Artificial intelligence has become one of the most widely discussed themes in markets. Throughout 2025, AI-related stocks have performed exceptionally well. However, many of the most prominent AI-linked companies have recently met resistance, with volatility increasing noticeably across semiconductor makers, cloud providers, and AI-exposed software businesses. This shift in momentum has brought renewed questions about whether the enthusiasm surrounding AI resembles a bubble or whether the recent volatility reflects a normal adjustment within a much longer-term investment cycle.
Some concerns may reflect real risks, but others stem from the growing narrative of an “AI bubble” itself. As with many emerging technologies, it is worth asking whether we are witnessing a bubble or simply a bubble of talking about a bubble. A balanced perspective requires understanding both the concerns that have emerged and the fundamental trends that continue to support AI adoption, compute demand, and investment activity across industries.
Corporate Focus, Adoption and Productivity Gains
Mentions of artificial intelligence on corporate earnings calls recently reached the highest level in ten years.[1] While some companies may emphasize AI because investors are paying close attention to the theme, the broader trend reflects meaningful integration across business operations. Organizations are adopting AI to improve workflows, enhance decision-making, reduce costs, and strengthen competitive positioning.
This corporate focus aligns with broader adoption trends. According to Federal Reserve survey data, generative AI usage has expanded rapidly, with more than half of U.S. adults now reporting they use these tools. Adoption is progressing faster than earlier technological shifts such as the personal computer or the early internet. As shown in the chart below, usage continues to rise steadily across both work and nonwork activities. Workers who use generative AI report notable time savings, and when aggregated across the labor force, these efficiencies translate into measurable national productivity gains. Early estimates suggest generative AI may have contributed up to 1.3 percent to labor productivity growth since late 2022, although the authors note that these gains cannot be attributed to AI alone.[2]
Chart. Generative AI Adoption: Overall, Work, and Nonwork
Source: Federal Reserve Bank of St. Louis
These developments indicate that the heightened attention from corporate America is grounded in real use cases and potential early productivity benefits rather than rhetorical positioning alone.
Investment Spending Is High, but Underlying Demand Is Also Strong
Capital spending on artificial intelligence infrastructure continues to rise, led largely by hyperscalers. These organizations operate expansive data-center networks with specialized, power-intensive compute designed to support advanced AI models. Although the pace of investment can raise concerns about overshooting, recent commentary suggests hyperscalers are investing aggressively as they work to scale infrastructure to support growing AI workloads. As illustrated in the chart below, aggregate capital spending by major hyperscalers is projected to grow at a compound annual rate of roughly 40 percent through 2027.
Chart. Hyperscalers Capital Expenditures
Source: Company data, LPL Research, Bloomberg (Consensus) 11/05/25
Disclosures: Past performance is no guarantee of future results. Any companies or options referenced are being presented as a proxy, not as a recommendation. Estimates may not materialize as predicted and are subject to change.
On Microsoft’s most recent earnings call, the company’s Chief Financial Officer noted that demand for Azure services is “significantly ahead of the capacity we have available,” explaining that even after spending $34.9 billion on capital expenditures in the latest quarter (nearly $10 billion more than the prior period) they still have not caught up and will increase spending in the next quarter.[3] In a separate interview with OpenAI CEO Sam Altman, Microsoft’s CEO explained that the company’s primary constraint is not excess compute but a shortage of power and usable data-center capacity, implying that some GPUs remain uninstalled because the necessary electrical and physical infrastructure is not yet in place.[4]
These comments highlight the real bottlenecks shaping today’s AI investment cycle. Companies are increasing capital spending not because demand is slowing, but because they must scale infrastructure quickly to support adoption. The need to expand capacity is becoming increasingly visible in the broader power and data center landscape.
Power demand underscores this challenge. In Texas, ERCOT received grid interconnection requests tied to data centers totaling 56 gigawatts in 2024 and 205 gigawatts in 2025, and the grid operator has noted that this rapid surge of large load projects is being built faster than traditional transmission planning can accommodate.[5]
Data center construction remains one of the few categories of nonresidential building that continues to grow, as shown in the chart below.[6] This pattern suggests that today’s investment is responding to real and persistent demand for compute resources rather than speculative expansion.
