Hero Background Image
In Depth

The AI Disruption: A Long-Term Opportunity

Spring 2026 | Download PDF

 

Uncertainty about AI-driven disruption has been a primary driver of market performance in early 2026. Investors, lacking visibility into the long-term impact of accelerating AI capabilities, sold off whole industries on the fear that AI tools would replace their business models. In contrast, areas of the market with exposure to positive AI trends—such as semiconductors and infrastructure—appreciated significantly. We believe this dispersion in the markets between “AI winners” and “AI losers” (Exhibit 1) was driven by misunderstanding and fear. More important, in our view, this short-term perspective overlooks long-term investment opportunities in both groups.

 

 

*MSCI Barra and Baron Capital analysis. Chart shows rolling 3-month MSCI Barra U.S. industry factor returns relative to each industry factor’s own history since 1996. Colored bars represent historical percentile bands of rolling 3-month returns; purple diamonds show the 3-month return as of March 31, 2026. Percentile labels rank historical 3-month returns from strongest to weakest, so lower percentiles indicate stronger relative performance and higher percentiles indicate weaker relative performance. Barra industry factor returns are model-derived and are not investable index returns.

 

In AI winners, companies across sectors—including Financials, Real Estate, Health Care, and Industrials—are benefiting from a generational change in how individuals interact with technology. These companies are providing crucial elements in the AI buildout or have successfully adopted AI capabilities into their businesses. They have been rewarded by the markets, but we believe that their growth potential is much larger—and will last much longer—than investors realize.

On the other hand, we believe many companies that have borne the brunt of AI displacement fears—the AI losers—have resilient strengths, including visionary management, competitive moats, and sustainable growth opportunities. These strengths may help them not only survive AI disruption but profit from AI advancements. We also note that, after recent market volatility, many are priced at what we view as attractive valuations despite solid fundamentals.

As bottom-up investors, we focus on secular growth trends that disrupt industries and create sustained, profitable growth opportunities—not geopolitical events (e.g., Iran-U.S. conflict) or sentiment swings. In our survey of industries recently impacted by AI, we describe why AI winners may offer more growth potential than widely perceived, while select AI losers may be underestimated and have competitive strengths that could emerge and deliver long-term outperformance. For skilled, patient investors, we believe the current environment offers extraordinary opportunities as the market eventually separates genuine AI losers from misperceived ones.
 

Technology

While semiconductors and infrastructure are direct beneficiaries of the AI revolution, the software industry has been experiencing its most dramatic valuation re-rating since the dot-com era. The median public software company fell about 25% in the first quarter of 2026, and multiples compressed to 10-year lows (Exhibit 2).1 This decline has been largely indiscriminate—driven not by deteriorating fundamentals but by fear that AI will disrupt software incumbents. AI disruption is a viable concern, but not all companies face the same degree of risk. Some will be disrupted, some will prove resilient, and some will emerge as significant beneficiaries.

 

 

We recently attended company and industry conferences, participated in expert panels and technical sessions, and spoke with IT teams at Fortune 500 companies and global systems integrators to better understand how AI adoption is expanding across workflows and budgets. We also met with venture capital investors and domain experts. This level of corporate access informs not just individual stock decisions but the broader analytical framework we apply across our strategies.

One of our conclusions is that companies in technology infrastructure (AI winners) have more growth potential than the market recognizes. Multiple AI scaling laws remain intact—across pre-training, post-training, inference-time compute, context windows, and memory—and frontier AI labs continue to demonstrate strong improvements in model capability with each successive generation. This demand continues to fuel the infrastructure buildout, which is of historic proportions. Leading technology companies have collectively guided to about $650 billion to $700 billion of capital expenditures in 2026, triple the level from two years ago. The vast majority of these investments are directed at AI data centers, semiconductors, and networking.2

This buildout has benefited many companies directly and indirectly, such as those that produce custom silicon. Broadcom confirmed an expanded agreement with Alphabet for next-generation Tensor Processing Units (TPUs) and a deal to supply approximately 3.5 gigawatts of compute capacity to Anthropic. Amazon’s custom chip business—spanning Trainium, Graviton, and Nitro—has reached a $20 billion annual revenue run rate, representing triple-digit growth rates.

