The AI Boom Is Entering a New Phase: $126 Billion Is Pouring Into Chips, Data Centres and Power

For the past several years, the artificial-intelligence investment story has largely revolved around one question:

Which company will build the most powerful AI chips and models?

That question helped fuel enormous gains across semiconductor and technology stocks.

But the AI investment cycle is beginning to change.

The next phase may be less about individual AI stocks—and much more about the enormous physical infrastructure required to keep artificial intelligence running.

Across Asia-Pacific, companies connected to high technology have raised approximately $125.8 billion through equity and convertible-bond deals in 2026, more than three times the amount raised a year earlier.

The money is increasingly flowing toward three critical areas:

Chips → Data Centres → Power

That raises an important question for global investors:

Is the next major AI investment opportunity shifting from the companies creating AI to the infrastructure required to power it?

Asia's AI Capital-Raising Boom Is Accelerating

The scale of capital flowing through Asian markets is remarkable.

Asia-Pacific companies have raised approximately $327.1 billion through equity deals in 2026, according to LSEG data.

That is around 53% higher than a year earlier.

High-technology companies alone have raised approximately $125.8 billion, accounting for around 38% of the region's total fundraising.

The number is important because it shows that the AI boom is increasingly becoming a capital-investment cycle, not simply a stock-market theme.

Companies are raising enormous amounts of money to build the infrastructure required for the next stage of artificial intelligence.

That includes:

• AI processors
• memory
• semiconductor manufacturing
• data centres
• networking
• optical communications
• electricity generation
• transmission infrastructure
• cooling systems

The AI boom is increasingly becoming physical.

The AI Investment Chain Is Getting Much Bigger

Every AI prompt looks digital.

The infrastructure behind it is anything but.

When someone asks an AI assistant a question, generates an image or asks an AI agent to perform a task, that request eventually reaches physical computing infrastructure.

The chain looks roughly like this:

AI applications

↓

Cloud computing

↓

Data centres

↓

GPUs and AI accelerators

↓

High-bandwidth memory

↓

Networking

↓

Electricity

↓

Power grids

This explains why the AI investment opportunity is broadening beyond traditional technology companies.

The more AI people use, the more infrastructure is required behind it.

Phase One Was Chips

The first phase of the AI boom was dominated by semiconductors.

Training increasingly powerful AI models required enormous computing resources.

That created extraordinary demand for:

• GPUs
• AI accelerators
• high-bandwidth memory
• advanced semiconductor manufacturing
• semiconductor equipment

South Korea provides one of the clearest examples.

SK hynix recently raised $26.5 billion in a Nasdaq share sale, according to Reuters.

The company has become one of the most important suppliers of high-bandwidth memory used in AI computing.

This connects directly with the surge in South Korean semiconductor exports and provides further evidence that AI demand is moving beyond stock prices into the real industrial economy.

The first AI investment equation was therefore relatively straightforward:

More powerful AI → more chips.

But the equation is now becoming much larger.

Phase Two Is Data Centres

AI chips need somewhere to operate.

That means data centres.

And AI data centres are dramatically different from conventional computing facilities.

Large AI clusters can contain enormous numbers of processors operating simultaneously.

That requires:

• huge buildings
• advanced cooling
• fibre-optic networking
• power-management systems
• backup electricity
• specialised servers

As AI models become larger and AI-agent usage expands, data-centre capacity could become one of the biggest constraints on the industry.

This is why capital is increasingly flowing toward data-centre operators.

Australian AI infrastructure company Firmus and Singapore-based data-centre operator DayOne are among companies that could potentially raise billions of dollars.

The AI investment narrative is therefore expanding from:

Who makes the chips?

to:

Where will all those chips actually run?

Phase Three Could Be Power

This may ultimately become the most important part of the story.

Data centres consume enormous amounts of electricity.

The more AI infrastructure companies build, the more power they require.

That creates a simple relationship:

More AI

↓

More computing

↓

More data centres

↓

More electricity

And electricity infrastructure cannot be expanded overnight.

