- The author is buying back into the equity drawdown: the semiconductor selloff is attributed mainly to positioning, with SNDK at 1,400s–1,500s and MU in the 950s as re-engagement levels.
- SOXL exposure is a separate quick trade, not a long-term thesis; the author would cut it if new local lows form. META’s continued AI spending is the longer-term bullish catalyst for semis.
- Crypto remains a cautious area: BTC looks stronger than attention-driven altcoin pumps. For PAPERTRADE, the author is waiting for LP below 2M; 50M of delta and a 2% BTC/ETH rise could trigger that reset.
Investment pitches
filter:
75 63 - The author sees little value in semis at current levels, preferring a larger dip or index breakout; re-entry markers are MU in the 950s and SNDK in the 1500s.
- Firmus’s failed $30B IPO attempt highlights neocloud valuation disparity: 900MW of contracted power versus NBIS’s stated 5B by end-2026 at a $60B valuation.
- SHAZ fell on shared-geography sentiment with Firmus, despite a $1.4B valuation; the post presents the contrast without declaring a trade.
- Crypto strength is narrowing to specific altcoins while majors and out-of-favor tokens weaken; DOGE fell over 10%, suggesting broad rallies may be ending.
67 - MU reported $54B revenue versus $51B expected and 86% gross margins; guidance points to tighter memory supply in 2027–28 than in 2026.
- Memory P/Es assume a fully cyclical business. If that assumption breaks, the author sees earnings growth plus multiple expansion supporting a potential 2–3x rerating.
- The muted earnings reaction suggests light semiconductor positioning; the author expects memory to reprice higher, though elevated yields and active ES futures selling remain a market headwind.
61 - Author expects a Treasury short squeeze: heavy short positioning and record-low gross exposure could push yields lower and fuel a QQQ breakout near all-time highs.
- Semis appear attractively valued because earnings have outpaced prices; SOXX is near highs, with MU’s 1 October earnings as the next pricing and guidance test.
- The author rotated some SOXL into TQQQ, expecting hyperscaler repricing and favoring 3× index exposure while cross-stock correlation remains low.
- PRL is an early-VC-style AI bet: mining costs exceeded market price through much of its history, yet inference cannot currently be bought with the token; upside rests on theme, roadmap, and execution.