The AI build-out is still growing, but the easy valuation story is over
The global AI infrastructure trade has entered a much harder phase. Financial Times analysis published on 15 November says manufacturing, utility and mining companies associated with data-centre construction have lost about $4.2 trillion in market value since June, erasing nearly half the gains they had built over the prior year. That does not mean data-centre demand has disappeared. It means investors are no longer willing to value every supplier as if AI capex can grow indefinitely without financing, regulatory or execution constraints. For India, this is an important test. The country still has a genuine data-centre growth opportunity, but Indian power, cooling, electrical and infrastructure stocks now have to prove that orders, margins and cash returns can justify the valuations attached to the AI theme.
⚡ Quick answer
Is the global AI infrastructure boom breaking?
Not necessarily. The stock-market premium has broken much faster than the physical build-out. US data-centre construction spending, equipment backlogs and power demand remain large, while India still has official projections for rapid capacity expansion. The new phase is about separating durable infrastructure demand from stocks that were priced for perfect execution.
Key takeaways
AI data-centre construction-linked stocks have lost about $4.2 trillion in market value since June, according to Financial Times analysis.
The sell-off reflects worries about debt-funded capex, tighter regulation, expensive valuations and whether AI returns justify infrastructure spending.
Physical demand has not vanished
several global suppliers still report large order backlogs.
In India, 34 of 42 AI-linked companies identified in one September basket fell an average 7% in October.
Indian data-centre capacity reached about 1.57 GW by August 2026 and is projected by the government to approach 8 GW by 2030.
India’s opportunity now depends more on commissioned capacity, customer contracts, power connections, cooling efficiency and return on capital than on AI branding.
Power equipment, cooling, fibre, engineering and grid suppliers may still benefit, but investors need company-specific proof.
Investor note
A structural theme can be right while a stock is still too expensive
The global sell-off is a reminder that a correct long-term thesis does not guarantee a good entry price. A company can operate in a real growth industry and still deliver weak shareholder returns if the valuation already assumes flawless execution for years.
1. What the $4.2 trillion sell-off is telling investors
The most important message is not that AI demand has ended.
It is that markets are repricing the cost of building AI infrastructure.
The Financial Times estimates that companies tied to data-centre construction have lost about $4.2 trillion in market value since June. The group includes industrial, utility and mining businesses whose valuations had expanded sharply as investors treated them as direct beneficiaries of the AI capex boom.
The correction has been severe because the market is now questioning several assumptions at once: whether hyperscalers can keep funding enormous capex, whether local regulation will slow construction, whether electricity systems can support the build-out and whether every supplier deserves a technology-like valuation multiple.
That is a valuation reset, not automatically an infrastructure collapse.
2. Why the physical data-centre boom has not disappeared
The other side of the story is important.
The same FT analysis notes that US data-centre construction spending reached about $96 billion in the year to September. Several large industrial suppliers still report strong order books, including large power-generation and electrical-equipment backlogs.
This creates the central tension for investors.
Share prices can fall while order books remain healthy.
That happens when the market decides that future growth is real but was priced too aggressively. The investment question shifts from “is AI demand growing?” to “how much growth is already in the valuation, and how much free cash flow will actually arrive?”
3. Why India cannot ignore the global reset
India is not isolated from this repricing.
Mint reported that 34 of 42 Indian companies identified as AI-linked in September fell an average 7% in October. Some data-centre and cooling-linked names fell much more sharply.
That matters because Indian investors had increasingly treated power equipment, cooling, telecom infrastructure, data-centre operators and related industrial names as indirect AI winners.
When the global market starts questioning the economics of AI infrastructure, those Indian proxy trades can also lose their valuation premium even if domestic demand is still expanding.
The Indian market therefore faces the same discipline test: evidence now matters more than labels.
4. India’s data-centre growth story is still substantial
The long-term domestic numbers remain large.
