· 供应链分析

板块错杀提供买入机会,NBIS/IREN无交易对手风险且利用率高。

涉及标的:

中文翻译

超额收益(alpha)在于了解哪些类型的合同能让人免受当前恐惧的影响,以及行业内哪些个别组件存在定价错误。$MSFT、$META 等公司的算力容量合同是“照付不议”(take or pay)的。 $NBIS、$IREN 等公司由于与“七巨头”(Mag7)(拥有无限资产负债表)签订5年期合同,实际上消除了**交易对手风险**(counterparty risk)。 然而,它们下跌的原因不同($NBIS 因2500万股的自动行使机制(ATM)发行;$IREN 因市场担忧其为将剩余3GW的AI云容量管道变现而进行稀释,而非用于数据中心托管(colocation))。 然而,主要的普遍担忧是 $ORCL、$CRWV 面临来自 OpenAI 破产+无力支付的严重交易对手风险,但由于它们是两大主要玩家,这引发了整个板块的算法抛售。 但这种板块抛售为一些未受影响(除信贷收紧外)/被误解的公司提供了良好的买入机会。 此外,AI是一个增长的市场,目前可能存在算力紧张(例如 anthropic、gemini 和其他模型),但 $NBIS、$IREN 在超大规模云服务商合同上的利用率基本为100%。 显然,如果存在过度建设,可能会导致利润率压缩,但我们目前尚未看到这种情况。5年后会发生什么,我不知道。

英文原文

The alpha is knowing what types of contracts lead to isolation from current fears and where there's mispricing on individual components in the sector. Compute capacity contracts are take or pay for $MSFT, $META, and others. Companies like $NBIS, $IREN and others are effectively de-risked due to **no counterparty risk** from Mag7 (infinite balance sheets) and 5 Year contracts. However, their drops were different reasons ( $NBIS, 25M share ATM), $IREN fears over dilution for monetizing their rest of the 3 GW capacity pipeline for AI Cloud instead of colo. However, the main overarching fear was that $ORCL, $CRWV faces severe counterparty risk from OpenAI insolvency + inability to pay, but this causes a sector algorithmic selloff because they're the two largest players. But this sector sell-off is a good buying opportunity for some of the unaffected (minus credit tightening) / misunderstood companies. Also AI is a growing market, and there's likely compute strain for the time being (eg. anthropic, gemini, and other models) but it's basically 100% utilization for hyperscaler contracts on $NBIS, $IREN. Obviously if there's overbuild, there's probably margin compression but we're not seeing that right now. What happens after 5 years I don't know.

在 X 上查看原推 ↗