Artificial intelligence will be a major technology for the next generation and likely to shape our future. Effects on business automation, medicine, science, education, security, consumer services, and software could be far-reaching. Changes to come are very likely to be as important as the rise of personal computers or the internet. We do not view AI as a fad, but the current surge in AI capital spending is still a cycle.

The current AI boom is being expressed first through a massive infrastructure buildout. Data centers, semiconductors, and networking gear are expanding fast. Related infrastructure, including power and cooling systems, are also growing quickly. Research and development of software tools and cloud platforms are seeing rapid growth, too. McKinsey estimates that global data centers may need $6.7 trillion in capital by 2030 [1]. This includes $5.2 trillion for AI workloads. Goldman Sachs predicts that annual AI spending could hit about $765 billion in 2026 and $1.6 trillion by 2031 [2]. The incredible scale of all this is what is grabbing investors’ attention despite limited visibility into how such massive investments will generate sufficient profit.

Great Expectations

The market’s enthusiasm for AI is understandable, but the role of expectations in valuation cannot be overlooked. The multiples that most investors are looking at to justify today’s valuations depend heavily on discounting extraordinary growth in future earnings. In many cases, share prices are supported by high hopes for massive profit growth, but history shows that times of high expectations can amplify swings in both directions – up swings and down swings. Today, the S&P 500 technology sector trades at 6,728 which is 23 times the next 12-month earnings estimate of $289. This sounds very reasonable. However, that $289 estimate is up nearly 70% from a year ago, and more heady growth is expected, which is carrying all the weight in justifying valuations. In fact, the long-run earnings growth rate, which measures expected earnings growth by analysts over the next several years, has hit an astounding 30%, according to Bloomberg estimates (Chart A, below).

Chart A: S&P 500 Technology Sector Growth Rate Estimates

Why Investment Spending Creates Profits

Capital spending for AI and AI infrastructure has already become a powerful support for reported profits. In short, capital spending by firms comes back to firms as profit and juices up growth for a time. A hyperscaler’s (massive cloud service providers that offer highly scalable computing, networking and storage services) data center investment budget, for example, creates sales for chip makers, networking firms, and electrical equipment suppliers. It also benefits engineering firms, utilities, and construction contractors. The AI investment spending boom gets multiplied throughout the economy, feeds back into profits, and is the main reason why equity market earnings expectations are now as high as they are.

We include a granular discussion of the macroeconomic formula in an end note to this commentary, but for now it suffices to say that rising investment will increase business profits (savings) all else being equal.

Demand For AI Or Demand For AI Investments?

Increasingly, the AI buildout is also becoming a capital-markets phenomenon. Specialized investment vehicles and private credit funds help buy AI infrastructure, and vendor-supported financing also supports these purchases. These structures can help users adopt AI by lowering initial costs, but this kind of financing should not be seen as “business as usual.” The need for AI infrastructure now depends on two things: the demand for AI services and how eager investors are to invest for good returns.

This distinction may prove important as the cycle evolves. Capital is flowing not just because businesses need computing power. The flow of capital is also because investors think AI will bring amazing returns. A question that may become increasingly relevant over time is this: Are AI hardware providers selling chips and other equipment because the world needs this much AI, or because investors are seeking high returns from AI? Today, the answer is likely both.

The next phase of the cycle may further illustrate this dynamic. Several of the most valuable private companies associated with AI, including OpenAI, Anthropic, and SpaceX, are widely expected to eventually access public markets. While the timing and structure of any future offerings remain uncertain, the potential size of these businesses suggests that the supply of AI-related investment opportunities is set to expand. In that sense, the AI boom is not simply driving demand for chips, data centers, and power infrastructure; it is also creating demand for investment vehicles through which investors can seek exposure to the theme.

As long as user demand and investor interest support each other, the cycle can stay very strong (Chart B, below). Strong demand encourages investment. Investment supports revenues and profits. Rising profits reinforce confidence. Confidence attracts additional capital. Additional capital supports further investment. History has shown that capital cycles eventually hit an inflection point. From that inflection point on, investors focused on earning a return on capital invested.