One of the most important questions for investors in 2025 was whether the hundreds of billions of dollars devoted to AI infrastructure would yield a worthwhile return. In 2026, evidence appeared that a return-on-investment (ROI) inflection is underway. Anthropic’s Opus 4.5 release dramatically accelerated AI inference and triggered a surge in corporate AI adoption and model use. Anthropic’s annualized revenue rate jumped from about $9 billion at year-end 2025 to over $30 billion as of early April 2026, with more than 1,000 enterprise customers spending over $1 million annually—a figure that doubled in under two months. As of the first quarter 2026, OpenAI was generating $2 billion in monthly revenue, ChatGPT’s weekly active user base had exceeded 900 million, and its coding agent was serving over two million users, up five times in three months. Both companies are investing aggressively in compute, with expected capacity buildouts of 5 to 6 gigawatts each by year end.

These results surprised many investors who, in our view, are still underestimating the true potential of AI growth. On the other hand, we believe investors are too pessimistic about software.

Software market performance in the first quarter was largely driven by multiple compression rather than fundamentals, reflecting fear-based narratives about potential AI disruption. This overlooks, in our view, that many software companies have differentiated proprietary data that is inaccessible to outside AI applications. The companies can leverage this data to compound their competitive advantages, which can include more effective AI-driven analysis because they have highly accurate, verifiable (i.e., deterministic) data. In addition, the software beneficiaries will be market-share leaders growing faster than competitors, with pricing models aligned to usage or outcomes rather than headcount, and led by talented founders with the authority and willingness to self-disrupt.
 

Emerging Markets

Many emerging market (EM) companies play a crucial role in the AI value chain. Memory and leading-edge logic, the two most important enabling technologies, are disproportionately supplied by EM companies, which is driving the asset class’s share of AI spend higher.

Taiwan and Korea are well-known centers of semiconductor manufacturing, and smaller EM companies produce critical components for the ongoing development of advanced computing, advanced semiconductor design, manufacturing, packaging, bonding, and testing. Over the past year, EM companies have begun to outperform the U.S. hyperscalers and the Magnificent Seven complex, given their increasing share of AI spend. This performance, in turn, is attracting global investors, who are seeing emerging markets as a more diversified way to participate in the AI theme. Given our view that AI’s potential is still underestimated by investors, however, we believe select EM companies in the AI value chain are opportunities.

China, largely independent of the U.S., is developing its own AI ecosystem. This system is a couple of years behind the U.S. because the Chinese do not have access to leading-edge semiconductors and technology. However, due to these constraints, the Chinese have been forced to develop solutions that have delivered strong results for significantly less capital than deployed in the U.S. Agentic AI capabilities are starting to emerge in the region as well and are being employed across the country, offering opportunities to investors.

China has a number of other significant advantages as well. The Chinese economy offers a large addressable market with exposure to more than a billion consumers, and China has aggressively followed a roadmap for energy development and production. China’s engineering and technology skillset is deep, and its technology hardware is catching up to the U.S. We believe this AI opportunity will drive productivity growth within China, creating value and opportunity for investors who are on the ground level looking for bottom-up opportunities.

The U.S. and China are in a competition to see which AI ecosystem is adopted by the rest of the world, since regions like Europe, Canada, and Latin America do not have an organic source of AI potential and development. These regions have been U.S. allies but, as U.S. foreign policy continues to evolve, they are beginning to explore a deeper relationship with China. The global competition for share remains early, but we believe some applications that are not nationally security dependent could gain popularity in the rest of the world, benefiting select Chinese companies.

Chinese internet service companies, for example, are among the largest gaming companies in the world, often combining their gaming offerings with social platforms and thus benefiting from virtuous network effects. They can deploy AI tools across their gaming and social media services to improve both engagement and monetization.

