AI Business · · 5 min read
The AI Data Centre Boom: Why Are Trillions in Related Stock Value Vanishing?
Explain the paradox of massive AI data centre investment alongside plummeting stock values for related companies, and what it means for businesses.

The numbers are staggering. Bain & Company estimates global spending to build new compute capacity for AI could reach an eye-watering US$6.5 trillion by 2030.
This colossal investment aims to nearly triple the world's data centre capacity. It will add almost 150 gigawatts of new power in just five years. Hyperscalers might spend over US$1 trillion by 2028, planning campuses of 5 gigawatts or more. Each could cost up to US$200 billion.
Yet, despite this projected boom, a curious trend has emerged. Companies supplying the 'picks and shovels' for this infrastructure are seeing their market value evaporate.
Global manufacturing, utility, and mining companies have shed US$4.2 trillion in value since June. This wiped out almost half of their gains from the previous year. Caterpillar, a major supplier of generators, saw its market capitalisation slashed by 44%, or US$214 billion, since June 30.
We are witnessing a paradox: huge projected investment in AI infrastructure, yet a dramatic loss of investor confidence in the companies building it.
Investors seem to be questioning the sustainability and profitability of this AI data centre gold rush. The industry is pushing forward with massive scale, but is it sustainable?
But that's not the interesting part. The real story isn't just about the numbers; it's about the underlying forces creating this tension. That raises three questions.
Question 1
What's Driving the Trillion-Dollar Investment in AI Data Centres?
Artificial intelligence's sheer computational demands drive this massive build-out. AI models, especially large language models (LLMs), need immense processing power. Existing data centres simply cannot provide enough.
Bain & Company projects data centre critical capacity will grow by 20 to 30% over the next five years. Just five years ago, a 50-megawatt (MW) facility was large. Now, hyperscalers plan data centre campuses of 5 gigawatts (GW) or more. Each could cost US$150 billion to US$200 billion.
This means more than just servers; it's a complete infrastructure overhaul. Global electricity demand from data centres could more than double. It is forecast to rise from 589 terawatt-hours (TWh) in 2026 to 1,228 TWh in 2030.
For context, US capital investment in data centres now rivals historic spending on railroads, telecom networks, and the electrical grid as a share of GDP. Imagine upgrading every major highway in India simultaneously for new, super-fast electric vehicles. These vehicles would need constant, high-power charging stations everywhere. Is our existing infrastructure ready for AI's horsepower? The answer, often, is no.
And that brings us to the second question.
Question 2
Why Are 'Picks and Shovels' Stocks Losing Trillions Amid This Boom?
The paradox stems from clear demand for AI infrastructure, but major hurdles in delivering it. The economic justification for such scale is also unclear. Investors see these challenges and react. But why are investors reacting this way?
A major issue is the severe strain on supply chains, power infrastructure, and capital. Building AI-ready data centres is incredibly difficult. It needs more than just chips; it requires cooling systems, fibre optics, and crucially, reliable electricity.
Widespread local opposition is another factor. Reuters reported at least 75 data centre projects, worth US$130 billion, were blocked or delayed in Q1 this year. This nearly matches the US$156 billion worth of projects disrupted in all of 2025.
New York paused construction of its largest data centres in July. 14 other US states introduced similar legislation. Countries like Ireland, the Netherlands, and Denmark also restricted new grid connections for data centres.
For context, imagine a chai stall owner wanting to open new outlets. They have capital for equipment. But what if the municipality denies permits, or the electricity board cannot provide reliable power? Their expansion stalls, making the investment risky. This uncertainty hits 'picks and shovels' companies hard.
The market also questions the revenue side. Bain & Company reports the AI industry needs US$6 trillion in annual revenue by 2031. This is to justify the US$5 trillion to US$6.5 trillion in data centre capital deployed by 2030. Investors worry this revenue may not materialise quickly enough. This leads to a de-rating for companies building foundational infrastructure.
And that brings us to our third question.
Question 3
What Does This Contradiction Mean for Businesses Betting on AI?
This market reaction signals a crucial shift. The focus moves from AI's promise to its tangible, justifiable returns. David Crawford, chairman of Bain's global technology practice, suggests the industry needs a 'wave of innovation' to justify this investment. This wave must 'dwarf what mobile and cloud unlocked'.
Bain projects new product development as the biggest revenue contributor, estimated at US$4.2 trillion within five years. Enterprise productivity gains, in areas like software development or customer service, could bring in another US$1 trillion to US$1.4 trillion.
For businesses, merely adopting AI is not enough. Applications must deliver clear, measurable economic value. Is your AI strategy generating new revenue streams? Or is it dramatically improving productivity to justify infrastructure costs?
BlackRock CEO Larry Fink warned that a US data centre construction moratorium could push the country into recession. This highlights deep economic implications. It shows the delicate balance between enabling AI and managing its resource demands.
The takeaway: AI's potential is immense, but its path has real-world challenges. These include securing power, managing supply chains, and overcoming regulatory hurdles. Most importantly, it needs to generate substantial new revenue to be economically viable.
The Bigger Picture
The AI data centre boom testifies to technology's transformative power. But it is also a stark reminder of the physical and economic realities underpinning advancement.
Building AI's digital foundation is as complex as any major physical infrastructure project. It demands huge capital, reliable power, robust supply chains, and community buy-in.
For business leaders, the message is clear: the AI revolution is not just about algorithms. It is about concrete, megawatts, and political will. AI's true success will hinge on its ability to generate significant new economic value. This value must justify its monumental footprint. Can AI truly deliver on its promise without addressing these fundamental constraints?
The market is now calling for AI to deliver on its promise, not just its potential.
Quick Answers
What is driving the AI data centre boom?
The immense computational demands of AI models, especially large language models, are driving a massive global investment. This requires building new data centres and significantly increasing power capacity, with spending estimated up to US$6.5 trillion by 2030.
Why are stocks of AI data centre suppliers losing value?
Despite the boom, companies supplying data centre infrastructure ('picks and shovels') are losing trillions in market value due to challenges like strained supply chains, power infrastructure limitations, and significant local opposition blocking projects. Investors are also questioning if AI can generate the US$6 trillion in annual revenue needed to justify the capital deployment.
What does the AI data centre situation mean for businesses?
Businesses betting on AI should focus on applications that deliver clear, measurable economic value, whether through new revenue streams or significant productivity gains. The market is demanding tangible returns to justify the monumental infrastructure costs and real-world challenges of building out AI capacity.
Sources
- Data Centre Magazine15 Nov · 14:07 IST
- Financial Times15 Nov · 13:38 IST
Facts are credited to the outlets above. The analysis, calculations and Signal Check are by the Finnect AI Insights desk.