once the dust of a technology revolution clears. And history, he argues, has a brutally consistent answer.

The Pattern That Never Breaks

Every major technology cycle since the mainframe has followed the same three-act structure, and Damodaran traces it with the patience of a man who has watched four of them unfold in real time.

Act One: The Infrastructure Boom. The companies that build the foundational layer — the chips, the cables, the protocols, the operating systems — capture the first wave of investor enthusiasm. They post genuine, extraordinary revenue growth. Their stock charts go vertical. In the internet era, this was Cisco, Sun Microsystems, Nortel, and JDS Uniphase. In AI, it is Nvidia, to a degree AMD, and the hyperscaler capex arms race among Microsoft, Google, Amazon, and Meta — companies that have collectively committed over $200 billion annually to AI infrastructure, according to their latest earnings disclosures reported by Reuters.

Act Two: The Margin Squeeze. Competition enters. The foundational technology commoditises. Margins, which looked like permanent moats, turn out to be temporary premiums earned for showing up first. Cisco's gross margins in networking equipment declined steadily once Huawei, Juniper, and a dozen others learned to build routers. Damodaran's thesis, as Livemint reports, suggests the same dynamic is already visible in AI chips: AMD, Intel, Google's TPUs, Amazon's Trainium, and a swarm of AI chip startups are all racing to erode Nvidia's GPU pricing power. OpenAI faces an even more crowded field — Anthropic, Google's Gemini, Meta's open-source Llama, Mistral, and dozens of Chinese competitors including DeepSeek are all driving foundation model pricing toward zero.

Act Three: The Application Coronation. A company nobody was watching — or that did not yet exist during the infrastructure boom — builds the consumer-facing product that becomes indispensable. It rides the now-cheap infrastructure, captures a network effect, and compounds value for decades. Amazon was a bookstore during the router wars. Google was a Stanford research project. The application layer inherits the revolution at a fraction of the infrastructure cost.

Inside Talk

The conversation in investment circles — from Mint Street to Sand Hill Road — is shifting in a way the headlines have not yet caught up with. Fund managers who loaded up on Nvidia through 2023 and 2024 are quietly asking each other the same question: "What is the exit thesis?" The stock has delivered extraordinary returns, but Damodaran's framework forces a confrontation with an uncomfortable truth — terminal value.

In valuation, terminal value is the estimate of what a company is worth beyond the explicit forecast period, and it is driven by sustainable competitive advantages. The talk among analysts tracking semiconductor cycles, according to industry commentary reported by Bloomberg, is that Nvidia's current dominance in training chips faces a structural challenge: as AI workloads shift from training (where Nvidia's CUDA ecosystem is nearly unassailable) to inference (where the competitive field is far wider and price sensitivity is far higher), the margin profile changes. The whisper in trading desks is that Nvidia may end up like Intel — a company that dominated one era's architecture and spent the next era watching the margins compress.

OpenAI's position, meanwhile, draws a different kind of scepticism. Its reported $5 billion-plus annual revenue is impressive until you set it against the capital required to maintain frontier model leadership. The talk in venture corridors, per commentary tracked by The Information, is that OpenAI's fundraising cadence — round after round at escalating valuations — resembles less a self-sustaining business than a capital-consuming arms race where the prize is uncertain and the runway is measured in quarters, not decades. (This reflects industry chatter and unverified speculation, not confirmed fact.)

The Indian Investor's Blind Spot

This is where Damodaran's thesis hits closest to home for India Herald's readership. Indian retail investors have dramatically increased their exposure to US technology stocks. The Reserve Bank of India's Liberalised Remittance Scheme (LRS) data shows outward remittances for investment purposes have grown steadily, and a significant portion of that flow, per analysis reported by The Economic Times, has been directed toward US tech — particularly the "Magnificent Seven" stocks, of which Nvidia is a flagship.

The risk Damodaran is naming is not that Nvidia will collapse. It is subtler and more dangerous: that investors are paying a price that assumes Nvidia captures the ENTIRE AI value chain in perpetuity, when history shows the value chain always disaggregates. The Indian retail investor buying Nvidia at 60-70x earnings is implicitly betting against Damodaran's pattern — and against the pattern of every technology revolution in the last half-century.

India Herald's read of the deeper incentive structure here is this: the real wealth transfer in AI is not from old economy to new economy — it is from early-cycle infrastructure investors to late-cycle application investors. The money that pours into Nvidia and OpenAI today is, in Damodaran's framework, effectively subsidising the infrastructure that will be used at commodity prices by the application company that captures the permanent value. The infrastructure investor pays for the revolution; the application investor inherits it.

