The Invesco QQQ Trust (NASDAQ: QQQ) has climbed 10% over the past month and 19% year to date, trading near $727 after rebounding from early-April volatility that briefly pushed the VIX above 25.
That rally has been driven almost entirely by a handful of AI infrastructure names that dominate the fund’s index weight, raising serious questions about underlying risk.
The fund’s top five positions account for roughly 37% of net assets, with NVIDIA (NASDAQ: NVDA) alone carrying approximately a 10% weighting in the portfolio.
With a 0.20% expense ratio and a Nasdaq-100 wrapper, QQQ’s performance over the next 12 months will be determined far less by fund mechanics than by the trajectory of a small cluster of holdings.
The federal funds upper bound currently sits at 3.75%, and the 10-year Treasury yield has held around 4.56% for much of the year, but the true swing factor is whether hyperscaler capital expenditure budgets continue rising through 2026.
Microsoft spent $30.88 billion on capital expenditures in a single quarter, up 84% year over year, while Alphabet ran capex up 107%, Amazon reached $44.2 billion, and Meta Platforms raised its 2026 guidance to between $125 billion and $145 billion.
That spending flows directly into NVIDIA’s revenue, with the company’s Data Center segment growing 92% last quarter, and these five names together represent over 29% of QQQ’s total weight.
A critical early warning signal is GPU rental pricing, with one widely-read post this week on Reddit’s stock market community titled “GPU Rental Prices Notably Decline through 2nd half of May; H200 -38%,” pointing to supply beginning to catch demand.
Historically, falling rental prices precede what analysts describe as capex digestion, meaning that if two of the four major hyperscalers trim their 2026 spending guidance in the same reporting cycle, QQQ would absorb the bulk of any resulting selloff.
Investors should monitor Microsoft’s commercial remaining performance obligation, currently at $627 billion and up 99%, alongside Google Cloud’s backlog of $460 billion, with the next round of hyperscaler earnings due in late July.
QQQ’s deeper structural exposure stems from the fact that its top holdings are not independent positions, as NVIDIA sells chips to Microsoft, Amazon, Alphabet, and Meta, which in turn generate AI services revenue that funds further chip purchases.
A deceleration in hyperscaler capital spending would simultaneously compress NVIDIA’s growth rate and threaten its roughly 75% gross margin, hitting the fund’s single largest holding from multiple directions at once.
NVIDIA’s approximately 10% weighting is approaching levels that historically triggered the Nasdaq-100’s special rebalance mechanism, which capped a single stock’s index contribution during a similar episode in 2023.
If that methodology forces a trim, QQQ investors would receive reduced NVIDIA exposure automatically, regardless of how the stock itself performs, with Invesco publishing daily holdings on its fund page.
Investors seeking the same AI thesis with less concentration can access equal-weight Nasdaq products, which dilute top names to roughly 1% positions each, though those vehicles historically underperform when NVIDIA leads the market.
The two events most likely to determine whether QQQ’s 43% one-year run extends or stalls are the July hyperscaler earnings round and the next Nasdaq-100 special rebalance, both of which carry more weight than any Federal Reserve decision.
