Leveraged ETFs have been around for years, using derivatives to deliver two or three times the daily return of a stock or index. The big shift is how much money is now piling into the same handful of names.
A Bloomberg analysis of roughly 800 bullish leveraged equity ETFs shows the share tied to AI companies has climbed from 26% in 2022 to 58% in mid-2026, and the concentration in a handful of names is now large enough to move markets.
How the money is concentrated
Total global assets in leveraged ETFs sit at roughly $250B, small relative to the $22T in all ETFs worldwide. But their share of daily trading is larger than their share of assets. At a recent peak, leveraged ETF trading volume hit roughly $70B per day on a 30-day average basis.
Just 10 companies account for two-thirds of AI-related leveraged exposure. Memory chipmakers dominate.
SK Hynix and Samsung Electronics together attracted more than $15B in net inflows to leveraged single-stock ETFs since mid-2025. Micron Technology and Sandisk add another $4.5B between them. Nvidia and Broadcom anchor the AI hardware bucket.
The mechanism driving risk is straightforward. When a fund targets two times a stock's daily return, it must buy more exposure after the stock rises and sell after it falls.
Somewhat counterintuitively, inverse funds can also be forced to trade in the same direction as their underlying stock after large moves. Both effects concentrate activity in the final minutes of the session, alongside index rebalances and options hedging.
South Korea showed what that looks like
South Korea's financial regulator approved more than a dozen new leveraged ETFs tracking Samsung and SK Hynix earlier this year. The domestic funds attracted more than $9B in two months.
At the height of the volatility that followed, combined trading in those leveraged funds and their two underlying stocks made up 70% of all daily activity in Seoul.
"The scenario that worries me most is a sudden event very close to the close of trading, whether at a company or market level, that would force end-of-day rebalancing activity to happen in a very short period of time."
Rocky Fishman, Asym Research
The Korea episode was not purely an ETF story. Chip prices, earnings, and borrowing costs all played a role in the broader AI stock wobble.
But by layering heavy leveraged exposure onto two stocks that already dominated the benchmark index, the funds amplified every swing in both directions.
The math erodes returns over time
The daily reset mechanic is a structural drag on long-term holders. Because gains and losses compound daily at a multiple, volatility chips away at a fund's value even when the underlying stock ends flat.
A stock that doubles over a year while swinging sharply along the way could leave a 2x leveraged fund down 11.1%, not up 200%.
SK Hynix has already logged multiple double-digit single-day price swings this year. That volatility tax is not hypothetical for anyone holding a leveraged product on the stock.
Despite this, holding periods for bullish ETFs have grown well beyond the recommended single day, especially for single-stock products.
The general upward trend of AI stocks has masked the fee drag and decay for investors who have been on the right side of the trade.
Closures are rising alongside launches
A total of 73 leveraged and inverse ETFs closed in 2026 through late July, versus 22 closures in all of 2025. That represents roughly 43% of all US-listed ETF closures this year. Some closures stem from funds that never attracted enough assets. Others are more abrupt.
The GraniteShares 2x Long LCID Daily ETF plunged roughly 51% on a single day in July, wiping out its net asset value and triggering a delisting from the Nasdaq.
Meanwhile, the SEC opened a new comment period on ETF innovation and novel investment strategies on June 30.
ETF experts say the regulator faces a market that won't slow itself, and that the next external market event will expose the limits of how much leverage the system can absorb.