AI is Eating the Index? Should You Add a Defensive ETF Sleeve?

AI stocks keep climbing.

Technology companies now represent almost 36% of the S&P 500. NVIDIA, Microsoft, Broadcom and Alphabet alone account for roughly one-fifth of the index.

And the technology-sector label does not even capture the entire AI trade. Alphabet and Meta are classified as communication-services companies. Amazon is classified as consumer discretionary. Other companies in sectors such as industrials are also benefiting from spending on data centres and AI infrastructure.

Even globally diversified all-equity ETFs are not immune. VEQT currently has approximately 21% of its equity exposure in technology stocks. You might be worried that you are too exposed to AI.

However, the practical question for your portfolio is not whether AI is a bubble or if the bubble is about to burst. Nobody knows. The better question is:

If you are uncomfortable with index concentration, can a defensive ETF sleeve reduce your risk without forcing you to abandon the market entirely?

A Useful Historical Analogy: Financials Before the Global Financial Crisis

Today’s market is not the same as the market of 2007. Still, the period before 2008 offers a useful portfolio-design case study.

By the end of 2006, financial companies represented approximately 22% of the S&P 500. Financials were the largest sector in the index. FDIC-insured institutions reported a sixth consecutive year of record earnings in 2006. Full-year profits increased 8.8%. Trading income climbed 36.6%, while income from investment-banking activities rose 20.1%.

At the time, those results looked like evidence of a highly profitable industry. In hindsight, the financial sector had also become a major source of risk.

Investors faced a question that feels familiar today:

Should you accept the index as it is, or add a defensive sleeve just in case?

Hindsight makes this question look easy but real investors do not receive that signal. A more useful test is to ask what happens when the investor acts early or waits until the risk is already obvious.

The Backtest

We compared four portfolios:

U.S.-listed ETFs were used because they have sufficiently long trading histories for the test. Results are shown in U.S. dollars to separate the performance of the investments from movements in the Canadian dollar.

This is not a forecast of how any ETF will perform during a future AI-related downturn. It is a historical stress test designed to answer a narrower question:

How useful was a defensive sleeve when the investor acted too early — or too late?

What Happened?

The table below compares each 80/20 sleeve portfolio against remaining fully invested in the S&P 500 through December 2009 based on the following approach:

• Initial investment: $100 on October 2005 (the earliest scenario date)

• SPY benchmark: grows according to the S&P 500 ETF from October 2005 to December 2009

• Sleeve portfolios: switch from 100% SPY to the 80:20 allocation at the scenario’s entry date

• After switching, the 80:20 portfolio is held without rebalancing until December 31, 2009.

• The final value is the sum of SPY and sleeve components on Dec 31, 2009.

The results reveal three useful lessons.

Lesson 1: Diversification Has a Cost Before It Has a Benefit

The difficult part of diversification is not identifying a risk after it becomes obvious. It is accepting the cost of protection while the dominant market theme is still working.

An investor who moved 20% of an S&P 500 portfolio into aggregate bonds two years before the October 2007 market peak would initially have lagged the index.

That underperformance was the price of caution.

But by December 2009, the same bond-sleeve portfolio is $4 ($107 vs $103) ahead of the all-S&P 500 portfolio. An investor who added the sleeve at the market peak received a larger benefit of roughly $10 ($113 vs $103).

The investor who waited until one year after the peak had a different experience. By October 2008, a substantial portion of the decline had already happened. The late investor missed much of the protection and then participated less fully in the 2009 equity-market rebound.

Risk management becomes less useful after the risk has already materialized.

Lesson 2: A Sector Sleeve Is Not the Same as Asset-Class Diversification

Consumer staples helped. But the effect was more modest than gold or bonds.

Consumer-staples companies may sell products people continue to buy during a recession. Their shares can still decline when investors sell equities broadly.

Bonds and gold serve a different role. They introduce return drivers outside the equity market.

During the Global Financial Crisis, gold produced the strongest result in this particular backtest. That does not make gold a guaranteed hedge against the next downturn. A future sell-off may have entirely different causes.

Lesson 3: Choose the Sleeve Based on the Risk

Not every concern about AI concentration calls for the same response.

If the concern is that a small group of mega-cap technology companies has become too dominant, an equity sleeve may be enough. A value ETF, small-cap ETF, equal-weight ETF or defensive-sector ETF can reduce reliance on the market leaders while keeping the portfolio fully invested in stocks.

If the concern is that an AI-driven sell-off could pull down the broader equity market, another stock ETF may not provide much protection. In that case, a bond ETF, cash-equivalent ETF or modest gold allocation may be more relevant.

Before adding another ETF, check whether your existing funds already own many of the same companies. A portfolio can contain several ticker symbols while remaining heavily exposed to the same market theme.

What About VEQT and XEQT?

A globally diversified all-equity ETF can still be a sensible starting point.

Funds such as VEQT and XEQT do not eliminate exposure to the largest U.S. technology companies. Nor should they. Those companies are an important part of the global equity market.

What the funds do is reduce the extent to which your portfolio depends on a single country, sector or handful of companies.

That may be enough for investors with a long time horizon and a high tolerance for equity-market volatility.

But global equity diversification should not be confused with asset-class diversification. VEQT and XEQT remain 100% equity portfolios. If stocks fall broadly together, owning more stocks will not provide the same protection as owning assets with different return drivers.

The Bottom Line

The rise of AI does not mean investors should abandon index funds.

It does mean investors should look beneath the headline number of holdings.

An index can own hundreds of companies while becoming increasingly dependent on a small number of winners. A defensive ETF sleeve can reduce that dependence, but every sleeve has a cost. If the dominant theme continues to rise, the diversified portfolio may lag.

The 2007–2009 backtest offers a useful lesson. Diversification did not require perfect timing to help. But the type of diversification mattered, and waiting until the crisis was already obvious reduced the benefit.

A consumer-staples sleeve changed the composition of the equity portfolio. Bonds and gold introduced more distinct sources of diversification.

The question is not whether you can predict the end of the AI boom.

It is whether your portfolio is resilient enough that you do not need to.

Sources:

Financial Sector Performance Pre-GFC – FDIC Quarterly Banking Profile Q4 2006. Reports 8.8% increase in full-year profits, trading income +36.6%, investment-banking income +20.1%. FDIC QBP Dec 2006

Historical S&P 500 Financials Weight – Bespoke report on sector weights. Financials represented 22.27% of S&P 500 at the end of 2006, falling to 8.58% by March 2009 low. Bespoke S&P 500 sector weights