So Who Is the Next Amazon?

This is the question Damodaran refuses to answer — and his refusal is itself the most important part of the thesis. The point is not to name the winner. The point is that the winner, by definition, cannot be named yet. Amazon was not investable during the Cisco boom. Google had not IPO'd. The next Amazon of AI is, right now, probably a startup with fewer than 50 employees, or an idea that has not yet been incorporated, or a pivot that a struggling company has not yet made.

What Damodaran does specify, according to the Livemint report, is the characteristics this winner will have: it will operate at the consumer or enterprise application layer, not the infrastructure layer. It will use AI as a capability, not sell AI as a product. It will build a network effect or a data moat that compounds over time. And it will likely appear in a sector that seems completely unrelated to "AI" as currently understood — just as Amazon was a "retail" company, not an "internet" company, and Google was an "advertising" company, not a "search" company.

The candidates that the market currently treats as application-layer plays — companies attempting to embed AI into healthcare, education, legal services, financial advice, creative tools — are interesting but unproven. The history Damodaran cites suggests that the ultimate winner may not even be from a category investors are currently watching.

What This Means for the Next Twelve Months

The forward projection, in India Herald's assessment, is that Damodaran's thesis is not a call to sell Nvidia tomorrow. Infrastructure booms can last years, and Nvidia's revenue growth remains extraordinary by any standard. The thesis is a call to understand what you own and why — and to recognise that the price you pay today encodes an assumption about permanence that technology revolutions systematically punish.

Watch for three signals in the coming quarters: first, Nvidia's inference-to-training revenue mix, which will reveal whether the margin-rich training monopoly is diversifying into the more competitive inference market. Second, OpenAI's path to profitability — or lack thereof — which will test whether foundation models can be a business or only a cost centre. Third, and most importantly, the emergence of consumer AI applications with genuine retention and network effects — the signal that Act Three has begun.

The most consequential question for any investor in AI right now is not "which infrastructure stock do I buy?" It is the question Damodaran keeps posing and the market keeps ignoring: when the pipes are cheap and the chips are commodities, who builds the thing that a billion people cannot live without?

Reported and written with AI assistance under India Herald's editorial standards; a human editor governs publication.

This report is journalistic, not investment advice; markets carry risk.

Key Takeaways

• Aswath Damodaran argues that Nvidia and OpenAI parallel Cisco and Sun Microsystems — infrastructure plays that boom first but do not capture the permanent, compounding AI wealth.
• Every major tech cycle (mainframes, PCs, internet, mobile) followed the same arc: infrastructure builders peak early, then an unknown consumer-application company inherits the revolution at commodity infrastructure prices.
• Indian retail investors have significantly increased US tech exposure via LRS, and many are implicitly betting against Damodaran's historical pattern by paying 60-70x earnings for Nvidia.
• The real AI wealth transfer is from early-cycle infrastructure investors to late-cycle application investors — the infrastructure money effectively subsidises the cheap pipes the eventual winner will use.
• Damodaran says the 'next Amazon' of AI will use AI as a capability, not sell it as a product, and will likely emerge from a sector not currently associated with AI.

Frequently Asked Questions

Why does Aswath Damodaran say Nvidia is not the next Amazon?

Damodaran argues that Nvidia is an infrastructure play analogous to Cisco during the dot-com era. Historically, infrastructure companies capture the early premium in tech revolutions but do not retain the permanent, compounding wealth — that accrues to consumer-facing application companies like Amazon and Google that build on the now-cheap infrastructure.

What kind of company does Damodaran predict will capture the most AI wealth?

According to his thesis reported by Livemint, the biggest AI winner will operate at the consumer or enterprise application layer, will use AI as a capability rather than selling it as a product, and will build a durable network effect or data moat — likely in a sector not currently associated with AI.

Should Indian investors sell Nvidia stock based on Damodaran's thesis?

Damodaran's thesis is not a short-term sell call. Infrastructure booms can persist for years. However, it warns that paying premium valuations (60-70x earnings) for Nvidia implicitly assumes permanent dominance — an assumption that every previous tech cycle has punished. This report is journalistic analysis, not investment advice.

What is the historical pattern Damodaran cites for tech wealth transfers?

In every major cycle — mainframes, PCs, the internet, mobile — the infrastructure builders (IBM, Cisco, Nokia) boomed first and faded or stagnated, while an initially obscure application-layer company (Microsoft, Amazon, Apple) captured the durable, generational wealth.