In our first-quarter outlook, we flagged four key themes: elevated equity valuations, rising market concentration, geopolitical uncertainty, and the extraordinary amount of capital flowing to artificial intelligence. Three months later, each theme has proven remarkably resilient. As reflected in the charts below, economic growth has held up better than expected and earnings estimates have stabilized and are trending upwards. While geopolitical tensions have reinforced inflationary pressures, AI remains the dominant force driving market leadership.
Markets Are Beginning to Ask a Different Question
More important than market performance this quarter was the evolution of the conversation surrounding AI. The debate is shifting from whether AI will be transformative to how that transformation will occur, who will capture the economic value, and whether current levels of investment will ultimately earn attractive returns. As this report from S&P Global highlights, there is growing concern over whether the pace of investment spend will be supported by comparable revenue growth. Whether the capital expenditures earn an attractive return is, in our view, the defining question of the decade.
Great Technologies Don’t Always Produce Great Investments
For long-term investors, that distinction is everything. History reminds us that transformative technologies and transformational investments are rarely the same thing. Railroads reshaped commerce. Electricity transformed manufacturing. The internet changed nearly everything. Each became indispensable. Yet each also punished investors who financed the revolution indiscriminately. As competition intensified, capital flooded in, and profitability often lagged the underlying promise of the technology.
While Jeremy Grantham is often associated with market skepticism, we find one of his observations particularly relevant today: the potential of AI is not the argument against a bubble, but rather the precondition for one. Truly transformative technologies attract capital because their potential is real. If AI were merely another speculative fad, companies would not be committing trillions of dollars of capital and significant portions of free cash flow to pursue it. The more compelling the opportunity appears, the stronger the incentive becomes to invest aggressively, sometimes beyond what future returns ultimately justify.
Railroads changed the world and still produced some of the largest busts in financial history. Once the dust settled, however, demand remained, the infrastructure endured, and the surviving businesses compounded value for decades. AI may ultimately follow a similar path—creating enormous economic value while producing a much less straightforward outcome for investors.
Importantly, we are not questioning the technology. We are questioning the economics surrounding today’s investment decisions.
On the other hand, many investors we respect analyze the same data and reach the opposite conclusion. Their argument deserves consideration. Managers such as Gavin Baker of Atreides describe AI as one of the most significant technological adoption cycles in modern economic history, pointing to demand that appears constrained less by customer interest than by the physical availability of power, advanced semiconductors, and computing infrastructure. Recent revenue growth among leading AI developers has been remarkable by any historical standard, suggesting that at least some portion of today’s capital spending is responding to real and rapidly growing demand rather than speculative enthusiasm.
There is also another way to interpret one of the key risks we and others have highlighted. The cost of AI inference has fallen dramatically, raising legitimate questions about the durability of current business models. Yet history suggests that falling costs can also expand markets rather than diminish them. Lower computing costs enabled the internet economy, just as declining semiconductor costs expanded the market for personal computers and smartphones. AI may ultimately follow a similar path, where lower prices unlock new use cases faster than they compress industry profits.
Whether falling costs erode economic returns or accelerate adoption sufficiently to expand the overall opportunity set remains, in our view, one of the defining unanswered questions of this cycle. Reasonable investors can disagree.
What we find most interesting is that both the bullish and cautious perspectives ultimately converge on the same issue: capital allocation. The central question is no longer whether AI will transform the economy—we believe it will. The question is whether the unprecedented capital being deployed today will generate returns that justify the investment. History suggests those are not always the same thing.
The Capital Allocation Challenge
Historical comparisons remind us that transformative technologies often experience periods of excessive optimism. They are less useful, however, in explaining what makes today’s investment cycle unique. Previous infrastructure revolutions were built around long-lived assets that generated durable cash flows over decades. The AI buildout may prove economically different.
Rather than focusing solely on the magnitude of today’s capital spending, we think investors should focus on three forms of duration: the economic life of the assets being built, the durability of the revenues they are expected to generate, and the duration of the capital committed to finance them.
Previous infrastructure revolutions relied on assets expected to generate value for decades. AI infrastructure appears different. As the chart below illustrates, many of the most important components of today’s buildout have useful lives measured in years rather than decades. At the same time, competition is reducing the cost of AI inference and raising questions about how durable today’s revenue models will ultimately prove.