Press Information Bureau data says India’s installed data-centre capacity increased from around 375 MW in 2020 to about 1.57 GW by August 2026. The government projects capacity could reach nearly 8 GW by 2030.
PIB also says nearly $70 billion of investment is already underway, with another $90 billion of projects announced.
Those are meaningful numbers, but announced investment is not the same thing as commissioned capacity.
That distinction is becoming more important after the global sell-off. Investors now need to ask how many projects receive power, complete construction, sign customers and generate acceptable returns.
5. Policy still gives India a relative advantage
India is also trying to make the economics of data-centre investment more attractive.
The Union Budget 2026–27 introduced a tax holiday through 2047 for eligible foreign cloud service providers using India-based data-centre infrastructure for global operations, subject to the notified framework.
That long horizon matters because data centres are capital-intensive assets with long payback periods.
The government has also given data centres infrastructure status, while broader semiconductor, AI and electronics programmes are designed to deepen the domestic technology ecosystem.
In a world where capital becomes more selective, policy certainty can become more valuable rather than less.
6. Why India could still benefit from global AI turbulence
The original thesis behind this RegalTicker page remains intact, but it now needs a tougher version.
Global AI turbulence can help India only if India offers a better risk-adjusted build environment.
Reuters Breakingviews has argued that large Indian projects can sometimes be completed faster than comparable developments in the US, where power constraints, local opposition and permitting delays are becoming more important.
If global developers become more disciplined, lower-cost and faster-to-build locations can gain relative appeal.
But a global capex slowdown would still reduce the size of the opportunity. India benefits from reallocation only if spending moves rather than disappears.
7. Power equipment remains important, but the market will demand proof
AI data centres require enormous amounts of electricity.
That creates a real opportunity across transformers, switchgear, substations, grid automation, transmission and backup power.
RegalTicker’s AI data-centre power-grid analysis explains why this theme is broader than one or two “AI stocks.” India’s national grid expansion, renewables build-out and industrial electrification already require large capital spending even before AI demand reaches its full potential.
That broader demand can protect some suppliers from a pure AI slowdown.
But investors should distinguish companies with disclosed data-centre orders from companies that are only being marketed as AI beneficiaries.
8. Cooling could be one of the cleaner picks-and-shovels themes
Dense AI computing creates heat, and heat has to be removed continuously.
That is why direct-to-chip liquid cooling, immersion cooling, closed-loop systems, chillers, pumps and related thermal-management technologies are becoming part of the AI infrastructure stack.
PIB has specifically highlighted advanced cooling and water-management technologies as part of India’s data-centre build-out.
The opportunity is credible, but again, broad sector exposure is not enough. Investors need to check whether a cooling company has actual data-centre customers, what share of revenue comes from the theme and whether margins justify the valuation.
9. Indian IT is exposed differently from infrastructure suppliers
Indian IT services companies sit on a different part of the AI chain.
They are not generally funding the physical data-centre build-out themselves. Their exposure comes through enterprise AI adoption, cloud migration, data engineering, cybersecurity, governance and implementation work.
That makes the global infrastructure sell-off neither clearly bullish nor clearly bearish for Indian IT.
A slower hyperscaler capex cycle can reduce technology enthusiasm. But cheaper or more efficient computing can also accelerate enterprise adoption over time.
This is why RegalTicker’s global AI reset and Indian IT analysis separates the broad India rotation from the harder earnings question facing IT services.
10. The biggest risk for Indian investors is still valuation
The global rout makes one thing clearer: theme quality and stock quality are different questions.
A data-centre supplier can grow revenue rapidly and still disappoint investors if the stock was priced for even faster growth.
A power-equipment company can win orders and still produce weak returns if working capital expands, margins compress or capacity arrives late.
A cooling company can have the right technology and still be a poor investment if competitors enter and pricing power disappears.
The fundamental analysis framework matters more now than it did during the easy phase of the AI rally.