Chart B: The Self-Reinforcing Capital Spending Cycle

This self-reinforcing cycle can shape dynamics within an economy and among different firms’ ability to compete for capital and investor attention. For example, today’s capital spending growth is focused heavily on business investment in equipment and intellectual property, largely related to AI. Conversely, investment outside of AI is contracting (Chart C, below). An increasingly large share of corporate investment, financing activity, and earnings growth is being tied to a single theme: AI. Capital investment is now narrowly concentrated around the AI thematic just as S&P 500 market capitalization,  earnings growth expectations, and stock performance have all become concentrated around artificial intelligence.

Chart C: AI vs Non-AI Investment

The Return-on-Capital Test

The question is what happens when the cycle matures. Will end-user demand for AI services ultimately prove large enough to justify the extraordinary amount of capital being deployed today, or will profits fall short?

The likely long-term winners will be companies that can run a marathon, not a sprint. They need strong balance sheets. Durable profits are important, too. They must fund innovation without relying on always favorable capital markets and they must also get through the cooling-off periods. During these times, investors will almost certainly want more proof of returns on their capital than what they are asking for today.

In our view, the correct posture is to recognize AI as the big technological change that it is, but remain diligent in our strategy of owning quality at a reasonable price. This is why we are invested in AI through a variety of portfolio companies from data centers, to capital equipment providers, to supportive plant, energy, and infrastructure. At the same time, we recognize that today’s boom in capital spending, profit growth, and investor excitement is also subject to the dynamics of past capital spending cycles. For this reason, now is not the time to throw in the towel on maintaining a strong defense.

The long-term success of the cycle depends on demand for AI services. It needs to grow enough to justify the trillions of dollars of capital now being invested in anticipation of AI-related returns. Investing math tells us that when prices are high, all else being equal, future return expectations should be lower, not higher.

What It Means For Our Investors

The investment opportunity may improve as the cycle matures. Disappointments can lead to chances to buy strong innovators at better prices. The goal is to stay invested in strong businesses while avoiding overpaying for unrealistic expectations and remaining flexible enough to invest when the market shifts from excitement to discipline.

We will look for proof that businesses may be able to generate returns on today’s capital investments as we will look to choose quality companies at fair prices.

——————————————————–

[1] Jesse Noffsinger, Mark Patel, and Pankaj Sachdeva, The Cost of Compute: A $7 Trillion Race to Scale Data Centers, McKinsey & Company, April 28, 2025.

[2] George Lee and Lucas Greenbaum, Tracking Trillions: The Assumptions Shaping the Scale of the AI Build-Out, Goldman Sachs Global Institute, May 1, 2026.

Note on Macroeconomic Accounting and Relation to Profits:  The starting point for a macroeconomic profit identity is the familiar macroeconomic formula for gross domestic product (GDP). Gross domestic product (GDP) is equal to Consumption (C) + Investment (I) + Government Spending (G) + Exports (X) – Imports (M). GDP is also approximately equal to gross domestic income (GDI), which can be expressed as income that is consumed (C), saved (S), or paid in taxes (T). Setting GDP = GDI and substituting in the constituents for each, we get GDP = C + I + G + (X – M) = C + S + T. We can further divide savings into two parts: household savings (SH) and business savings (SB). Business savings represent the portion of corporate earnings retained within the firm or distributed as dividends to shareholders. We can now solve for business savings using basic algebra. Rearranging the formula yields: SB = (I – SH) + (G – T) + (X – M). In other words, business savings are equal to investment minus household savings plus the government fiscal deficit plus net exports. All else being equal, increases in investment, government spending, or exports and decreases in household savings, taxes, or imports will raise business savings, and vice-versa. Macroeconomics shows us that, holding all else constant, if businesses collectively decide to spend more on factories, software, data centers, equipment, or inventories, that spending creates income somewhere else in the corporate sector, some of which ultimately appears as profits, retained earnings, or dividends. This accounting relationship helps explain why periods of heavy capital spending are frequently associated with rising aggregate profits in the short run. The implication is that large investment booms can support corporate profits even before the underlying projects prove economically successful. The spending itself becomes income somewhere else in the economy.