New demand may require:

• power plants
• renewable generation
• nuclear energy
• natural gas
• transmission lines
• transformers
• battery storage
• grid upgrades

This means the AI investment boom is gradually becoming an energy-infrastructure story.

The next major bottleneck for AI may not necessarily be chips.

It could be electricity.

Why Power Could Become AI's Biggest Constraint

Technology companies can order more processors.

Semiconductor manufacturers can gradually expand production.

But adding large amounts of electricity capacity is much more difficult.

Power projects can take years to approve and build.

Transmission infrastructure can take even longer.

That creates an important mismatch.

AI demand can grow exponentially.

Electricity infrastructure generally grows gradually.

If AI data-centre construction continues accelerating, power availability could become increasingly valuable.

That may create opportunities across industries that investors historically would not have considered part of the technology sector.

AI Is Blurring the Line Between Technology and Utilities

Traditionally, investors viewed technology companies and utilities as completely different sectors.

Technology represented growth.

Utilities represented stable, defensive businesses.

AI is beginning to blur that distinction.

Technology companies increasingly need enormous amounts of reliable electricity.

Utilities need enormous amounts of capital to meet that demand.

The result is a new relationship:

Technology growth creates utility investment.

This could make electricity networks, generation capacity and grid infrastructure increasingly important components of the AI investment ecosystem.

The $126 Billion Figure Shows Investors Are Funding the Buildout

The strongest evidence comes from capital markets themselves.

Investors are providing large amounts of money to companies that can demonstrate genuine exposure to AI infrastructure.

But investors are also becoming more selective.

The market is increasingly distinguishing between:

companies that simply tell an AI story

and

companies generating real earnings from the AI buildout.

That distinction may become extremely important.

As the AI cycle matures, simply adding “AI” to a corporate strategy may no longer be enough.

Investors increasingly want evidence of:

• contracts
• revenue
• capacity utilisation
• cash flow
• infrastructure demand
• sustainable earnings

That could mark an important transition in the AI market.

Asia Is Becoming the Physical Backbone of AI

Much of the AI software innovation receives attention in the United States.

But a large part of the physical AI supply chain sits in Asia.

Consider the ecosystem:

United States

AI models, cloud platforms and chip design.

Taiwan

Advanced semiconductor manufacturing.

South Korea

High-bandwidth memory and memory chips.

Japan

Semiconductor equipment and advanced materials.

China

AI infrastructure, manufacturing and domestic semiconductor expansion.

Singapore

Data centres and regional cloud infrastructure.

India

Rapidly expanding data-centre, digital and cloud demand.

The AI boom is therefore not simply a Silicon Valley story.

It is becoming one of the largest infrastructure investment cycles across Asia.

Readers can track the major Asian, U.S., European and Indian equity markets through the LiveWorldMarket Global Indices & Futures Hub.

China Is Becoming an Important Part of the AI Infrastructure Cycle

China's latest economic data reinforce this trend.

The country's official manufacturing PMI returned to expansion in September, rising to 50.1 from 49.8 in August.

Production improved to 51.7, while new orders reached 50.5.

AI-related industrial activity helped support manufacturing even while property and domestic demand remained weak.

This reinforces the idea that China's economy is becoming increasingly divided between its technology-led industries and weaker traditional sectors.

Read China's Economy Is Splitting in Two: AI Profits Are Surging While Traditional Industries Struggle for our detailed analysis of this divergence.

China Is Building Its Own AI Supply Chain

China is also investing heavily in domestic semiconductor capacity.

Yangtze Memory Technologies is among the companies reportedly considering major fundraising.

The broader objective is clear.

China wants greater control over:

• AI chips
• memory
• semiconductor manufacturing
• cloud computing
• networking
• data centres

U.S. technology restrictions have accelerated this push.

Rather than slowing China's AI ambitions entirely, restrictions have also strengthened incentives to develop domestic alternatives.

That creates a global semiconductor race.

Read Trump-Xi Meeting Could Be the Next Big Catalyst for AI and Semiconductor Stocks for more on the geopolitical side of the AI supply chain.

AI Is Moving From Training to Inference

Another reason infrastructure demand could remain strong is the transition from training AI models to using them at scale.