 

Financials

In the Financials sector, potential disruption from AI weighed on many subsectors, including financial data providers, insurance brokers, and wealth managers. Alternative asset managers faced higher redemption requests from retail investors due to concerns in their private credit business about lending standards and exposure to software companies facing AI risks.3

Financial software and information services companies sold off after Anthropic’s release of specialized Claude Cowork plug-ins, which enable AI to function as domain-specific analysts across legal, finance/accounting, sales/marketing, and customer support. This development caused investors to worry that AI agents could directly replace expensive human-led, subscription-based business workflows. Accounting and tax preparation software providers, along with data analytics companies, fell dramatically as a result.

Within these pockets of extreme underperformance, several companies with strong competitive characteristics were punished as well. As their advantages are combined with AI tools to outperform competitors, increase market share, and drive growth, we believe these companies will ultimately be recognized by the markets. As technology adoption accelerates, the gap between outperformers and underperformers within Financials is already widening. Firms that are embracing and using transformative technologies are expected to grow three times faster than traditional banks from 2023 to 2028.4 Notably, since the Global Financial Crisis, the number of capital-light, service-oriented companies—rather than traditional branch-based institutions—among the top 25 financial businesses has more than doubled, indicating that the industry shift is already well underway.

Looking forward, we believe financial companies are poised to be among the primary beneficiaries from the explosive growth in AI capabilities. Many are well-capitalized, have ample access to cash flow and expertise, and have massive datasets that are well suited for AI-driven applications. Financial services firms continue to invest heavily in innovation. The sector spent $425 billion on technology as of year-end 2024, more than any other industry, and that figure is projected to rise each year (Exhibit 3). This is creating strong tailwinds for financial businesses positioned to capitalize on tech-driven change—an opportunity that we believe has been largely underestimated by the markets.

 


 

Real Estate

In the first quarter of 2026, tangible assets emerged as AI winners. Real Estate, for example, tends to offer greater near-term certainty and lower risk of AI-driven disruption compared to segments of the digital economy. Investors purchased asset-intensive businesses—HALO (heavy assets, low obsolescence)—including select REITs, homebuilders, and residential building products companies. Some REITs were relatively insulated from AI-related disruption, benefiting from well-covered dividends, contracted cash flows, annual rent escalators, and other structural advantages. At the same time, the asset class was supported by several secular themes: higher demand for data centers, an aging population (boosting Health Care real estate), housing affordability pressures (increasing rental demand), suburbanization (benefiting retail), and the rise of remote work (fueling storage needs).

We note, however, that concerns about AI-driven disruption did pressure some parts of the sector. Providers of information and marketing services to the commercial and residential real estate industries fell, as did companies that provide brokerage services and advisory and consulting. Many of these companies, however, remain high-quality opportunities—the declines were not reflected in the fundamentals, which remained solid.

As AI tools improve and usage rises, AI-driven change will continue to impact Real Estate. While visibility remains limited, we see potential beneficiaries in:

Data Centers: AI workloads are driving explosive demand for data center real estate around the world. Given the many factors needed to make a data center site suitable—including power sources, water, favorable regulations, and government support—this real estate can be extremely valuable.

Industrial: AI’s physical manifestation—robotics, chip manufacturing, and supply chain automation—will require more industrial facilities (hubs that can accommodate robotics and on-site servers).

However, real estate categories that may face AI-related headwinds include:

Office: The long-term prospects for office real estate are mixed, in part since AI-driven automation and hybrid work models may lead to reduced traditional office demand.

Residential: AI infrastructure may reshape regional housing markets—job loss may weaken some residential markets, while other markets attracting large AI campuses may see housing demand spike.

We have been selective with the addition of office and residential companies due to AI-related job loss concerns and our expectations for superior long-term growth prospects in other segments of Real Estate. In certain instances, however, individual office and residential real estate companies are considered for purchase if we view them as attractively valued and offer prospects for solid population and job growth, which should lead to improvements in occupancy, rents, and cash flow growth.

Lastly, though Baron Capital does not predict the path of interest rates, we note that AI could lead to greater productivity and lower numbers of workers for companies. This runs counter to the Federal Reserve’s mandate for full employment, and AI-driven job loss may result in a secular trend toward lower interest rates.
 