Capital is always scarce, regardless of the size of the balance sheet. Every dollar invested in AI infrastructure is a dollar unavailable for another opportunity. While the transformative effect of AI is no longer a debate, the better question is whether the next dollar invested in AI is likely to earn a more attractive return than the next best alternative. Ultimately, that is the essence of capital allocation.
The next phase of this investment cycle is unlikely to be defined by whether AI transforms the economy. It will be defined by whether the returns generated justify the unprecedented capital being deployed.
Portfolio Implications
Assuming that our analysis is directionally correct, the obvious question becomes: What should investors do about it?
The answer, in our view, is not to make a dramatic portfolio shift. It is to become increasingly intentional about where risk is taken and how capital is allocated. Attempting to avoid AI altogether would likely create a different—and potentially greater—risk: missing one of the most important drivers of long-term productivity and economic growth.
Equities: Diversification Requires More than Asset Allocation
As the chart below illustrates, the increasing concentration of AI-related companies within public equity markets has reinforced a theme we’ve been discussing with clients throughout the year: diversification increasingly requires more than simply owning different sectors or geographies.
Market concentration is a bigger risk than the usual index-weight description suggests. For much of the past decade, today’s largest technology companies competed in adjacent markets. Increasingly, they are competing for leadership in AI. That convergence raises an important investment question: are these companies creating value, or competing away one another’s future returns?
Broad market exposure remains a cornerstone of our long-term investment philosophy, and passive investing (particularly in U.S. large-cap equities) has historically been one of the most effective ways to participate in long-term economic growth. We continue to believe it deserves an important role in most portfolios.
That said, elevated valuations have changed the way we think about liquidity. Our philosophy has always been that when capital needs to be redeployed, we prefer to fund those decisions by selling assets trading at or above our estimate of fair value rather than assets we believe remain undervalued. Today, U.S. large-cap equities generally meet that definition. We are not advocating broad reductions in equity exposure, but current valuations provide a more attractive opportunity to rebalance than we have seen in several years.
At the same time, we are looking beyond traditional definitions of diversification. Increasingly, we are evaluating managers based not only on sectors, styles, or geographies, but also on how they are positioned relative to AI. Some managers are investing alongside the companies building the underlying infrastructure, while others are identifying businesses that stand to benefit from AI adoption without bearing the extraordinary capital burden required to build it. We believe both perspectives have a role in a diversified portfolio.
Private equity remains a strategic allocation for us, but it is not immune to the same forces shaping public markets. As the data below from PitchBook highlights, after signs of recovery in 2025, exit activity and transaction volumes slowed again during the first half of the year.
Geopolitical uncertainty, higher interest rates, and questions surrounding AI’s impact on business models caused buyers to become more selective. The same question we are asking of public markets applies equally to private markets: not whether AI will create value, but which businesses are positioned to capture it. Rather than viewing this as a reason to reduce exposure, we view it as further evidence that manager selection is becoming increasingly important. In an environment where financial engineering is less powerful and valuation expansion is harder to rely upon, we believe experienced managers who can create value operationally are likely to become increasingly differentiated.
Fixed Income: Looking Beyond Yield
The improving yield environment has created one of the better fixed income opportunity sets we’ve seen in several years. With long-term Treasury yields now above 5% and inflation remaining above the Federal Reserve’s long-term target, investors are once again being compensated for taking duration risk. While spreads remain relatively tight by historical standards, the overall opportunity set is considerably more attractive than it was just a few years ago when cash and high-quality bonds offered little real return.
That said, we continue to believe selectivity matters. As discussed earlier, the AI investment cycle is increasingly influencing credit markets in ways that extend well beyond publicly traded technology companies. Howard Marks recently highlighted the extraordinary growth of private credit—an asset class that has expanded from approximately $150 billion two decades ago to nearly $2 trillion today. At the same time, Jefferies’ research suggests software businesses may face increasing competitive pressure as AI accelerates innovation and compresses competitive advantages. These developments reinforce our preference for managers that emphasize disciplined underwriting, strong covenant protection, and senior positions in the capital structure.
Across our portfolios, managers remain broadly underweight software and technology-related credit, preferring businesses with more durable cash flows and less dependence on continued capital investment. We continue to believe private credit deserves a role in long-term portfolios, but only where investors are being appropriately compensated for illiquidity and where the underlying risks are well understood.
