For the past few years, the artificial intelligence boom has been dominated by one question:
Who is building the best AI?
In 2026, investors are increasingly asking a different question:
Who is actually going to make money from AI?
That shift matters.
The first phase of the AI boom was largely about building models, buying GPUs and constructing data centers. Companies spent enormous amounts of money trying to secure computing capacity before competitors did.
Now the industry is entering a more complicated phase.
The AI infrastructure is being built at extraordinary scale, but investors want to see something more important than impressive models and enormous capital expenditure:
profits.
Big technology companies are expected to spend hundreds of billions of dollars on AI infrastructure in 2026. One recent estimate puts combined capital expenditure by Amazon, Microsoft, Alphabet and Meta at around $760 billion.
At the same time, Nvidia has partnered with major financial institutions including Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR to establish financing platforms intended to mobilize more than $500 billion of third-party capital for AI compute infrastructure.
This raises a fascinating question:
If trillions of dollars are flowing into AI, where will the profits ultimately end up?
The answer may not be the companies that generate the most headlines.
The First AI Gold Rush Was About the "Picks and Shovels"
The classic gold-rush analogy is useful.
During historical gold rushes, many prospectors hoped to become rich by finding gold.
But businesses selling the equipment needed by prospectors could make money regardless of which individual miner succeeded.
AI has followed a similar pattern.
The "picks and shovels" of AI include:
GPUs
Advanced semiconductor manufacturing
Networking equipment
Cooling systems
Nvidia became perhaps the most famous example.
Its GPUs became foundational to modern AI infrastructure, allowing the company to capture enormous economic value from the AI buildout.
But the next phase may be different.
The question is no longer simply:
Who sells the equipment?
It is increasingly:
Who collects recurring revenue from all the AI infrastructure being built?
AI Spending Is Entering a Massive New Phase
The scale of AI investment is difficult to ignore.
Goldman Sachs has noted that estimates for AI capital expenditure vary substantially depending on assumptions about infrastructure construction, replacement cycles and utilization.
Meanwhile, IDC forecasts that the global semiconductor market could reach approximately $1.29 trillion in 2026, with AI infrastructure, memory and hyperscaler demand among the major growth drivers.
This creates a huge economic opportunity.
But there is also a huge problem:
Someone has to pay for all of it.
And eventually, investors will want to know whether the revenue generated by AI is large enough to justify the infrastructure spending.
The Real AI Gold Rush May Be Moving From Models to Infrastructure
The first generation of AI excitement focused heavily on models.
Which model is smarter?
Which company has the best reasoning?
Which AI chatbot has the most users?
Which company can build the largest model?
Those questions still matter.
But the economics of AI increasingly depend on infrastructure.
Every powerful AI model needs:
Computing power
Electricity
Data centers
Networking
Storage
Cooling
Software
Human talent
As AI agents become more capable and AI inference expands, the amount of computing required to serve users could continue increasing.
This means infrastructure providers could remain important even if the AI model landscape changes dramatically.
Nvidia Is Still One of the Biggest Winners
It would be difficult to discuss AI profits without Nvidia.
The company sits near the center of the AI computing ecosystem.
But Nvidia's role is becoming broader.
The company is not simply selling chips.
Its recent financing initiative with major financial institutions is designed to help mobilize capital for AI compute infrastructure.
That potentially gives Nvidia influence across several layers of the AI economy.
It can participate in:
Hardware → Software → Systems → Infrastructure → Financing
This is one reason Nvidia has become such an important company in the AI economy.
However, being a major beneficiary does not mean Nvidia will capture every dollar generated by AI.
As the industry matures, value is likely to spread across the ecosystem.
The Hyperscalers Could Be the Next Major Winners
The largest cloud companies may have one of the most important advantages in the next phase of AI.
Companies such as Microsoft, Amazon and Alphabet already operate enormous cloud platforms.