Health Care

Health Care is one of the largest sectors in the U.S. economy, accounting for almost a fifth of GDP in 2024. Health Care encompasses a diverse array of sub-industries, and it is a dynamic sector undergoing changes driven by legislation, regulation, and advances in science and technology. Health Care companies can also be responsible for a range of tasks, from administration, to diagnoses, to drug development, to patient experience, to billing. Given the scale and expense of these operations, investors have long recognized the potential beneficial applications of AI.

In 2026, however, the stocks of certain Health Care companies were punished because of investor concern about potential AI-driven disruption. These included companies that offer AI-driven diagnostic solutions—such as AI algorithms to enhance cancer screenings—as well as companies that sell products and services used in scientific research and companies that offer AI-guided, robotic support for surgical procedures. As in other sectors, we believe these AI losers have been diagnosed prematurely.

We believe AI is not a threat but a tool that can improve business operations and generate breakthrough discoveries. Health Care companies hold enormous amounts of proprietary data that can provide fertile areas for AI-driven analysis. AI can help accelerate drug development and, ultimately, drive overall R&D investment higher. For firms that specialize in diagnostic imaging, AI can reduce the time required to conduct and review scans, as well as enhance accuracy of diagnoses. This could increase capacity while enabling earlier and more effective cancer detection.

Technology companies have recognized the potential of AI in Health Care. NVIDIA and Lilly, for example, have announced a partnership to help discover new medicines and improve clinical trials, and Lilly executives have cited AI for identifying inefficiencies in the company’s drug manufacturing processes.5 In mid-April, OpenAI launched model GPT-Rosalind, which specializes in biology knowledge and scientific research. The model is designed to “help research accelerate the early stages of discovery” and, among other objectives, shorten the drug development timeline.6

The federal government is also supporting AI use in Health Care. In April, the FDA announced that it will launch an AI pilot to accelerate clinical trials of medicines. The agency expects AI can help companies collect and submit study data in real time, supplanting a normally laborious and often manual process. This can also make studies more accessible to companies with fewer resources and outside traditional study centers. The potential benefits are significant. More than 90% of candidates fail drug clinical trials, and drug discovery can cost hundreds of millions of dollars and require several years.7

These examples show why Health Care offers some of the largest—and least understood—opportunities for AI-driven change.
 

Industrials

While industrial firms have been relatively unexposed to AI disruption, they have benefited from the growth of AI infrastructure. Non-residential construction spending is currently being driven by favorable trends in data center building, as well as the reshoring of manufacturing and other industrial categories, which we believe can continue for several more years. Federal and state-level funding for new and existing projects continues to support steady growth in infrastructure-related spending. AI is driving electrical load growth for the first time in 20 years, which is also leading to demand for grid modernization, electrification, and broader energy independence globally.

We believe these secular tailwinds are more powerful than generally recognized in the markets, and well-run and competitively advantaged companies that provide skills, services, and products into infrastructure are opportunities. The manufacturers of specialized equipment used in data centers, the power grid, and energy-intensive industrial applications will also benefit. (About 1,500 data centers are under construction in the U.S. and, as noted, hundreds of billions have been allocated by U.S. technology companies for further data center investment.8) Suppliers of basic raw materials used in construction and infrastructure development (crushed stone, sand, and gravel for concrete and asphalt) are seeing strong demand. Finally, the demand for energy to power data centers has led to a renewed interest in previously moribund industries, such as nuclear power.

Industrials also offer exposure to robotics, which can fulfill a large number of relatively simple, repetitive tasks (e.g., manufacturing) and are a focus of leading technology companies. Industrials are also using AI to improve their operations (e.g., in food processing). Finally, industrials provide niche, valuable labor skills needed for infrastructure capex, from electricians to nuclear engineering.

Industrials have typically been considered relatively safe from AI disruption because the work is driven by materials and labor, but also because of the nature of risk-tolerance in the space. Many of the companies have deep, long-lasting relationships with customers based on reliability and trust. It is notable that industrial production has real-world consequences that set high standards for work quality and materials. Companies have strong incentives to avoid experimentation or risks because a failure—from shoddy construction, to accidents, to code violations—can have long-lasting, even catastrophic consequences.
 