Real Assets: Infrastructure Beyond Technology
Real assets remain one of the more attractive areas in long-term portfolios. Unlike public equities, where valuations have grown demanding, many real asset sectors still reflect more balanced expectations and offer differentiated returns.
Artificial intelligence has also expanded our definition of infrastructure. Data centers have understandably received significant attention as demand for computing power has accelerated, but they represent only one component of a much broader investment opportunity. The AI ecosystem will require substantial investment in electrical generation, transmission infrastructure, industrial facilities, equipment leasing, fiber networks, and other physical assets needed to support the digital economy. As the chart below highlights, however, the opportunity set extends beyond AI.
We are approaching these opportunities selectively. We are not interested in owning AI infrastructure simply because it is associated with a compelling narrative. We are focused on experienced operators, reasonable leverage, disciplined underwriting, and investments where returns are supported by durable cash flows rather than optimistic assumptions about future demand.
More broadly, we value real assets as a source of diversification whose return drivers extend beyond a single technology theme.
Conclusion
Reasonable investors can look at today’s AI landscape and reach very different conclusions. Some see the early stages of a productivity revolution whose economic impact may exceed even the most optimistic forecasts. Others see echoes of prior technology cycles where capital flowed faster than profits ultimately materialized. Both perspectives may contain elements of truth.
What seems increasingly clear is that the debate is evolving. The market has spent the last two years asking whether artificial intelligence will transform the economy. We believe that question is increasingly settled. The more important investment question is whether the extraordinary capital being deployed today will ultimately earn returns commensurate with the opportunity.
History offers a consistent lesson. Transformative technologies create enormous value for society, but not always for the investors who finance them. Railroads, electricity, and the internet each reshaped the world. Along the way, they also experienced periods of intense competition, excessive capital investment, and significant investor disappointment before durable winners emerged.
AI may ultimately prove even more consequential. The challenge for investors is not determining whether the technology matters. It is determining where value will accrue, who will capture it, and whether current prices adequately reflect both the opportunity and the risks.
As long-term investors, our objective is not to predict every twist in that evolution. It is to allocate capital thoughtfully amid uncertainty. That means focusing on valuation, diversification, manager selection, and the durability of future cash flows rather than becoming overly dependent on any single outcome. For portfolios, we believe the appropriate response is neither enthusiasm nor avoidance. It is disciplined capital allocation. That means remaining invested, diversifying sources of return, and maintaining the flexibility to rebalance and redeploy capital as opportunities evolve.
About Thierry J.D. Brunel
Thierry joined Matter in 2013, bringing years of experience in family office and wealth management. He previously worked in investment research and portfolio management roles at Convergent Wealth Advisors and GenSpring Family Office. At Matter, Thierry leads the investment committee, advising families on portfolio strategy and governance. A Wake Forest University graduate, Thierry has a diverse international background. He’s active in his community, serving as an assistant coach for the John Burroughs School Varsity football team in St. Louis.
Other Investment Blogs
Important Information
This material is for information purposes only. The views, opinions, estimates and strategies expressed herein are based on current market conditions and are subject to change without notice. Any companies referenced are for illustrative purposes only, and are not intended as a recommendation or endorsement. Any views, strategies or products discussed in this material may not be appropriate for all individuals and are subject to risks. Investors may get back less than they invested, and past performance is not indicative of future results. Asset allocation/diversification does not guarantee a profit or protect against loss.
Nothing in this material should be relied upon in isolation for the purpose of making an investment decision. Certain information contained in this material is believed to be reliable; however, Matter LLC does not represent or warrant its accuracy, reliability or completeness, or accept any liability for any loss or damage (whether direct or indirect) arising out of the use of all or any part of this material. No representation or warranty should be made with regard to any computations, graphs, tables, diagrams or commentary in this material, which are provided for illustration/reference purposes only. Any projected results and risks are based solely on hypothetical examples cited, and actual results and risks will vary depending on specific circumstances.
Forward-looking statements should not be considered as guarantees or predictions of future events. Nothing in this document shall be construed as giving rise to any duty of care owed to, or advisory relationship with, you or any third party. Nothing in this document shall be regarded as an offer, solicitation, recommendation or advice given by Matter LLC and/or its officers or employees.