They have:
Data centers
Enterprise customers
Distribution
Cloud software
Developer ecosystems
Global infrastructure
That gives them multiple ways to monetize AI.
Instead of simply selling AI models, they can sell the infrastructure required to run those models.
This is potentially very powerful.
Microsoft: AI Through the Cloud
Microsoft has a major advantage because AI can be distributed through its existing enterprise ecosystem.
Businesses already use:
Azure
Microsoft 365
GitHub
Dynamics
Security products
AI can be integrated into those products and sold as additional services.
The company's cloud infrastructure also allows it to monetize AI workloads directly.
That creates multiple potential revenue streams.
Amazon: AWS Could Become an AI Toll Road
Amazon's biggest AI economic opportunity may not be its consumer business.
It may be AWS.
Companies developing AI applications need computing infrastructure.
AWS can provide that infrastructure.
Amazon therefore has the potential to earn money whether a customer uses one particular AI model or another.
This is an important characteristic.
A company that provides the infrastructure layer can potentially benefit from the overall expansion of AI rather than betting everything on a single model.
Recent results have reinforced investor attention on cloud growth and AI infrastructure demand.
Google: Search, Cloud and AI Models
Alphabet has a different combination of advantages.
It has:
Search
YouTube
AI research
Advertising
Massive data infrastructure
If AI changes how people search for information, Google has to adapt.
But Google can also monetize AI through its cloud platform and enterprise products.
That creates an unusually broad AI exposure.
Meta Could Monetize AI Without Selling AI Models Directly
Meta is another interesting case.
The company can use AI to improve:
Advertising
Recommendations
Content discovery
Messaging
Creator tools
Its biggest AI opportunity may therefore come from making its existing businesses more valuable.
This highlights an important point:
AI profits do not necessarily require selling AI as a standalone product.
A company can use AI to make its existing business more profitable.
The Next Big Winners May Be "AI Toll Roads"
This could be one of the most important investment ideas of the next phase.
A toll road doesn't care which individual car wins a race.
It makes money when vehicles use the road.
The AI equivalent could be infrastructure companies that earn revenue whenever businesses consume AI computing.
Examples include:
Data-center operators
Networking companies
Memory suppliers
Power providers
Semiconductor equipment manufacturers
These companies may benefit from the growth of the overall AI ecosystem.
Memory Could Be an Unexpected AI Winner
AI requires enormous amounts of memory.
As models become larger and inference workloads increase, high-bandwidth memory becomes increasingly important.
This creates opportunities beyond GPUs.
The semiconductor supply chain includes companies involved in:
DRAM
HBM
NAND
Advanced packaging
The AI boom has already pushed demand across this broader ecosystem. Counterpoint Research reported that leading wafer-fabrication-equipment manufacturers saw strong growth in 2025, driven in part by AI-related investment in leading-edge logic, HBM and advanced packaging.
The lesson is simple:
AI is bigger than GPUs.
Semiconductor Equipment Companies Could Keep Benefiting
Every new generation of AI chips requires sophisticated manufacturing.
That creates demand for semiconductor production equipment.
The companies supplying the machines used to manufacture advanced chips may therefore benefit from continued AI capital expenditure.
This is an example of an indirect AI winner.
Investors don't necessarily need to find the company selling the final AI product.
Sometimes the better opportunity is the company selling the equipment needed to build it.
Data Centers Are Becoming Strategic Assets
AI data centers are different from traditional server facilities.
AI systems require enormous computing density.
They also require:
Advanced cooling
High-capacity electrical connections
Specialized networking
Large amounts of power
Expensive computing hardware
As a result, data-center capacity is becoming increasingly valuable.
The challenge is that building these facilities takes time.
Land can be acquired relatively quickly.
GPUs can sometimes be purchased quickly.
But electricity infrastructure, transmission capacity and large-scale construction can take years.
That creates scarcity.
Electricity May Be the Real "Gold" Behind the AI Gold Rush
One of the most important developments in AI economics is the growing importance of power.
AI chips cannot operate without electricity.
