History Rhymes

“The only thing new in the world is the history you do not know.”

– Harry S. Truman

It often feels as though we are living through an especially volatile period in history, with major geopolitical developments and technological breakthroughs making headlines almost daily. The first half of 2026 has certainly reinforced that perception: the year began with the apprehension of Nicolás Maduro, followed by the rapid escalation of the conflict with Iran, continued shifts in trade and tariff policy, and astonishing advances in artificial intelligence.

The reality is that the world has always been an unpredictable and unnerving place, and people from almost every era have believed they faced a uniquely difficult period. Investors have always had to contend with wars, natural disasters, technological change and periods of irrational market exuberance. While the precise nature of our challenges differs from those of previous generations, history still offers valuable lessons. Unfortunately, there is no precise roadmap on how to proceed, but the past does provide some hints as to how the future may unfold.

We have firmly exited the post-global financial crisis (GFC) environment, which was characterized by low interest rates and subdued inflation, and are still adjusting to a new market regime with higher interest rates, stickier inflation, and rapid advancements in AI that require massive investments in new infrastructure. As this paradigm continues to evolve, new risks and opportunities are emerging in ways that may rhyme with the past.

Recent Economic & Market Activity

Investors entered the year expecting inflation to decline, the Federal Reserve to begin cutting interest rates, and consumer spending to receive a boost from enhanced tax refunds under the One Big Beautiful Bill Act (OBBBA). Collectively, these forces were expected to keep US GDP growth in the solid, if unspectacular, 2% range. Expectations for GDP have proven correct, but not for the reasons anticipated at the start of 2026.

The Iran conflict caused energy prices to spike and inflation to remain elevated, which offset much of the benefit of lower-income consumers’ tax refunds and altered the course of Fed policy. The market is now pricing in rate increases later this year as the Federal Reserve appears increasingly focused on the price stability side of its dual mandate.

GDP growth and corporate earnings in the US have been supported by two key drivers: 1) AI capital expenditure (capex) growth and 2) consumer spending by higher-income households. While both are expected to remain supportive over the next twelve months, they represent a fairly narrow foundation for economic growth. There is also a risk that any reversal in enthusiasm for AI-driven equities could reduce the wealth effect supporting higher-income consumers. A broadening of activity would provide a more stable environment for further gains and would be a welcome development in future quarters.

In capital markets, global equities continue to produce stellar returns, with every broad market we track in positive territory and posting double-digit returns over the past twelve months. These returns were driven by strong earnings growth: S&P 500 earnings per share (EPS) year-over-year growth exceeded 23% in the second quarter, which allowed the forward price-to-earnings ratio to decline to 19.7x from 22.0x at the start of the year. Given elevated valuations, returns driven by earnings growth are far more sustainable in an environment where multiples are unlikely to expand further.

We have seen a rotation away from the dominant technology firms that have driven broad index returns over the past five years, as the US value sector, small-capitalization stocks, and international equities all outperformed the US growth index. This rotation represents a healthy development for diversified portfolios and should reward investors who have actively sought exposure beyond mega-cap US technology companies.

Fixed income markets generated more modest returns as geopolitical uncertainty and a resetting of Fed policy expectations resulted in yield volatility and limited price appreciation. The dollar strengthened modestly after a year of weakness, as the US benefited from a flight to safety, energy independence, and expectations for higher overnight rates relative to other currencies.

Private markets were dominated by reports of investors seeking to redeem capital from private credit funds beyond the limited quarterly redemption gates many funds offer. Concerns emerged that loans to software companies were at risk due to competitive threats from AI, and investors rushed to exit from an asset class with inherently limited liquidity. The actual fundamentals in the direct lending space appear solid, with limited defaults and still-attractive yields. Several large IPOs, including SpaceX’s highly anticipated debut, have highlighted improving exit opportunities for private equity investors after several years of muted returns, with prominent private firms such as Anthropic and OpenAI also reportedly planning IPOs over the coming quarters.

AI: Echoes of Past Technology Revolutions

We have noted the importance of AI capex to GDP and earnings growth, and the scale of the spending is truly breathtaking. Spending by just the five “hyperscalers” approached $500 billion in 2025 and is expected to exceed $700 billion in 2026, more than 2% of US GDP. To put those figures into perspective, annual growth in total US consumer spending is roughly $1 trillion. In other words, AI capex by just five companies is approaching the scale of all annual growth in US consumer spending.

Beyond the staggering amount of capital being invested in data centers and compute capacity, the promise and potential of the technology are both extraordinary and disruptive. It is a true technological revolution that will have far-reaching impacts on workers, companies, and societies for years to come.

Fortunately, there are several past examples of technology breakthroughs that had enormous impacts on consumers and markets. Two in particular may serve as useful case studies: 1) the commercialization of electricity and 2) the dawn of the internet. In a recent white paper, Bubbles as a Feature Not a Bug, Jason Thomas of the Carlyle Group outlined a compelling case that recent investment in AI closely resembles the investments made in the 1920s to scale up electricity generation and distribution in the US. As large as the total addressable market (TAM) may be for AI applications, it is difficult to overstate the TAM for electricity in 1920. The seismic nature of electrification is at least comparable to the changes that AI will drive.

