While the term “Magnificent Seven” is still commonly used, it no longer adequately describes stock market leadership. It remains a convenient media label, but it increasingly obscures a more consequential shift - artificial intelligence has moved beyond a handful of mega-cap consumer technology companies and into a broad industrial and infrastructure investment cycle.

We have tried to make sure our clients have enough cash on hand for the next two to three years to support withdrawal requirements without needing to sell stocks. That cash is invested in very short-term instruments so it can still earn a return without taking undue price risk.

The Magnificent Seven has not disappeared as a group of influential companies. It has simply become too narrow—and too internally inconsistent—to serve as the market’s primary theme. Alphabet, Amazon, Apple, Meta, Microsoft, Nvidia, and Tesla have very different earnings drivers. Nvidia is tied closely to AI compute demand; Microsoft and Amazon benefit from cloud adoption; Alphabet and Meta combine AI investment with advertising; Apple depends heavily on consumer-device cycles; Tesla is an electric-vehicle, energy-storage, and manufacturing story. Their valuations, capital needs, competitive pressures, and sensitivity to the economy are increasingly different.

In our mutual fund and ETF portfolios, we reduced software, gold, and S&P 500 index exposure in early January. We put the proceeds into industrial materials, Asian and Emerging market equities instead. To date, those markets are outperforming the S&P 500. In March, we took gains in individual stock portfolios where single positions had become too large or had risen an extraordinary amount. Paying taxes is preferable to losing money.

AI is now an infrastructure cycle

AI requires far more than leading software platforms and chip designers. It needs memory, semiconductor manufacturing equipment, networking gear, servers, cooling, data centers, power generation, grid investment, and construction capacity.

These needs expand the investment opportunity to companies across the semiconductor, industrial, utility, networking, and infrastructure sectors. The more useful question is no longer, “Which Magnificent Seven company wins AI?” It is, “Which companies capture the spending needed to build and operate AI at scale?”

How Leadership has broadened

‍

The portfolio implication

We’re not suggesting that investors abandon the mega-cap leaders. Several remain outstanding businesses and central participants in AI adoption. The issue is that owning cap-weighted positions can leave a portfolio overly concentrated in the most visible layer of the AI theme.

A stronger approach separates AI exposure across platforms, chips, manufacturing equipment, networking, data-center infrastructure, and power. This gives investors more ways to participate if spending remains durable while reducing reliance on a few companies. The next market leaders may come from the factories, networks, power systems, and infrastructure behind them. Further, the companies that use AI to develop new products and/or expand their growth may end up being the long-term winners.

The Second Quarter AI Boom Gave Way to a July AI Bust. What Happened?

The broad AI-industrial theme sold off in July because investors stopped rewarding the *promise* of AI spending and began demanding proof of returns. The decline was not confined to the former Magnificent Seven; it extended through semiconductors, equipment, networking, servers, data centers, power, cooling, and related infrastructure. The entire sector had become crowded after a powerful run. Many AI-linked companies were priced for sustained hypergrowth. Once expectations became that elevated, even good results disappointed when they were merely in line rather than meaningfully better than expected.

Factors that drove the July pullback:

• Profit-taking and concentration. Investors often trim large winners after a strong first half, especially where positions have become oversized in indexes and portfolios.
• Valuation compression. Higher interest rates and reduced expectations for Fed easing lower the present value investors assign to long-duration growth earnings.
• Questions about AI monetization. Cloud platforms and software companies may be spending heavily on data centers and chips before revenue from AI products fully catches up. Markets then focus on capital-expenditure intensity, depreciation, and return on invested capital.
• Capex-cycle fears. Semiconductor and infrastructure shares discount future orders well in advance. Any sign of slower order growth, shipment constraints, customer digestion, or postponed projects can cause the entire chain to reprice.
• Crowded thematic positioning. When the same AI winners are widely owned, a shift toward risk reduction can force broad selling across companies with very different fundamentals.
• Geopolitical and policy risk. Export restrictions, tariff threats, supply-chain disruption, and uncertainty around global demand can directly affect chips and indirectly pressure the wider infrastructure complex.
• Rotation into neglected market sectors. Investors may move capital from expensive technology and AI infrastructure shares toward energy, materials, financials, defensives, or smaller companies when valuations and macro conditions favor a broader market.

