August, 2026
The era of artificial intelligence (AI) has arrived. Chatbots that simulate human conversation are now widely available, allowing anyone to interact with AI through simple question-and-answer interfaces. Taking this technology a step further, AI agents are already capable of performing complex tasks autonomously, including multi-step workflows involving decision-making, problem-solving, digital actions, and interactions with external systems. AI is also being rapidly integrated into physical technologies such as robotics, transportation, manufacturing, and consumer wearables, dramatically expanding their capabilities. Meanwhile, the models that power AI continue to improve at a remarkable pace, driven by companies and researchers around the world. As a result, the prospect of artificial general intelligence (AGI)—AI capable of performing any intellectual task a human can—is no longer such a far-fetched concept.
What we do know is that unprecedented amounts of capital are being invested to build the data centers that power AI. The computing capacity required to train and operate today’s advanced AI models is immense. As AI becomes more capable and adoption continues to grow, demand for these “AI factories” will increase accordingly. AI providers are racing to stay ahead of that demand to avoid capacity constraints, disappointing customers, or leaving revenue opportunities untapped.
It is estimated that more than $1 trillion will be invested globally this year alone to construct and equip AI data centers. Many of the companies funding these projects have taken on substantial debt and issued tens of billions of dollars of new equity despite uncertainty surrounding future demand and investment returns. Spending could increase even further next year. Beyond that, however, the outlook is less certain. Financing is only one challenge. Data centers also require enormous amounts of electricity and water, while facing growing opposition from local communities. This pace of investment is unlikely to continue indefinitely and may prove to be heavily front-loaded, with several years of exceptional spending followed by a meaningful slowdown. In fact, there are already signs the industry may have gotten ahead of itself, as some operators have begun leasing excess computing resources after previously suffering from capacity constraints.
This narrow group of AI-related stocks has increasingly traded as a single cohort and has become notably more volatile. After their substantial advances, investor enthusiasm may be giving way to valuation concerns as prices appear increasingly stretched relative to the broader market. At the same time, many companies with little direct exposure to AI infrastructure spending have recently begun to outperform, particularly during periods when AI-related stocks have pulled back. Market leadership appears to be broadening to include fundamentally strong businesses whose growth is less dependent on data center construction.
Our investment approach remains focused on identifying long-term winners rather than attempting to time cyclical market movements. Many of the companies we own are already benefiting from AI or are helping enable its adoption. As a transformational technology, AI will create compelling investment opportunities for years to come, and we expect to participate in its continued growth and success.