Like AI, the initial financial beneficiaries of the electricity buildout were the direct suppliers of products that represented bottlenecks in the system. Manufacturers of turbines and transmission lines drew in capital much as semiconductor and memory chip makers do today. Investors looking to play the theme piled into electric holding companies, which saw their earnings and valuations triple between 1925 and 1929. Those stocks eventually struggled as their return on invested capital fell short of expectations, but the technology itself was a smashing success: over 70% of homes and 80% of manufacturing were electrified by 1930.

The long-term beneficiaries of electrification were the downstream users and new industries the revolution created. Radio is an obvious example, but nearly every company exploited the new technology to grow, become more efficient, and create new products. Durable goods manufacturers; firms that produced items such as refrigerators, vacuums, and radios; saw their stocks return an annualized 11% in the 1930s despite the Great Depression. This is a reminder that the ultimate winners from an AI revolution may be the downstream users of the technology as adoption accelerates, and not just the creators of the hottest new large language model.

The internet mania of the 1990s exhibited some similar characteristics, as makers of routers, chips, and fiber optics saw initial gains but ultimately corrected when their investments failed to generate acceptable returns on capital. The benefits of the internet eventually accrued to industries such as media and retail that adapted their business models, or to entirely new sectors such as social media that were enabled by the new platform technology.

Two major risks stood out in these previous cycles: 1) the lack of an acceptable return on capital initially invested in the infrastructure fueling the revolution and 2) the source of that capital. In past cycles, much of the capital came from banks, the bond market, or equity investors in unprofitable companies, which created financial shocks when issuers defaulted on their debts or could not raise fresh equity. Today, most of the capital being invested comes from the free cash flow of the hyperscalers. This somewhat reduces the systemic risk from the capex buildout, but it still leaves equity markets exposed to firms that may be committing hundreds of billions of dollars to projects that may not earn acceptable financial returns. The ultimate profitability of these investments will likely be the primary driver of future performance for the hyperscalers, which today account for roughly 20% of the S&P 500.

Portfolio Implications

While a discussion of past cycles is interesting, the more important question is what lessons investors can apply today. For those with fully invested portfolios, trying to time the end of a bubble is neither realistic nor likely to be successful, potentially causing investors to miss attractive returns and realize unnecessary capital gains. A more productive approach is to understand a portfolio’s current concentration in AI-related holdings, take profits where exposure has become excessive, and add to sectors that may be overlooked today but are likely to be downstream beneficiaries of the technology over time.

Examining the top ten holdings and sector allocations of various indices provides a snapshot of a portfolio’s hyperscaler and AI exposure. A comparison of the top ten positions and sector allocations in the market-capitalization-weighted S&P 500, Russell 1000 Growth, Russell 1000 Value, and EAFE (International Developed Markets) Exchange Traded Funds (ETFs) is revealing.

Even a quick glance at this data is instructive. The S&P 500 and growth indices are both highly concentrated and heavily exposed to either the AI hyperscalers; Alphabet, Microsoft, Amazon, Meta; or to companies benefiting from those AI capex dollars – NVIDIA, Broadcom, & Micron. The US value index is less concentrated in AI and offers broader sector exposure in areas like financials, healthcare, and energy that can provide diversification away from the AI theme. International developed markets have even less direct AI exposure and are far more diversified than the US core and growth indexes, which carry significant concentrations in their top ten holdings.

We have been actively trimming growth and core positions to add to value and international equities over the past few quarters, and we believe that remains the appropriate strategy as the AI revolution evolves. While we suspect AI will ultimately mimic previous rapid shifts in technology, in which initial winners made way for a broader set of beneficiaries, the timing and magnitude of that transition is not something we feel capable of predicting at this point. There is also the possibility that AI resembles the internet era, in which several companies with entirely new business models – most notably Alphabet, Meta, and Amazon – became long-term massive winners. We would not want to forgo exposure to the next generation of winners by trying to avoid AI exposure altogether today.

AI is also having an impact on private markets. Private equity and venture capital firms are investing heavily in AI-related companies and industries, while real estate funds are raising capital to invest in data centers and related infrastructure. Our private market allocations have limited direct AI exposure. Our most recent private equity fund does have limited exposure to direct AI themes but also carries significant allocations to sectors such as consumer products, logistics, music royalties, industrial & energy services, and insurance. In real estate, we have no data center exposure and remain focused on housing, retail, and hospitality opportunities. Our private credit allocations have very limited AI-related exposure and are more oriented towards broad-based corporate loans and asset-backed real estate lending. We view these alternative allocations as additional diversifiers for portfolios permitted to own private assets.

Learning From History

The fear that we are experiencing another technology-driven bubble – one that will end like previous manias surrounding electricity or the internet – is top of mind for many investors. It is tempting to take an extreme position: to go all-in on maximum AI exposure and ride the wave until it crashes, or to avoid all risk and stay on the sidelines until the bubble bursts. We believe there is a more reasonable third option, in which investors harvest profits from their technology exposure to reallocate capital to less frothy portions of the equity market, add alternative asset classes with diversified return drivers, and maintain appropriate allocations to cash and fixed income to protect near-term spending needs. This more measured approach may not be revolutionary, but it is informed by the lessons of past cycles. We hope you are having a wonderful summer and enjoyed a memorable Fourth of July weekend celebrating the 250th birthday of our country. Please feel free to reach out with any questions, and we look forward to visiting with you soon.

Sincerely,