For the past several years, investors have been rewarded for owning a relatively concentrated group of companies tied to artificial intelligence and semiconductors. NVIDIA, Broadcom, TSMC, AMD, memory manufacturers and other semiconductor companies have benefited from an extraordinary wave of spending on AI infrastructure.
That trend has not disappeared. In fact, the underlying demand for AI infrastructure remains remarkably strong. TSMC recently reported a 45% year-over-year increase in July sales and raised its 2026 revenue-growth outlook to more than 40%, underscoring that demand for advanced AI chips remains robust.
Yet investors are beginning to ask a different question:
If AI continues to grow, where should the next dollar of investment go?
That question helps explain some of the recent rotation within technology and the broader market. Rather than simply moving “out of AI,” investors appear to be looking beyond the first beneficiaries of the AI boom and toward the infrastructure, industries and businesses that can benefit from the next phase of adoption.
In other words, the AI trade may be evolving from a chip story into an economy-wide investment story.
This Isn’t Necessarily an Exit From AI
It is important not to confuse a change in market leadership with the end of the AI investment cycle.
AI-related capital expenditures remain enormous. UBS estimates that global AI capital expenditures could reach approximately $821 billion in 2026 and $986 billion in 2027.
TSMC, one of the most important suppliers in the semiconductor ecosystem, expects 2026 capital expenditures of approximately $60–64 billion, higher than its previous forecast.
Those numbers do not look like the beginning of an AI spending collapse.
Instead, they suggest that the market is entering a more complicated phase.
During the first phase of the AI boom, investors largely focused on the companies selling the “picks and shovels”: GPUs, advanced processors, memory and semiconductor manufacturing.
Now investors are increasingly asking:
- Who supplies the electricity?
- Who builds the data centers?
- Who provides networking and optical connectivity?
- Who manufactures the power-management equipment?
- Who owns the real estate?
- Who provides cooling?
- Who benefits from increased productivity?
- Which industries can use AI to expand margins?
- And, ultimately, who turns AI investment into sustainable cash flow?
That is where the next stage of the opportunity may lie.
From GPUs to the Entire AI Infrastructure Stack
The first major shift is occurring within technology itself.
The AI infrastructure stack is much broader than GPUs.
As AI systems become larger and more complex, bottlenecks are emerging throughout the system. Memory, networking, optical components, power management, advanced packaging and semiconductor manufacturing equipment are all becoming increasingly important.
Franklin Templeton’s semiconductor research highlights this evolution, noting that specialized memory, CPUs, custom accelerators and other components are becoming increasingly important as AI workloads expand. The firm also identifies server power components and memory as potential constraints on further AI capacity growth.
This matters for investors because it changes the question from:
“Which company sells the best AI chip?”
to:
“Which companies enable the entire AI computing system?”
That creates a much broader investment universe.
Networking companies, optical suppliers, memory manufacturers, semiconductor-equipment companies and power-management businesses can all participate in the AI capital-spending cycle.
The result may be less of a narrow “NVIDIA trade” and more of an AI infrastructure ecosystem.
The Next Bottleneck: Electricity
Perhaps the most important second-order consequence of AI is energy demand.
AI data centers require enormous amounts of electricity. As hyperscalers build increasingly large computing campuses, access to reliable power is becoming a constraint on how quickly new capacity can be deployed.
This creates opportunities far outside traditional technology stocks.
Utilities, electrical-equipment manufacturers, grid infrastructure companies, independent power producers, natural-gas infrastructure, nuclear energy and renewable generation can all potentially benefit from the growing electricity requirements of data centers.
In this sense, AI is becoming an energy and infrastructure story.
The market has begun to recognize this connection. UBS has identified power and resources as one of the areas that could benefit from rising data-center demand, alongside industrial companies and other infrastructure providers.
This is one reason we believe investors should look beyond the obvious AI beneficiaries.
The company selling the processor gets paid when the processor is installed.
The utility, power-equipment manufacturer or infrastructure owner may continue benefiting from the electricity required to operate that computing capacity for years.
Data Centers Are Becoming a Real-Estate and Construction Story
The next layer is physical infrastructure.
AI requires buildings.
Those buildings require land, electrical connections, cooling systems, transformers, generators, construction equipment, fiber connectivity and specialized engineering.
This creates opportunities for companies that investors historically would not have considered part of the technology sector.
Industrial companies can benefit from demand for electrical equipment and machinery.
Engineering and construction firms can benefit from new data-center projects.
Real-estate owners with access to power and suitable land can benefit from increasing demand for data-center capacity.
Cooling and HVAC providers can benefit because high-density computing generates enormous amounts of heat.
This is an important characteristic of the current AI cycle: the physical world is becoming an increasingly important part of the digital economy.
From AI Infrastructure to AI Productivity
Perhaps the most significant transition, however, is still ahead.
The first phase of AI investment was primarily about building the infrastructure.
The next phase is about using that infrastructure to improve business productivity.
That distinction could have major implications for investors.
If AI can meaningfully improve productivity, companies across almost every sector could benefit.
Financial-services companies can automate portions of research, customer service and back-office operations.
Healthcare companies can use AI for administrative work, drug discovery and diagnostics.
Manufacturers can use AI for predictive maintenance, design and automation.
Retailers can improve inventory management and personalization.
Professional-services firms can automate portions of research and document production.
Software companies can incorporate AI directly into their products.
The winners of this next phase may therefore not be companies that sell AI technology.
They may be companies that use AI better than their competitors.
Why Software Could Re-Emerge
One of the more interesting implications is the potential shift from hardware toward software.