Competitive Advantages in an Era of AI Disruption

We believe AI accelerates disruptive change, reduces barriers to entry, and makes it easier for companies to copy features and functionality from competitors. We also believe, however, that structural, competitive moats remain durable and many are widening. Simpler product- or workflow-based moats, in contrast, are becoming less defensible. As we have discussed, we believe investors have largely underestimated the competitive resilience of many companies that have been caught in fear-driven selling. We have identified several competitive advantages that will help companies not just survive AI disruption, but flourish because of it. These advantages include:

  1. Unique, deterministic, and proprietary data. Data—proprietary, clean, secure, well-organized, and accessible—has been called the “new gold for companies.”9 In the AI era, we believe companies that can capture or create unique, proprietary data have an enormous competitive edge. These companies can generate revenue by selling their data for independent ratings, research, risk management, and analytics. Management can use increasingly sophisticated AI tools to analyze the data and develop new products, competitive insights, and strategic initiatives. Software companies with access to private customer data may generate better solutions or answers to queries. (It may also be unsafe, illegal, or out of policy for a company to use an external model, or to allow that external model to have access to its proprietary information.) The gap between the data “haves” and “have nots” is widening. In many companies, data is continuously generated as a byproduct of serving customers and cannot be replicated by a foundation model trained on public sources.
  2. Regulation. Heavily regulated industries—such as health care and finance—make up a sizeable portion of the U.S. economy. Effective regulations are necessary for economic growth; they establish clear rules, build confidence, and encourage development. While regulatory environments will evolve, these companies will always operate within a regulatory framework. They will need deep expertise, solid internal processes and controls, and regular contact with regulators and the regulatory process. They carry high barriers to entry and high switching costs. Health Care companies can be subject to extreme penalties for misuse or loss of patient information, and clinical trials and FDA approval may be required before Health Care software can be used. Financial companies must protect client data and defend themselves against fraud and cybercrime. Insurance companies rely on historical data for their analysis and models. Other companies must undergo rigorous regulatory procedures for specialty parts. After-market aerospace parts, for example, are critical to their customers even though they represent a small fraction of the overall cost. In this case, the question is not whether a customer can switch, but whether it is in their best interest to do so.
  3. Network effects. Companies that effectively meet the needs of participants create networks that are not easily replaced or replicated. Online markets, trading, and social media platforms bring together participants by serving their exclusive needs. These networks become more efficient and useful as more participants join, deepening their competitive advantages. Online market platforms, for example, benefit from network effects by attracting a high number of loyal consumers, which attracts merchants, who offer the widest variety of products at competitive prices. This attracts more loyal consumers and so on, resulting in a beneficial, upward spiral. On social media, active users with consistently high engagement can be more effectively targeted by advertisers, which results in high returns for ad expenditures. On exchanges, the wide availability of contracts and healthy volumes creates liquidity, which attracts more traders to the platform, which creates more contracts and liquidity, which benefits the traders. Companies that run payment networks connect millions of businesses with billions of consumers globally, enabling seamless commerce. In each of these examples, it is extremely difficult for a competitor to establish a rival network that can attract the number of participants (and thus generate the scale) needed to be successful.
  4. AI support. As companies look to leverage AI for their businesses—a critical, idiosyncratic, and complex task—some will look to new breeds of software companies. These companies integrate with and service their clients’ needs, tailoring and improving their offering through feedback, rather than selling an off-the-shelf software package. Such software provides high ROI and auditable security compliance at a reasonable cost. Even if cheaper, Large Language Model (LLM)-coded software solutions became available, they would still have to be integrated and maintained into the enterprise’s architecture, and they would have to link to outside LLMs for AI capability (which could be more expensive, and risky, to run). We believe that these companies will become even more valuable in an agentic AI world, where software autonomously executes tasks based on user goals, operates with its own enterprise privileges, and must be monitored and controlled.
  5. Manufacturing complexity and accumulated know-how. Many of the leading-edge components of technology require deep expertise, experience, and resources. These represent self-reinforcing competitive moats that have been developed over decades and are, in the current environment, unlikely to be successfully challenged. These companies enjoy a self-reinforcing cycle: the higher yields of their chips attract more customers, which enables the company to earn higher cross-cycle returns on its investment (while the diversity of its customers also reduces the overall cyclicality of the business). This funds more R&D, which enables the company to develop the next generation of chips more quickly and with higher yields than competitors, which widens their lead further. Competitors would have to make an extraordinary investment to build fabrication facilities, which would take years to construct. The facility would have to be staffed by an expert workforce that would produce sophisticated chips efficiently enough to be competitive. Even this might not be sufficient, because the company would need to win enough customers to amortize fixed costs and accumulate yield-learning data. In the meantime, lead chip developers will have innovated further, widening the capabilities gap. Rising chip complexity increases the switching costs for customers, as changing suppliers would take years, add major costs, and risk product delays, which could negatively impact their competitive positioning. Companies that design, develop, and supply semiconductors have also spent years building co-design relationships, custom silicon, and networking capabilities.
  6. Switching costs. Companies must consider the potential advantages of new AI capabilities with the potential risks of switching. In some cases, companies have sensitive, regulated information and processes that protect it. Other companies have deep relationships built on trust and reliability. In many cases, changing to a new AI solution would take years, add major costs, and risk product delays. Some companies serve a very specific customer base and provide increased value by giving each customer the benefit of understanding (using hard-to-compile domain-specific data) what is going on in their industry. Platforms for service trades such as plumbing and HVAC, for example, can offer lead generation, job bookings, dispatching, estimating jobs, customer communications, and payments/financing. Each trade has its own specific characteristics and regional data on pricing, competition, service times, and contract terms. It is not easy to switch software with these capabilities, particularly because it helps businesses automate their processes and minimize the overall personnel needed. Another example, integrated construction software, is required by many major general contractors in order for subcontractors to be able to participate in a construction project. The software combines computer-aided design software blueprints with job scheduling, cost estimations, materials costs, and change order management. In this manner, the job site can be coordinated among all the different parties involved in the construction project. This community-oriented platform is not easily replaced.

