A Redditor shared the kind of story that makes everyone reach for the calculator. A friend put roughly $1 million into Direxion Daily Semiconductor Bull 3X Shares (NYSEARCA:SOXL), watched the position bleed down to about $200,000 during the 2025 tariff-war selloff, refused to sell, and then cashed out near $7.5 million when the AI chip trade ripped back to life this spring. Seven and a half times your money, with a brutal 80% drawdown in the middle, in a fund that the prospectus itself warns about holding. That outcome hides a lot of mechanism behind one clean number.
The arithmetic that produced a 7.5x
Start with the round-trip on the underlying. SOXL is at $234, up 971% in one year, and it accelerated late. April 2026 alone delivered a 165% monthly gain, the best single month since SOXL’s inception, while the underlying PHLX Semiconductor Index rose 38%, a monthly move it had not produced since February 2000. A position that opened 2025 around the high $20s, drew down with the rest of the sector through a tariff-driven semiconductor selloff, and survived long enough to ride the spring 2026 recovery, lines up with the Redditor’s $1 million to $200,000 to $7.5 million arc. The shares never got cheaper than a tactical trader could stomach, but the dollar value of a paper position can collapse fast when 3x’ed.
The headline number assumes a holder who watched a 7-figure account fall by four-fifths and did not flinch, did not add, did not trim, did not get tapped on the shoulder by a spouse or a margin clerk. Real survivors of an 80% drawdown in a 3x fund are vanishingly rare, which is part of why the screenshot travels.
What actually did the work
SOXL is a 3x long bet on a basket of semiconductor names like NVIDIA (NASDAQ:NVDA | NVDA Price Prediction), Broadcom (NASDAQ:AVGO), AMD (NASDAQ:AMD), Taiwan Semiconductor (NYSE:TSM), Qualcomm (NASDAQ:QCOM), and Micron (NASDAQ:MU). The recent surge has been due to renewed confidence in AI infrastructure spending, an OpenAI data center lease, and hyperscaler capital expenditure commitments. Multiply a directional sector index by three, every day, with no path-dependent friction during a sustained uptrend, and you get the kind of curve that looks fake on a screenshot.
The same mechanism cuts the other way. When tariff headlines tore through semiconductors in 2025, SOXL spent the first half of that year sliding, with a January 2 to June 30, 2025 move from $28 to $25 understating the violence in between, because the daily reset compounds losses on red days the same way it compounds gains on green ones. A 33% drawdown in the SOX index over a few months becomes a 70%-plus drawdown in the 3x wrapper, and that is before volatility decay, the slow tax a leveraged fund pays when the chart chops sideways. Coverage from late April pointed this out explicitly, noting SOXL is a tactical instrument for active traders due to volatility decay and symmetric downside, and unsuitable for buy-and-hold investors. The Redditor’s friend did, in a sense, got lucky.
What to watch from here
The Redditor’s friend quit near the top of the move. SOXL has already given some of it back. One piece from late May framed it bluntly, noting that the setup is no longer as “cheap” as it once was.
Watch the things that drive the underlying chart. Hyperscaler capex guidance from the cloud majors, because the AI build-out is the entire fundamental story. NVIDIA and Broadcom forward commentary on data center order books. Tariff posture toward Taiwan and mainland China, which is what produced the 2025 air pocket in the first place. And the VIX, which is the heart rate of any 3x position you own.
There’s medium confidence and a bullish lean, the AI capex cycle remains intact, and the same fund that paid a brave (or stuck) holder 7.5x can take it back faster than it gave it. The $20 SOXL of the 2025 lows is not coming back without another tariff shock or a credible threat to the data center capex narrative. Plus, a $1 million bet at today’s $241 is a fundamentally different trade from a $1 million bet at $30.