And increasingly powerful AI data centers require enormous quantities of it.
The Financial Times recently reported that major data-center projects could significantly increase U.S. carbon emissions and are already affecting power-generation decisions.
This means AI is increasingly connected to:
Natural gas
Nuclear power
Renewable energy
Transmission
Grid infrastructure
Energy storage
In some locations, power availability may matter more than chip availability.
The AI Boom Is Creating a New Energy Economy
Imagine a company with the perfect AI model.
It has customers.
It has capital.
It has GPUs.
But it cannot get enough electricity.
The AI data center cannot operate at full capacity.
This is becoming a real constraint.
Reports about infrastructure delays increasingly point to power availability and data-center construction as bottlenecks.
That means companies able to provide reliable electricity could become strategically important to the AI industry.
Cooling Could Become Another Hidden Winner
AI processors generate enormous amounts of heat.
Traditional air cooling becomes less practical as computing density rises.
This creates demand for:
Direct-to-chip cooling
Cooling systems
Pumps
Heat exchangers
Thermal management
Again, these businesses may not be considered "AI companies."
But they are essential to the AI economy.
This is one of the central lessons of infrastructure investing:
The companies nobody talks about can sometimes become the companies everyone eventually needs.
Networking Is Another Critical Bottleneck
Thousands of AI processors need to communicate with one another.
That requires extremely fast networking.
As AI clusters become larger, networking performance becomes increasingly important.
This creates opportunities for companies involved in:
Ethernet
Optical networking
Switches
Interconnects
Transceivers
Network processors
AI is therefore creating demand across an increasingly complex hardware ecosystem.
What About AI Software Companies?
This is where things become more difficult.
AI software has enormous potential.
But competition is intense.
The cost of building AI applications can fall rapidly when models become cheaper and more accessible.
That can be good for customers.
But it can make it harder for individual AI startups to maintain high margins.
Imagine building an AI application that costs $1 billion to develop.
Then a competitor releases a model that makes your core technology much cheaper to reproduce.
Your competitive advantage could disappear.
This is one reason investors need to distinguish between:
AI usage
and
AI Adoption Does Not Automatically Equal AI Profits
This is perhaps the most important point in the entire AI investment story.
Companies can use AI extensively without generating enormous profits from it.
Goldman Sachs recently noted that AI adoption is accelerating, but the earnings impact across the broader S&P 500 remains relatively narrow so far. Only a small share of companies have quantified specific AI productivity benefits or linked AI directly to earnings gains.
That doesn't mean AI is failing.
It means the monetization process is still developing.
The $800 Billion Question
Estimates of global AI spending vary depending on what is included.
Axios recently reported estimates approaching $800 billion in AI development spending in 2026, while also noting that the profitability of the massive investments by major technology companies remains difficult to measure.
This creates a huge economic experiment.
Companies are spending enormous sums today based on the expectation that AI will generate much larger revenues tomorrow.
The next phase of the AI boom will test that assumption.
AI Agents Could Change the Economics
One potentially important catalyst is the rise of AI agents.
Traditional AI often assists humans.
Agents can potentially perform tasks independently.
For example, an AI agent could:
Research companies
Write software
Process invoices
Manage customer service
Analyze financial documents
Coordinate workflows
Monitor systems
Purchase goods
Schedule meetings
If businesses begin deploying thousands of agents, AI usage could increase dramatically.
That could increase demand for inference computing.
And inference could become one of the biggest long-term AI infrastructure markets.
The Inference Economy Could Be Bigger Than Training
Training large AI models gets most of the attention.
But once a model exists, it can serve millions of users.
Every request requires computing.
As AI becomes integrated into everyday software, inference could become a continuous revenue stream for infrastructure providers.
This creates a potentially powerful economic model:
AI application → user interaction → inference → compute consumption → infrastructure revenue
That is exactly the kind of recurring usage that investors tend to value.