AI Disruption: A Long-Term Investment Opportunity

AI is a generational paradigm shift in how enterprises and individuals engage with technology. While the lack of visibility into the impact of evolving AI capabilities led to sell-offs across an array of industries in the first quarter, we believe a number of companies have been oversold. Recent declines were valuation multiple compression rather than deteriorating fundamentals, and we believe the combination of geopolitical uncertainty and early stage AI apprehension has created an attractive buying opportunity for many companies. In addition, we believe the long-term potential of AI to increase productivity, strengthen businesses, and generate growth has been largely underestimated by investors.

During periods of heightened uncertainty, it is critical to maintain long-term perspective. This era is marked by change, but the characteristics of great companies remain the same. Going forward, we believe that, as the impact of AI becomes visible, investors will be able to correctly distinguish the AI losers from winners.

Ultimately, in our view, the true risk would be allowing short-term noise or rapidly shifting AI narratives to influence long-term decision-making. We continue to focus on companies whose competitive advantages that we believe are reinforced, not eroded, by the accelerating adoption of AI and whose business models support durable, compounding growth.

Stock Names and Rationales

These companies, which are held in Baron Capital funds as of March 31, 2026, are examples of holdings that offer exposure to AI themes or competitive advantages.*

  • Strateg(ies)Rationale
    Baron Discovery StrategyAdvanced Energy is a leader in power conversion and control, with strong market share driven by energy efficiency and power density in its core applications. The company remains well positioned for the current upcycle in semiconductor equipment business and is benefiting from a refocused data center business emphasizing high-value, sole-source solutions growing alongside AI infrastructure buildouts. Advanced Energy is also seeing signs of recovery in its industrial and medical markets, generates significant cash flow, and maintains an active acquisition pipeline.