11. What would prove India’s data-centre thesis is working
The next phase should be judged through operating evidence.
Investors should watch:
- commissioned megawatts rather than announced megawatts;
- signed hyperscaler and enterprise contracts;
- power-connection timelines;
- occupancy and utilisation;
- cooling efficiency and water intensity;
- order conversion for electrical and engineering suppliers;
- working-capital discipline;
- free cash flow;
- return on capital;
- valuation relative to realistic earnings growth.
These indicators will tell investors whether India is capturing real AI infrastructure economics rather than simply attracting announcements.
12. What the $4.2 trillion reset changes for the India thesis
The biggest change is psychological.
Investors are no longer assuming that every company touching AI infrastructure deserves a premium valuation.
That is healthy for the long-term theme.
A tougher capital market can eliminate weak projects, expose overleveraged developers and force buyers to demand better economics. It can also redirect investment toward locations where construction, power, taxes and demand create stronger returns.
India may still benefit from that process.
But the bar is now higher.
India needs to prove that it is not only cheaper or faster to build in, but also capable of delivering reliable power, efficient cooling, contracted demand and durable returns.
Investor note
The next winners may look more industrial than technological
If the AI data-centre theme survives the valuation reset, the durable Indian beneficiaries may be companies that quietly convert orders into cash flow—transformers, electrical equipment, cooling, fibre, engineering and infrastructure—rather than companies with the loudest AI branding.
Frequently asked questions
Have AI data-centre stocks really lost $4.2 trillion?
Financial Times analysis says manufacturing, utility and mining companies associated with data-centre construction have lost about $4.2 trillion in market value since June 2026.
Does that mean the AI data-centre boom is over?
No. Construction spending and supplier backlogs remain substantial. The sell-off is better understood as a valuation and financing reset rather than proof that physical demand has disappeared.
How large is India’s data-centre market?
Government data says installed capacity reached about 1.57 GW by August 2026 and could approach 8 GW by 2030.
Which Indian sectors could benefit?
Potential beneficiaries include power equipment, cooling, fibre, networking, engineering, renewable energy, storage and data-centre operators. Actual company exposure must be verified.
Why could the global sell-off help India?
If global developers become more selective, locations with faster construction, clearer policy, competitive costs and strong demand can become relatively more attractive. That helps only if capital is reallocated rather than cancelled.
What is the biggest risk for Indian AI-linked stocks now?
Valuation. The global correction shows that even strong demand cannot protect a stock whose price assumes years of perfect execution.
Verify through official sources
Official references
- Financial Times — 15 November 2026 analysis of the $4.2 trillion market-value loss across data-centre construction-linked stocks and the continuing order-backlog debate.
- Press Information Bureau — Official India data-centre capacity, investment, power-demand and cooling information.
- Press Information Bureau — Budget 2026–27 cloud and AI infrastructure tax-holiday framework through 2047.
- Reuters Breakingviews — India-versus-US data-centre construction and investment context.
- Reuters — Indian proxy AI trade, listed-company exposure and valuation-risk context.
- Mint — Indian AI-linked stock performance and the transmission from the US AI sell-off.
Final takeaway
India’s data-centre opportunity survived the sell-off, but the investment test got harder
The $4.2 trillion global reset does not erase India’s data-centre opportunity. It changes how investors should value it. The next phase will reward execution rather than narrative: real megawatts, real customers, reliable power, efficient cooling, disciplined capital allocation and cash returns. If India delivers those things, global AI turbulence can still strengthen its relative position. If it does not, the same valuation reset now hitting global infrastructure stocks will eventually reach every weak domestic proxy too.
Educational disclaimer: This article is for education and general market information only. It is not investment advice or a recommendation to buy or sell any data-centre, power, cooling, IT or infrastructure stock. AI infrastructure spending, regulation, project timelines and valuations can change quickly. Verify current company filings, exchange disclosures and government information before making financial decisions.