But AI Infrastructure Is Not Risk-Free
The biggest danger is overbuilding.
Companies are constructing enormous amounts of infrastructure based on expectations of future AI demand.
What happens if demand grows more slowly?
Data centers could become underutilized.
GPU rental prices could fall.
Power infrastructure could be stranded.
Debt burdens could rise.
Investors could discover that the expected returns were too optimistic.
Goldman Sachs has emphasized that estimates of the AI buildout depend heavily on assumptions about utilization, infrastructure design and replacement cycles.
There Is Also a Financing Risk
The AI infrastructure boom increasingly relies on external financing.
That makes interest rates and credit conditions important.
If capital becomes expensive, some projects may no longer make economic sense.
Nvidia's new financing initiative demonstrates how important private capital is becoming to the AI buildout. The company says its partnerships are intended to mobilize more than $500 billion of third-party capital.
This could accelerate the AI buildout.
But it also means investors should pay attention to:
Debt
Lease obligations
Contractual commitments
Financing costs
Cash flow
Customer concentration
Investors Are Starting to Look Beyond Nvidia
The latest market conversation reflects this shift.
Reuters reported on August 17 that investors are increasingly looking for the next generation of AI winners as concerns about AI capital expenditure ease, with attention moving toward semiconductors and hyperscale cloud providers.
This is significant.
The market is gradually moving from:
"Who is winning the AI model race?"
to:
"Who has the strongest economics?"
The Companies That Could Make the Most Money
Rather than trying to predict one winner, it may be more useful to divide the AI economy into several groups.
1. AI Chip Leaders
These companies sell the fundamental computing hardware.
Potential advantages:
High demand
Pricing power
Technological leadership
Large customer base
But competition and technological change remain major risks.
2. Hyperscalers
Cloud companies can monetize AI through infrastructure, software and enterprise services.
Their advantage is scale.
They already have millions of customers.
3. Semiconductor Equipment
These companies benefit when the entire semiconductor industry increases investment.
They don't need to predict which AI model wins.
They sell equipment to the companies manufacturing the chips.
4. Memory Companies
AI workloads require enormous amounts of high-performance memory.
Memory suppliers can therefore benefit from increased AI computing demand.
However, semiconductor memory is cyclical, so supply and pricing remain important risks.
5. Data-Center Operators
Data-center companies can potentially generate recurring revenue from AI infrastructure.
The challenge is capital intensity.
Building facilities is expensive.
6. Energy Companies
AI data centers require power.
Utilities and energy infrastructure providers could benefit if demand continues increasing.
7. Networking Companies
Large AI clusters require enormous amounts of data movement.
Networking could become increasingly valuable as AI systems scale.
8. Cooling and Thermal Management
Higher computing density requires better cooling.
This is another infrastructure layer with potentially strong demand.
9. AI Application Companies
This is potentially the biggest prize—and the hardest to identify.
A company that develops an AI application with a genuine competitive advantage could eventually capture enormous profits.
But the failure rate may be high.
The Biggest Winner Could Be the Company That Owns the Customer
There is another principle investors should consider.
Technology changes quickly.
Customers often don't.
A company with strong relationships with millions of businesses may have an enormous advantage.
This is why cloud platforms, enterprise software companies and large technology ecosystems could remain powerful.
They can introduce AI products to customers they already serve.
The AI Gold Rush May Become a Margin Battle
The next phase may not be about who has the biggest model.
It could be about who has the best economics.
Companies will increasingly compete on:
Cost per AI query
Energy efficiency
Hardware utilization
Customer retention
Revenue per user
Gross margin
Free cash flow
Return on invested capital
This is where the AI boom begins to look less like a technology experiment and more like a normal industry.
What Investors Should Watch
If you're trying to identify the future winners of the AI economy, don't focus only on headlines.
Watch the numbers.
Revenue Growth
Is AI actually generating new revenue?
Gross Margins
Can the company monetize AI profitably?
Free Cash Flow
Is AI producing cash or consuming it?
