Summary: The episode’s central lesson is simple: a great long-term AI thesis does not guarantee a good short-term trade. The week ending October 2, 2026 showed three very different outcomes across CoreWeave, Nebius, and IREN. CoreWeave (CRWV) — Reclaim: After previously failing around $88–$89, CoreWeave closed back above that zone at about $89.62 and above its 20- and 50-day EMAs. Strong demand, pricing power, backlog growth, and analyst optimism supported the move, while convertibles, potential share issuance, and insider selling remained major risks. The key takeaway was that buyers appeared willing to absorb those concerns once price reclaimed the critical level. Nebius (NBIS) — Continuation: Nebius successfully defended its prior breakout around $230 despite dipping below it intraday. It closed the week around $242.81 and remained above its major moving averages. Rising GPU prices and infrastructure demand supported the bullish thesis, but enormous capital expenditures and convertible debt highlight how aggressively the company is spending to capture future AI demand. The technical lesson: breakouts are not just confirmed at resistance—they are proven when former resistance holds as support. IREN — Invalidation: IREN produced arguably the strongest fundamental headlines, including large Microsoft- and Nvidia-related contracts, rapid AI-cloud growth, and a major analyst upgrade. But the stock broke below its $42–$43 support area, closed around $41.76, and fell below its major EMAs. Concerns about massive future capital spending, potential dilution, losses, and short interest outweighed the good news in the short term. The long-term business thesis may still succeed, but the prior long trade setup failed. The broader issue across all three companies is AI demand versus financing risk. Compute remains scarce and companies have demonstrated strong pricing power, but building the infrastructure requires extraordinary amounts of capital. With borrowing costs elevated, investors are closely scrutinizing debt, convertible securities, equity issuance, and dilution. The episode’s main takeaway is: Trade the levels, not the hype. You can be completely right that AI infrastructure will be enormous in 2030 and still lose money by entering a stock at the wrong price in 2026. CRWV demonstrated a reclaim, NBIS demonstrated continuation, and IREN demonstrated invalidation. Price action provides a way to distinguish the long-term corporate story from the short-term trade.
