· 供应链分析
涉及标的： $CIFR$CRWV$IREN$MSFT$NBIS$ORCL$SLNH$WULF
$IREN 在加密货币挖矿方面拥有良好的利润率和高吉瓦(GW)容量，但这并不能直接转化为高性能计算(HPC)优势。 与此同时，他们正采取 $CIFR 和 $WULF 的路径，通过中间件编排将业务导向超大规模云服务商，但这会在长期内大幅压缩利润率。 $NBIS 直接与 $MSFT 进行全栈合作，实际上可以随着时间的推移提高其毛利率。 大家都指向 $CRWV 作为小型转型的案例，但最近我们学到的残酷现实是，如果许多矿工试图建立自己的编排和堆栈，他们很可能会失败。 $ORCL 是软件护城河的完美例子：如果一个市值 8000 亿美元的超大规模云服务商都感到困难，那么一个市值 40 亿美元的比特币矿工，或者像 $SLNH 这样市值 1 亿美元的微盘股，又该如何转型呢？ 如果他们选择直接转型，许多人在估算比特币矿工基本面时参考 $CRWV 将吉瓦(GW)转化为收入/利润的方式，但实际回报在执行过程中可能会令人失望。 Nebius 从一开始就是第一名，甚至在 GPU 利用率上超越了 $CRWV（根据其最新的白皮书），其全栈带来的长期盈利能力将在财报中胜过投机。
$IREN has good margins for crypto mining and large GW capacity but that doesn’t translate directly into HPC. They’re going the $CIFR $WULF route in the meantime with middleware orchestration funneling to hyper scalers but that cuts into margins a lot in the long run. $NBIS is direct to $MSFT full stack and can actually increase their gross margins over time. Everyone points to $CRWV as the example to minor pivots but the harsh reality as we learned recently is that if many miners try building their own orchestration and stack, they will likely fail. $ORCL is the perfect example of the software moat where if a $800B hyperscaler has trouble how would a $4B bitcoin miner pivot, or a $100m microcap like $SLNH do it. If they go direct, a lot of people estimating bitcoin miner fundamentals use $CRWV in terms of translating GW to revenue/profit, but the actual returns will likely be underwhelming during the execution. Nebius is #1 from the very start, even beating out $CRWV (from their most recent whitepaper with gpu utilization) and long term profitability from full stack will shine when it comes to earnings vs speculation.