Capital Expenditure
How much money is being spent to support AI?
Data-Center Utilization
Are expensive facilities being used?
Customer Concentration
Does one customer account for a huge percentage of revenue?
Recurring Revenue
Does the company get paid repeatedly for AI usage?
Return on Invested Capital
Are AI investments generating attractive returns?
The "Picks and Shovels" Strategy Still Works—But It Is Changing
The first AI gold rush rewarded companies supplying the infrastructure.
The second phase could reward companies that monetize the infrastructure.
That could mean:
Cloud providers charging for compute.
Data centers charging for capacity.
Utilities selling electricity.
Networking companies selling connectivity.
Memory companies selling HBM.
Software companies charging for AI agents.
Enterprises using AI to reduce costs and increase productivity.
The winners could therefore be spread across the economy.
What Could Kill the AI Gold Rush?
Several things could slow the boom.
AI Models Become Much More Efficient
If AI becomes dramatically cheaper to run, fewer GPUs may be needed per unit of output.
That could reduce hardware demand even while AI adoption grows.
AI Revenue Disappoints
If customers don't pay enough for AI services, infrastructure returns could suffer.
Interest Rates Rise
Higher financing costs could make massive data-center projects less attractive.
Energy Constraints
Insufficient power could delay infrastructure projects.
Regulation
Governments could impose restrictions on AI development, data centers or energy consumption.
Competition
Fierce competition could push AI prices down faster than costs fall.
The AI Bubble Question
Is AI a bubble?
The honest answer is:
Parts of the AI market could be overheated without AI itself being a bubble.
The technology is real.
The demand is real.
The infrastructure buildout is real.
But that does not mean every company benefiting from the AI narrative will generate attractive long-term returns.
Some businesses will succeed.
Others will disappear.
Some infrastructure projects will become highly profitable.
Others may become stranded assets.
That is normal in major technological transitions.
The Next Phase Is About Economic Moats
The most valuable AI companies may ultimately be those with durable competitive advantages.
A moat could come from:
Proprietary technology
Distribution
Customer relationships
Data
Manufacturing expertise
Energy access
Capital
Brand
Switching costs
Having a powerful AI model is useful.
Having a business that competitors cannot easily replicate may be even more valuable.
Who Will Actually Make Money?
If the AI boom continues, the likely winners may fall into three broad categories.
The Infrastructure Owners
They own the physical systems that AI needs.
The Infrastructure Suppliers
They sell the chips, memory, networking, cooling and equipment.
The AI Monetizers
They use AI to generate recurring revenue or significantly improve existing businesses.
The biggest winners may be companies that combine all three characteristics.
The Most Important Shift in AI Investing
The AI story is moving from technology scarcity to economic scarcity.
At first, the scarce resource was advanced AI models.
Then it was GPUs.
Then data-center capacity.
Now the scarce resources increasingly include:
Electricity
Capital
Customers
Profitable AI applications
Skilled workers
High-quality data
Infrastructure
The companies controlling these resources could capture a disproportionate share of future AI profits.
Final Thoughts
The AI gold rush is not ending.
It is evolving.
The first phase rewarded companies capable of building powerful models and supplying the computing infrastructure needed to train them.
The next phase will be much more demanding.
Investors will want proof that AI can generate sustainable revenue, strong margins and real free cash flow.
That could create opportunities across the entire AI ecosystem.
Nvidia and other semiconductor companies may continue to benefit from demand for computing hardware.
Cloud providers could become AI toll roads.
Data-center operators could benefit from recurring demand for compute.
Energy and networking companies could become critical infrastructure suppliers.
And AI software companies could capture enormous profits if they discover applications that customers are willing to pay for repeatedly.
But not everyone will win.
The AI industry could experience the same process seen in previous technological revolutions: enormous investment, explosive growth, intense competition, consolidation and eventually a smaller group of companies capturing most of the profits.
That is why the most important question for investors is no longer:
"Which company has the coolest AI?"
