🔥 Buy the Quiet, Sell the Headlines? A Beginner’s Guide to USO, Oklo & Vistra
🧠 The Big Idea: “Buy When It’s Quiet, Sell When War News Returns” A popular trading idea is simple: buy energy stocks when geopolitical tensions appear to be cooling, then sell when war headlines return. But does this strategy actually work? Looking at USO, Oklo and Vistra, the answer is only partly yes. The strategy has historically worked much better for oil prices and USO than for nuclear and power stocks such as Oklo and Vistra. The biggest lesson for investors is that not every energy stock reacts to war headlines in the same way. Understanding what actually drives each company is more important than simply trading the headline. ⸻ 📊 The Recent Trade: What Happened? The trading window around August 25–31 provides a useful example. Stock Aug 25 Aug 28 Aug 31 🛢️ USO $126.15 (-4.6%) $129.
上周我说VST/CEG过去两周跑赢XLU的7个百分点是地缘溢价要回吐。今天甲骨文2.8GW燃料电池订单出来,这个判断必须拆开——地缘溢价该回吐的还在回吐,但AI电力基本面溢价被一笔单子重新托住了。 2.8GW是什么概念:相当于2.5个核电机组,相当于旧金山全年用电的15%。单笔订单历史无先例。甲骨文选燃料电池不选电网扩容,是因为9-12个月通电 vs 3-5年扩容。它愿意为“快”付溢价,等于承认:在AI数据中心的成本模型里,推迟一年拿电=推迟一年确认AI收入,这个代价比燃料电池溢价大得多。 真正的信号:AI数据中心的瓶颈从芯片切换到电力。 产业链要分三层定价: Bloom Energy——地缘溢价0%,AI基本面100%。甲骨文一单锁死两年产能。最纯AI电力敞口,但PE 40倍脆弱。 VST/CEG——地缘溢价3-4%该吐(已经吐了一半),AI溢价3-4%由基本面托住。向下空间比我上周判断的小。 ORCL——12%涨幅不是订单利润,是“供电问题被移除”的重估。当前对FY2027E PE 26x,嵌入“AI RPO转化+供电问题分阶段解决”。到$340的下一档需要剩余75%电力需求再落实40%+Q4毛利率不恶化。 七天观察: ① 谷歌大会会不会跟进燃料电池合作(验证结构性趋势) ② Bloom Q2产能利用率(超85%=涨价权) ③ 台积电Q2对CoWoS的表述(如果CoWoS扩产但电力约束仍紧=2026H2限速阀是电) 争议预判:如果瓶颈真从芯片切到电力,英伟达PE 23x、博通PE 16x、Bloom PE 40x这个估值分布就是错的——电力比芯片贵。要么市场已经领先定价了转换,要么电力是泡沫。我倾向前者但不足以满仓。 $甲骨文(ORCL)$