I’m undertaking a 1000-day reinvention project, blogging here daily to track my progress. In Wednesday Wealth, I write about money management, retirement, estate, and long-term care planning.
In So, Why Don’t You Own It? Understanding Underweights in Equity Investing, strategists from Allianz Bernstein (AB) write:
Today’s equity markets are arguably the most concentrated, interconnected and exposed to correlated risks in the modern era. In this fragile environment, we believe investors need more than just exposure to stocks that have driven recent market returns. Disciplined stock selection and clear risk objectives are essential—as well as conviction in what not to own.
AB says you need to deliberately choose where to underweight your portfolio relative to market weights, because if you just by default take on market weights you may be opening yourself up to significant risk of correction or crash.
Because today’s equity markets are extremely concentrated in the AI buildout right now, investors may find themselves at risk in case of any wobble in this sector. The AB strategists’ research found that the 10 largest contributors to US equity risk have seen their share of the S&P500 benchmark more than double since the global financial crisis.
Simply diversifying geographically or into small caps won’t necessarily help you, because some global indices such as MSCI Emerging Markets and MSCI World Growth and small cap indices such as Russell 1000 Growth as well feature outsized exposure to large stocks with values dependent on AI buildout spending.
While AB acknowledges that the megacap tech companies are “strong businesses with durable competitive advantages,” there is good reason for concern that the tremendous spending on the AI buildout will not translate into attractive long-term returns. That means investors need to deliberately assess what place megacaps and marquee IPOs such as Anthropic or OpenAI should have in their portfolios. AB says “FOMO isn’t a substitute for an investment process.”
They cite the analog to the dot-com bubble:
Today’s technology leaders aren’t dot-com replicas. Many have healthier businesses, deep competitive moats and strong operating cash flows.
But there are parallels. In the late 1990s, circular investment helped inflate fragile business models. Lucent Technologies, for example, extended billions of dollars in vendor financing to smaller internet and networking companies; when the bubble burst, uncollectible loans contributed to its demise, destroying $250 billion of shareholder value.
Today, a similar concern is emerging as chipmakers, cloud providers and AI model developers are increasingly each other’s investors, suppliers and customers. This raises the risk that capital circulating within the AI ecosystem is artificially inflating reported demand.
Even if the AI cycle is built on stronger foundations than the dot-com era, potential vulnerabilities deserve attention. Hyperscaler capex could outstrip cash flows, and we believe the underperformance of this group so far in 2026 may signal a shifting investor focus toward free cash flow, returns on capital and financial resilience. Meanwhile, imminent IPOs from currently unprofitable tech firms (notably OpenAI and Anthropic) offer another echo of the dot-com era. Technical market drivers may be changing too, as stretched cash flows limit share buybacks.
AB suggests that “passive ownership doesn’t eliminate risk; it repackages it by embedding yesterday’s winners more deeply into portfolios.” So should you underweight them? They write:
Underweight positions in large, outperforming stocks can contribute to weak relative returns in a concentrated market, but they are a calculated, strategic trade-off aimed at preserving a portfolio’s risk profile. We believe real discipline requires knowing what a portfolio owns, what it avoids and why those decisions support the portfolio’s long-term objectives.
My plans for managing this risk
One thing I’ve been consistently surprised by is how important it is to know exactly what a particular ETF holds. You can’t just, for example, buy an emerging markets passive ETF and get true diversification from the S&P 500. Consider the iShares MSCI Emerging Markets ETF (EEM). It is heavily overweight big tech companies riding AI euphoria:
And, more important than that, right now is a good time to reduce any market-cap-weighted ETFs in favor of value, commodities, real estate, and precious metals (also a kind of commodity, I should say).
My newly developed Flow Allocation Model will leave me underweight the megacap tech companies, and will have me reducing my current exposure to the AI buildout trade.
Here’s what my asset allocation will look like once I’ve rebalanced (noting that any sleeves with their preferred ETFs below the 200-day simple moving average will actually be moved to cash in weekly 25% tranches):
The AI-concentrated portions of this include U.S. large-cap momentum, U.S. small-cap momentum, and Emerging market equities: just 15% of the entire portfolio. Compare that to 40% if you are only in the S&P 500, or almost 50% for the Nasdaq 100.
I’ll be gradually moving to this new allocation over four weeks, with the help of a web application I’ve built to guide my trades each week.
It feels good, but a little scary, to be taking this contrarian approach.