Every major technological shift creates two stories. The first is about what technology can do. The second is about who can turn that capability into a durable, profitable business.
With artificial intelligence, the first story has moved at extraordinary speed. The second is only the beginning.
The AI Boom Has Already Reshaped the Market
Artificial intelligence feels as though it has been part of the business conversation forever. In reality, ChatGPT was released to the public only in November 2022. In less than four years, AI has moved from a technological curiosity to the centre of corporate strategy, capital expenditure and equity-market enthusiasm.
Nvidia became the defining supplier of the AI infrastructure boom. The ‘Magnificent Seven’ accounted for roughly 45% of the increase in the market capitalisation of all US stocks between 2022 and 2025. Meanwhile, spending on chips, data centres, cloud capacity and large language models became important enough to contribute materially to US economic growth.
The technology is real. Its impact is real. But neither of those facts automatically makes every AI company a good business – or every AI-linked stock a good investment.
A Big Market Is Only the First Step
Most bullish arguments begin with the total addressable market, or TAM. AI could assist workers, replace selected tasks, automate entire functions and eventually reshape industries across the world. Depending on which version of that future one assumes, the market can appear almost limitless.
That is precisely why TAM estimates must be treated carefully. A forecast of a $22 trillion AI market is not meaningful unless it explains what AI will replace, which workers or industries will be affected, how widely adoption will spread and who will actually pay for the products.
For investors, the logic has to move through four separate bridges:
- A large potential market must become actual customer spending.
- Customer spending must become company revenue.
- Revenue must generate sustainable operating profit.
- Profit must translate into cash flow after the investment required for growth.
The Business Model Is Already Changing
The first generation of consumer AI was built largely around subscriptions. That model works beautifully for traditional software because serving one more user often costs almost nothing. AI is different: every query, response and generated output consumes computing resources.
Unlimited usage at a fixed price can therefore turn a popular customer into a loss-making one. That is why the industry is moving towards usage-based pricing, particularly in enterprise products. Subscriptions may remain in the mass market, but they are likely to include limits, tiers or usage caps.
The open-versus-closed model debate also has a commercial dimension. Closed models can offer more control, customisation and customer stickiness, giving their owners greater pricing power. But they cost more to develop and maintain. Open models can spread faster and suit standardised applications, but may be harder to monetise at premium margins.
Cheaper Tokens Do Not Necessarily Mean Cheaper AI
The cost of an AI token has fallen sharply. On the surface, that should improve margins. Yet the products being built are also becoming more powerful, more output-heavy and more computationally demanding.
The result is a crucial split in the market:
- Mass-market AI will compete on scale and cost. Standardised tools can benefit directly from falling token prices, especially when users do not need the most powerful model for every task.
- Premium AI will compete on capability and specialisation. Products replacing expensive skilled labour may command higher prices, but their delivery costs could remain stubbornly high as each upgrade consumes more computing power.
- More powerful is not always more valuable. A model that delivers five times the capability a customer needs may simply be five times harder to monetise.
Where Will the Moats Come From?
Early technology leadership does not guarantee long-term dominance. As AI products become easier to compare, companies will need a durable reason for customers to stay.
In mass-market AI, the strongest advantages may come from lower unit costs, scale and proprietary access to useful data. In premium enterprise AI, the winners may be companies that combine technical capability with deep customer integration. A model built around a client’s workflows and private data becomes harder to replace – and therefore more valuable.
Brand may matter too, but trust will matter more. Businesses are giving AI systems access to internal data, strategic information and sensitive processes. In this market, trust will not be created by philosophical statements about safety. It will be earned through behaviour, reliability and the responsible handling of data.
The Costs Beyond Computing
AI is often discussed like a software business, but its physical footprint resembles an infrastructure build-out. Data centres require land, electricity, water and enormous upfront capital. As communities and governments react to that footprint, approvals may become slower and resources more expensive.
Other constraints are likely to grow as well:
- Data-privacy scandals could lead to tighter rules and higher compliance costs.
- Job displacement could trigger political pressure to preserve employment or compensate affected workers.
- The concentration of wealth created by successful AI companies could intensify calls for new taxes.
- Different regulatory regimes could make the same AI business far more valuable in one geography than another.
What Today’s Valuations Are Assuming
The cleanest way to test an ambitious valuation is to reverse-engineer it. Instead of asking what an AI company could be worth, begin with what investors are already willing to pay and calculate the future revenue and margins required to justify that price.
Consider the framework in the source analysis. At a hypothetical $2 trillion valuation, with a 30% after-tax operating margin and a 10% cost of capital, Anthropic would need approximately $1.2 trillion in annual revenue if the business matures in ten years. If maturity takes fifteen years, the required revenue rises towards $2 trillion.
Applied to the industry as a whole, an aggregate valuation of $5 trillion and a blended operating margin of 20% would require roughly $5 trillion in annual revenue after ten years – or more than $8 trillion if maturity takes fifteen years.
These calculations do not prove that AI is a bubble. They show how much success is already embedded in the price. Rapid growth from a small base is encouraging, but it is not enough on its own to justify any valuation.
A Practical Framework for AI Investors
Before valuing an AI company, investors should ask five questions:
- Market focus: Is the company targeting mass-market tools or premium, specialised applications?
- Unit economics: Does serving an additional customer improve margins, or does greater usage increase costs just as quickly?
- Competitive advantage: Is the moat based on scale, proprietary data, technology, integration, trust or regulatory protection?
- Reinvestment: How much more must the company spend on chips, data centres, talent and product development to deliver its promised growth?
- Regulation: How exposed is the business to restrictions on data, infrastructure, employment and geography?
The Big Picture
AI may prove to be one of the most important technological changes of our time. But technological importance and investment value are not the same thing.
The winners will not necessarily be the companies with the largest models, the loudest founders or the most dramatic TAM slides. They will be the businesses that identify a customer willing to pay, deliver the product at a sensible cost, build a defensible advantage and convert growth into cash flow.
This is still a young industry. Certainty is unavailable, and conviction without evidence is dangerous. Investors do not need to dismiss the AI revolution. They do need to price it like a business.
SEBI Registered Investment Advisor: Company Name: ORIM ADVISORS PRIVATE LIMITED. SEBI Registration Number: INA000018294. CIN: U74999MH2021PTC373405.
Address: 13/C, Mini Land, Tank Road, Bhandup West, Mumbai, Maharashtra 400078. Email: connect@orim.in
Disclaimer: Investments in the securities market are subject to market risks. Read all the related documents carefully before investing. Registration granted by SEBI, membership of BASL, and certification from NISM in no way guarantee performance of the intermediary or provide any assurance of returns to investors. Past performance should not be considered as a guarantee of future returns. The views expressed in this blog are for educational and informational purposes only and should not be construed as investment advice or a recommendation to buy or sell any security. Full disclosures: https://orim.in/sebi-disclosures/