Companies that are part of the expanded AI theme can trade together during a risk-off episode even though the businesses have different fundamentals. That correlation is a market-positioning phenomenon and can present buying opportunities in some of the stocks.

What to Watch After the AI Selloff

The next phase depends on hard evidence rather than lofty expectations about the future. The list below shows what will likely influence stock direction going forward.

• AI capital-expenditure guidance from the hyperscalers (Alphabet, Amazon, Meta, Nvidia, etc.)
• Order backlogs and lead times for chips, memory, networking, servers, and cooling
• Gross-margin trends, especially for hardware suppliers
• Data-center leasing, power availability, and construction timelines
• Bond yields and real rates, which influence long-duration equity valuations
• Evidence that enterprise AI adoption is producing revenue growth or productivity gains

General Market Outlook – The Good, the Bad, and Our Take

The Good – We thought Jeremy Siegal said it best

“We are seeing double-digit earnings growth despite second-quarter real GDP growth of only about 1.5%, while the magnitude and breadth of earnings beats have been exceptional. Outside of rebounds from severe recessions, it is difficult to find many comparable periods. Earnings are strong, forward guidance is strong and corporate finances remain healthy.”  Jeremy Siegal, Strong Earnings Support Stocks Despite Softer Economic Data

The Bad

Higher interest rates, particularly in longer maturity Treasuries, are likely to stick around. There are three pressures on long term interest rates – inflation, growing supply, and a shrinking pool of foreign buyers. Inflation caused by higher oil prices and other costs of the Iran war may not abate quickly. The U.S. deficit is currently projected to be $2 trillion, driven by higher spending on defense, mandatory programs like Medicare and Social Security, and rising net interest costs on the national debt. Interest expense is actually the fastest growing line item in the Federal Budget. Lastly, foreign holdings of U.S. Treasury securities as a share of total marketable debt have fallen from roughly 34 percent in 2013 to approximately 23 percent in 2026.

What About Gold?

The big difference between this gold bull market and previous speculative runs is who’s buying. “Central banks may be accumulating gold because gold has no counterparty. A Treasury bond is somebody’s liability. A bank deposit is somebody’s liability. A currency is ultimately the liability of a sovereign monetary system. Gold is simply an asset sitting outside that chain of promises, and that distinction becomes much more important when governments are running massive deficits, geopolitical alliances are shifting and investors start questioning how much purchasing power their currencies will retain over time. Sovereign buying doesn’t guarantee that gold goes straight up, but it gives the market a structural bid that simply wasn’t present to the same degree in many earlier cycles.

The larger issue is that gold isn’t really trading like a commodity anymore. It’s increasingly trading like an alternative monetary asset. The underlying question isn’t whether jewelry demand rises a few percent or whether another mine opens in Nevada. It’s confidence—confidence in currencies, sovereign debt, central banks and the ability of governments to finance ever-larger promises without steadily depreciating the units in which those promises are denominated. That kind of confidence rarely collapses overnight. It erodes slowly, then suddenly. And once investors begin treating gold not as a trade but as monetary insurance, traditional valuation models start becoming much less useful.” Kerry Lutz, Financial Network Substack

Our take

We continue to believe in AI, but the financing needs of the companies that build AI are at risk of crowding out from the needs of the U.S. Treasury.  It is unclear how this will play out but ultimately, as we said in our blogs on AI, we think the baton will pass from companies that make AI possible to those who can use AI to make themselves more productive. Although we like to think our portfolios are magnificent, they are considerably less concentrated than the S&P 500. We own many AI infrastructure providers as well as companies we believe will benefit from using AI. Additionally, we maintain our exposure to gold and gold producers. Our fixed income exposure is short term, and we believe TIPS are increasingly attractive.