Training requires enormous computing resources.

But inference happens every time somebody actually uses AI.

Every:

• AI search
• chatbot conversation
• generated image
• coding request
• AI-agent task

requires computing capacity.

As AI adoption expands, inference could become an increasingly important source of demand.

AI agents could accelerate this dramatically.

An agent asked to book a trip may need to perform dozens of computing operations behind the scenes.

That means:

More AI users → more inference → more data-centre utilisation → more electricity demand.

Read AI Agents Are Becoming the New Catalyst for Chip Stocks for our analysis of this next phase of consumer AI demand.

AI Infrastructure Could Become a Multi-Trillion-Dollar Cycle

The scale of planned global AI investment is extraordinary.

Major technology companies and AI developers are committing hundreds of billions of dollars to computing infrastructure.

Anthropic, for example, expects to spend at least $518 billion over a decade building AI infrastructure with several partners, according to an IPO prospectus reviewed by Reuters.

OpenAI's Stargate initiative has also outlined a roughly $500 billion AI infrastructure ambition.

This suggests the current $126 billion Asian fundraising wave could represent only one part of a much larger global investment cycle.

But enormous spending also creates enormous expectations.

Eventually, AI needs to generate enough revenue to justify the infrastructure being built.

The Biggest Risk: What If AI Revenue Cannot Justify the Spending?

This is where the investment story becomes more complicated.

AI infrastructure is extremely expensive.

Data centres require enormous upfront capital.

Chips become obsolete quickly.

Electricity contracts can be long-term.

Debt financing can become expensive when interest rates are high.

If AI revenue grows slower than expected, companies could find themselves with:

large infrastructure commitments + high financing costs + insufficient returns.

This is why investors should not assume that every company connected to AI infrastructure will automatically benefit.

The key question is increasingly:

Who earns enough money from AI to justify the capital being invested?

Credit Markets Are Beginning to Ask That Question

Signs of caution are already emerging.

Some debt linked to major AI infrastructure projects has weakened as investors reassess how much borrowing the sector can absorb.

The concern is not necessarily that major technology companies will default.

Many have exceptionally strong balance sheets.

The issue is return on investment.

If trillions of dollars are invested in AI infrastructure, the industry eventually needs enormous recurring revenue to generate acceptable returns.

That could make AI revenue growth versus AI capital expenditure one of the most important financial ratios of the next several years.

High Bond Yields Make the Calculation Harder

The AI infrastructure boom is occurring in a very different interest-rate environment from previous technology investment cycles.

U.S. Treasury yields are near multi-decade highs.

That means financing large infrastructure projects is expensive.

Companies must therefore generate higher returns to justify investment.

This creates another tension:

AI demand encourages more infrastructure spending.

But:

High interest rates increase the cost of financing that infrastructure.

Read U.S. 10-Year Yield Back Above 5%: Why Strong Growth Is Now a Risk for Stocks for the broader bond-market implications.

AI Stocks Have Already Passed the First Test

Despite rising yields, AI-related equities have remained remarkably resilient.

That suggests investors still believe AI earnings growth can overcome the higher cost of capital.

But the next phase may be more demanding.

Investors may increasingly ask:

Which companies own valuable AI infrastructure?

Which companies generate cash from it?

Which companies depend heavily on debt?

Which companies have reliable electricity access?

Which companies can scale without destroying returns?

That could produce a much wider dispersion between AI winners and losers.

Read AI Stocks Keep Rising While Bond Yields Hit Multi-Decade Highs: Is the Stock Market Becoming Too Dependent on AI? for the equity-market side of this debate.

India Could Become an Important AI Infrastructure Market

India has several characteristics that could make it increasingly important in the AI infrastructure cycle:

• enormous digital population
• rapid cloud adoption
• expanding data-centre market
• growing electricity demand
• large technology-services industry
• increasing semiconductor ambitions

AI inference could become particularly important.

If hundreds of millions of Indian consumers eventually interact regularly with AI assistants, the computing requirement could be enormous.