It is:
"Which companies can turn AI demand into durable cash flow?"
The companies that answer that question successfully may become the true winners of the AI gold rush.
And the next decade could reveal that the biggest AI fortunes were not necessarily made by the companies with the most impressive technology—but by the companies that figured out how to monetize it at scale.
Frequently Asked Questions (FAQ)
1. What is the AI gold rush?
The AI gold rush refers to the enormous wave of investment, innovation and business activity surrounding artificial intelligence. It includes spending on AI models, chips, data centers, cloud computing, software, electricity and related infrastructure.
2. Who is making the most money from AI?
AI chip and infrastructure companies have captured significant value so far, but the long-term winners are likely to include semiconductor companies, hyperscale cloud providers, data-center operators, energy suppliers, networking companies and AI application developers.
3. Is Nvidia still the biggest AI winner?
Nvidia remains one of the most important beneficiaries of AI infrastructure demand. Its position in AI computing gives it significant exposure to continued infrastructure spending. However, the industry is becoming broader, and other companies may capture increasing portions of the AI profit pool.
4. What are AI "picks and shovels"?
AI picks-and-shovels are the products and services that companies need to build and operate AI systems. They include GPUs, memory, networking equipment, semiconductor manufacturing equipment, data centers, cooling systems and electricity.
5. Could cloud companies become the biggest AI winners?
Yes. Cloud providers have a major advantage because they can monetize AI through computing, software and enterprise services. Their existing customer relationships and infrastructure could make them important long-term beneficiaries.
6. Why is electricity so important to AI?
Large AI data centers require significant amounts of electricity. As computing demand grows, access to reliable and affordable power could become one of the biggest constraints on AI expansion.
7. Will AI create more profits than it consumes in investment?
That remains one of the biggest unanswered questions. Technology companies are spending enormous amounts on AI infrastructure, but the full financial return from those investments is still developing.
8. Could AI infrastructure become an investment asset class?
Potentially. Nvidia's partnerships with major financial institutions to mobilize more than $500 billion of third-party capital demonstrate how AI compute is increasingly being treated as infrastructure that can attract institutional financing.
9. What is the biggest risk in the AI investment boom?
One of the biggest risks is overbuilding. If companies construct infrastructure faster than AI demand grows, some facilities and equipment could generate lower-than-expected returns.
10. Are AI software companies good investments?
Some could become extremely valuable, but software is also one of the most competitive areas of AI. Investors should look for companies with recurring revenue, strong customer relationships, defensible technology and improving margins rather than assuming every AI startup will succeed.
11. Could AI make existing companies more profitable?
Yes. A company does not necessarily need to sell AI directly to benefit from it. AI can reduce costs, improve productivity, automate customer service, improve advertising and increase the value of existing products.
12. What should investors watch when evaluating an AI company?
Important indicators include revenue growth, gross margins, free cash flow, capital expenditure, customer concentration, recurring revenue, infrastructure utilization and return on invested capital.
13. Is the AI boom a bubble?
Some AI stocks or infrastructure projects may be overvalued, but that does not necessarily mean AI itself is a bubble. The technology has genuine applications and growing adoption. The key issue is whether valuations and infrastructure spending ultimately match the economic value AI creates.
14. What is the biggest AI investment opportunity for the next few years?
There is no guaranteed winner. The strongest opportunities may emerge among companies controlling critical infrastructure, providing recurring AI services, or using AI to create substantial productivity and revenue gains.
15. What is the biggest lesson for AI investors?
Don't confuse AI adoption with AI profitability. A company can use AI extensively while still generating poor returns. The long-term winners will likely be businesses that can turn AI demand into sustainable revenue, strong margins and free cash flow.
Disclaimer: This article is for educational and informational purposes only and is not financial, investment, legal or tax advice. AI and technology stocks can be highly volatile. Investors should conduct independent research and consider their own financial circumstances before making investment decisions.
Post a Comment