The semiconductor companies have benefited enormously because customers have had to spend billions of dollars building AI computing infrastructure.
But eventually, investors will demand evidence that this infrastructure generates attractive returns.
That means software and application companies could become increasingly important.
The key question is no longer simply:
“How much AI are companies buying?”
It becomes:
“How much economic value are companies generating from AI?”
That is a much more difficult question—and potentially a much more important one.
Companies capable of using AI to reduce labor costs, increase revenue, improve customer retention or accelerate product development may experience expanding margins.
This is where the AI investment thesis could broaden from a capital-expenditure story into an earnings-growth story.
Financials Could Benefit From a Broader Market Rotation
Another potential destination for capital is financial services.
This does not necessarily mean that investors are abandoning technology for banks.
Rather, financial stocks can provide a different combination of valuation, income and economic sensitivity than the highly valued technology leaders that have dominated the AI narrative.
If economic growth remains healthy while AI investment expands productivity, financial companies could participate in the resulting economic expansion.
Banks, asset managers, exchanges, insurers and alternative-asset managers can also benefit from increased economic activity and continued growth in financial markets.
For diversified portfolios, this can provide an important counterbalance to concentrated exposure to mega-cap technology.
Industrials May Be One of the Biggest Beneficiaries
The industrial sector deserves particular attention.
AI data centers need transformers, switchgear, electrical distribution systems, cooling equipment, generators and construction services.
At the same time, manufacturers are beginning to incorporate AI into factories, logistics networks and supply chains.
That creates two separate AI-related growth opportunities:
AI infrastructure demand and AI-enabled industrial productivity.
Industrial companies therefore represent an interesting bridge between the technology sector and the traditional economy.
The market may increasingly reward companies that can demonstrate tangible earnings growth from these trends rather than simply associating themselves with the AI narrative.
Healthcare: The Longer-Term AI Opportunity
Healthcare may be another area where the economic benefits of AI take time to emerge.
AI has potential applications across drug discovery, medical imaging, clinical documentation, administrative automation and personalized medicine.
But unlike the semiconductor industry, healthcare’s AI opportunity is unlikely to be captured through one dominant hardware supplier.
Instead, value could be distributed across pharmaceutical companies, biotechnology firms, healthcare providers, technology platforms and specialized software companies.
For long-term investors, this makes healthcare particularly interesting as a diversification opportunity within the broader artificial-intelligence theme.
Defense and Cybersecurity
AI is also increasingly intertwined with national security.
Governments are investing in autonomous systems, intelligence analysis, cybersecurity, satellite technology and advanced computing.
At the same time, the more organizations rely on AI and digital infrastructure, the greater the potential importance of cybersecurity.
This creates another layer of beneficiaries beyond traditional semiconductor companies.
Defense and cybersecurity companies can potentially benefit from secular spending trends that are independent of whether a particular AI chipmaker beats or misses quarterly expectations.
That diversification can be valuable in a portfolio.
Why Investors Are Becoming More Selective
There is another reason for the rotation: valuation and concentration risk.
The AI rally has created enormous gains in some companies. That success creates a natural problem for investors.
When a small group of stocks becomes an increasingly large percentage of an index or portfolio, future returns become more dependent on a handful of companies continuing to exceed very high expectations.
UBS has recently highlighted this issue, arguing that the sharp rise in semiconductor and AI-linked stocks has increased the risk associated with single-stock concentration. At the same time, the firm continues to view AI as a major long-term investment theme.
That distinction is critical.
A stock can be an excellent company and still be a poor investment at the wrong valuation.
As expectations rise, the margin for error becomes smaller.
Investors may therefore be rotating not because they believe AI is over, but because they believe the easiest money in the AI trade has already been made.
What Comes Next?
We believe the next phase of the AI investment cycle is likely to be broader and more fragmented.
Instead of one or two dominant beneficiaries, capital could increasingly spread across several layers:
AI hardware → networking → memory → power → data centers → industrial infrastructure → software → productivity → broader economic growth.
This progression is important for investors.
The first stage rewarded companies that enabled AI.
The next stage may reward companies that monetize AI.
And eventually, the most important beneficiaries may be companies that use AI to become structurally more profitable than their competitors.
That could mean greater opportunities in industrials, utilities, energy infrastructure, software, healthcare, financial services and selected consumer companies—alongside continued opportunities within semiconductors.
The Bottom Line for Investors
The recent movement away from the most crowded AI and semiconductor trades should not automatically be interpreted as a vote against artificial intelligence.
In many ways, it may represent the opposite.
AI is becoming too important to remain confined to a handful of semiconductor companies.
The investment opportunity is expanding.
The next phase of the cycle is likely to involve the infrastructure required to power AI, the companies building and connecting data centers, the software companies monetizing AI, and the businesses using AI to improve productivity.
For investors, this creates both opportunity and risk.
The opportunity is that the AI theme could become much larger than the companies that initially led it.
The risk is that investors may simply chase the next group of AI-related stocks without considering valuation, fundamentals or portfolio concentration.
A more disciplined approach is to think about AI as an economic transformation rather than a single sector.
That means looking across the entire value chain—from chips and memory to electricity and infrastructure, and ultimately to the companies capable of converting artificial intelligence into higher productivity and sustainable cash flow.
The AI story may not be ending.
It may simply be moving from the obvious beneficiaries to the broader economy.
This article is for educational and informational purposes only and should not be interpreted as individualized investment advice or a recommendation to buy or sell any security. Market conditions, valuations and investment risks can change rapidly. Investors should consider their objectives, risk tolerance and overall portfolio before making investment decisions.