That creates potential demand for:

local data centres + cloud capacity + fibre networks + electricity infrastructure.

This could gradually make AI infrastructure an important component of India's digital investment story.

AI Could Transform Electricity Demand

The most underappreciated part of the AI boom may ultimately be electricity.

For decades, electricity demand in many developed economies grew slowly.

AI could change that.

Data centres require continuous power.

They also require reliable power.

Interruptions can be extremely costly.

That means technology companies increasingly care about:

• grid reliability
• generation capacity
• long-term power contracts
• renewable energy
• nuclear energy
• battery storage
• natural gas generation

The AI investment cycle could therefore eventually influence energy markets almost as much as technology markets.

The Next AI Trade May Look Very Different

The first AI trade was relatively obvious:

Buy semiconductor companies.

The next phase could be much broader.

Investors may increasingly examine:

Data-centre operators

Companies owning or developing computing facilities.

Power generators

Businesses supplying electricity to AI infrastructure.

Grid equipment

Companies manufacturing transformers, switchgear and transmission equipment.

Cooling infrastructure

AI servers produce enormous amounts of heat.

Optical networking

Large AI clusters need extremely fast data transfer.

Memory

Inference and training require huge amounts of high-speed memory.

The AI ecosystem is expanding outward.

Three Scenarios for the AI Infrastructure Boom

Scenario 1: AI Demand Keeps Accelerating

Consumer and enterprise adoption continues growing rapidly.

AI-agent usage expands.

Inference demand rises.

Data-centre utilisation remains high.

Under this scenario, infrastructure investment could continue for years.

Scenario 2: AI Grows, but Investors Become More Selective

AI adoption remains strong, but markets demand proof of profitability.

Capital remains available for companies with real contracts and cash flow.

Companies relying primarily on AI hype struggle to raise money.

This could create a healthier but more selective investment cycle.

Scenario 3: Infrastructure Gets Ahead of Demand

Companies build too much capacity.

AI revenue fails to grow quickly enough.

Data-centre utilisation disappoints.

Financing costs remain high.

Investment slows sharply.

This is the key downside risk.

What Investors Should Watch Next

The next phase of the AI boom may require monitoring a completely different set of indicators.

Watch:

• AI infrastructure fundraising
• hyperscaler capital expenditure
• data-centre construction
• electricity demand
• power prices
• grid investment
• HBM demand
• semiconductor equipment orders
• optical networking demand
• cloud revenue
• AI inference volumes
• AI revenue versus capital expenditure
• corporate debt linked to AI projects

These indicators could reveal whether AI is developing into a sustainable infrastructure cycle—or an investment boom running ahead of revenue.

The Bigger Picture: AI Is Becoming an Infrastructure Story

Artificial intelligence began as a software story.

Then it became a semiconductor story.

Now it is rapidly becoming an infrastructure story.

The latest $125.8 billion raised by Asia-Pacific high-technology companies shows how quickly that transformation is happening.

The next phase of AI may require enormous investment in:

chips

↓

data centres

↓

networking

↓

electricity

↓

power grids

The most important constraint on AI may therefore eventually be something surprisingly old-fashioned:

physical infrastructure.

And that could dramatically broaden the AI investment landscape.

The question investors should increasingly ask is no longer simply:

“Which AI stock will win?”

It is:

“Who will build—and power—the infrastructure required to run the AI economy?”

That could become one of the defining global investment themes heading into 2027.

Related LiveWorldMarket Analysis

China's Economy Is Splitting in Two: AI Profits Are Surging While Traditional Industries Struggle

AI Agents Are Becoming the New Catalyst for Chip Stocks

AI Stocks Keep Rising While Bond Yields Hit Multi-Decade Highs

U.S. 10-Year Yield Back Above 5%: Why Strong Growth Is Now a Risk for Stocks

Trump-Xi Meeting Could Be the Next Big Catalyst for AI and Semiconductor Stocks

Global Indices & Futures Hub

Disclaimer: This article is for informational and educational purposes only. It does not constitute investment advice or a recommendation to buy or sell any security. AI investment, capital expenditure, interest rates and financial markets can change rapidly.