Table. Data Center Construction vs Nonresidential Structures Ex Data Centers
Financing Structures Are Evolving, but Large Investors Remain Financially Strong
Some large AI projects are being financed through special purpose vehicles (SPVs) and other nontraditional arrangements. These structures introduce complexity but allow companies to fund large data center buildouts without adding the full amount of debt to their balance sheets. Meta recently secured about 60 billion dollars for data center development, with approximately half raised through an SPV that does not appear as corporate debt. Although SPVs are common in project finance, their growing use in AI infrastructure has drawn attention because similar off-balance sheet structures played a role in Enron’s collapse and in the mid-2000s mortgage credit boom.[7]
In some cases, semiconductor suppliers are helping to finance AI developers that rely on their technology, including companies that are not yet profitable but continue to attract funding. This can create a circular dynamic in which suppliers support customer growth, while customers depend on investor capital to continue scaling.
Table. Example of Deal Circularity
Source: LPL Research, Bloomberg 10/08/25
Disclosures: Past performance is no guarantee of future results. Any companies or options referenced are being presented as a proxy, not as a recommendation.
Circularity is not necessarily a problem, but it can create vulnerabilities if revenue expectations prove too optimistic or if capital availability tightens. The interconnected financial relationships among chipmakers, cloud providers, and AI developers highlight the importance of monitoring how funding moves through the ecosystem. While today’s structures differ from the leverage driven excesses of past speculative periods, they remain an area where thoughtful caution is appropriate as AI investment accelerates.
Market Volatility Does Not Necessarily Signal the End of the Investment Cycle
Recent volatility in AI-related stocks has prompted questions about whether the market is entering a broader downturn. Periods of volatility are a normal part of market behavior and often serve to reset expectations after extended gains. With equities rising for six straight months, it is not surprising to see greater scrutiny of AI leaders, and this reassessment can be healthy for market discipline.
Despite strong performance this year, sentiment remains unusually cautious. AAII bearish sentiment has been above its historical average for most of the past year[8], while consumer sentiment is still near levels last seen during the 2022 inflation shock[9]. The Fear & Greed Index, as of November 19th, also reflects a guarded tone. Markets, in other words, continue to climb a wall of worry.
Meanwhile, corporate fundamentals remain supportive. Analysts have been raising earnings expectations, and FactSet now projects S&P 500 profit growth of 11.7 percent in 2025 and 13.9 percent in 2026.[10]
Conclusion
AI’s growing visibility and the pace of related investment inevitably raise concerns about a potential bubble. Yet the market backdrop remains defined more by caution than euphoria, even as adoption and investment continue to advance. Recent volatility may reflect a healthy reassessment rather than an end to the AI trend.
With sentiment still restrained and fundamentals improving, we will continue to follow developments with measured optimism, remaining attentive to both the risks and the persistent structural drivers shaping AI’s long-term investment cycle.
Content in this material is for general information only and not intended to provide specific advice or recommendations for any individual. All performance referenced is historical and is no guarantee of future results. All indices are unmanaged and may not be invested into directly.
All investing involves risk including loss of principal. No strategy assures success or protects against loss.
The economic forecasts set forth in this material may not develop as predicted and there can be no guarantee that strategies promoted will be successful.
Stock investing includes risks, including fluctuating prices and loss of principal. Any company names noted herein are for educational purposes only and not an indication of trading intent or a solicitation of their products or services. LPL Financial doesn’t provide research on individual equities.
[1]FactSet, John Butters, “Highest Number of S&P 500 Earnings Calls Citing ‘AI’ Over the Past 10 Years,” September 5, 2025.
[2] Bick, Alexander; Blandin, Adam; Deming, David. “The State of Generative AI Adoption in 2025.” Federal Reserve Bank of St. Louis, On the Economy Blog, November 2025.
[3] Ford, Brody & Matt Day. “Microsoft Data Center Crunch Persists Despite Heavy Spending.” Bloomberg, October 29, 2025.
[4]Yahoo Finance, “Microsoft CEO Satya Nadella: Company doesn’t have enough electricity to install all the AI GPUs,” November 2 2025.
[5] Martin, Arcelia; Inside Climate News. “Explosion of data centers causes planning struggles for Texas power grid.” The Texas Tribune, October 30, 2025.
[6] LPL Research. “Weekly Market Commentary: AI Infrastructure — A New Pillar of Economic Growth.” November 10, 2025.
[7] Arroyo, Carmen. “Meta, xAI Spread Risks of AI Splurge With Off-Balance-Sheet Debt.” Bloomberg, October 31, 2025.
[8] AAII. “AAII Sentiment Survey: Pessimism Takes Flight.” November 13, 2025.
[9] Grossman, Matt & Goldfarb, Sam. “U.S. Consumer Confidence Slides in November.” The Wall Street Journal, November 7, 2025.
[10] FactSet Research Systems Inc., “Earnings Insight: November 14, 2025,” FactSet, November 14 2025