2025-12-11 · 供应链分析

甲骨文因OpenAI信用风险大跌，新云板块错杀，建议买入非OpenAI依赖标的。

涉及标的： $CIFR $CRWV $GOOGL $IREN $MSFT $NBIS $ORCL $WULF

中文翻译

甲骨文($ORCL)的财报结果及其对$NBIS和$IREN等“新云(Neocloud)”板块的影响：甲骨文财报EPS超预期且积压订单创纪录，但盘后仍下跌12%。甲骨文自9月11日高点以来已下跌39.8%，并拖累了整个板块。原因如下：抛售不仅是对营收轻微不及预期的反应，更是算法做空和投资者对AI资本支出(Capex)周期可持续性，以及该板块主要租户OpenAI偿债能力的担忧。甲骨文宣布2026年资本支出增加150亿美元至近500亿美元，这与据报道的与OpenAI的3000亿美元合作伙伴关系密不可分。最初，OpenAI作为前沿大语言模型(LLM)，向甲骨文、Coreweave等承诺了诱人的资本支出，推动了板块的初步重估。然而，随着超过1万亿美元的债务义务以及来自Anthropic、Gemini、XAI等的竞争加剧，市场严重怀疑甲骨文、Coreweave等是否为一家无法通过经营性现金流履行义务的租户建设基础设施。市场有效发出了信号：甲骨文正在为无法履行承诺的OpenAI提供不可持续的债务融资“供应商融资(Vendor Financing)”。因此，下跌是理性的：抛售由信用风险和资本密集度的理性重估驱动。 OpenAI资金担忧是合理的：OpenAI缺乏资金履行合同的假设，由其收入(130亿美元)与义务(600亿美元/年)之间的巨大错配所支持。信用担忧是真实的：甲骨文信用违约互换(CDS)利差的扩大表明“信用事件”降级或违约的概率上升。此外，我们看到这在“新云”板块中引发了传染效应：$NBIS从140美元跌至90美元区间，$IREN从80美元跌至40美元区间，$CIFR从20多美元跌至17美元区间。但这对于$WULF、$NBIS、$IREN等新云公司是否是买入机会？是的。这对$ORCL是否是好的买入机会？不是。前瞻展望： $ORCL(大部分)和$CRWV(25%积压订单)是两大主要依赖OpenAI的玩家，这一叙事可能因OpenAI的融资活动而瞬间反转(+30%以上波动)。如果OpenAI在2026年以高估值进行超额认购的IPO，且其新GPT模型击败Gemini/Claude，情况可能改变。然而，许多其他玩家与OpenAI隔离。新云板块的原始逻辑是Mag7从其现金牛业务(Azure, AWS, GCP)向下漏斗资金至：$NBIS, $IREN, $CIFR, $WULF等。但随着最大玩家($ORCL, $CRWV)下跌，算法交易拖累了整个板块。如果单独看公司，$CIFR和$WULF由$GOOGL背书，$IREN/$NBIS由$MSFT资助。这些是来自超大规模云厂商/Mag7的锁定合同积压，而非OpenAI。这种因误解风险而导致的非理性抛售，为新云板块提供了绝佳的买入机会，但不包括与OpenAI绑定的$ORCL和$CRWV。

英文原文

Oracle [ $ORCL ] earning results and its effect on the neocloud sector like $NBIS & $IREN: Oracle reported earnings with a beat on EPS and a record backlog but, dropped 12% after hours. Oracle is down 39.8% from September 11th highs and brought down the sector with it. Here's why: The sell-off was not merely a reaction to marginal revenue miss, but both an algorithmic short and investor selloff on the sustainability of the AI capex cycle and the creditworthiness of the sector's primary tenant: OpenAI. Oracle's announcement of a $15 billion increase in 2026 capital spending to nearly $50 billion was inextricably linked to a reported $300 billion partnership with OpenAI. Originally, OpenAI was the frontier LLM, with promising capex promises to Oracle, Coreweave and others, contributing to the initial repricing of the sector. However, with over $1t+ in obligations and increasing competition from Anthropic, Gemini, XAI, and others, the markets have serious doubts on whether Oracle, Coreweave, and others are building for a tenant that cannot currently fund its obligations from operating cash flow. WE're seeing the market effectively signaling that the market Oracle is creating an unsustainable debt-funded "vendor financing" for OpenAI, which cannot fulfill its promises. So, the drop was rational: The sell-off was driven by a rational repricing of credit risk and capital intensity. OpenAI Funding Fear is Valid: The hypothesis that OpenAI lacks the funds to honor its contracts is supported by a glaring mismatch between its revenue ($13B) and its obligations ($60B/year). Credit Fears are Real: The widening of Oracle's CDS spreads sees a rising probability of a "credit event" downgrade or default. Furthermore, we're seeing this trigger a contagion effect across the "Neocloud" sector from $NBIS dropping from $140 to $90s, $IREN dropping $80 to $40s, $CIFR dropping from $24s to $17s. But is this a buying opportunity for Neoclouds like $WULF, $NBIS, $IREN, and others? Yes. Is this a good buying opportunity for $ORCL? No. Forward Outlook: $ORCL (large portion) , $CRWV (25% backlog) are the two players largely dependent on OpenAI and this narrative can flip in an instant (+30%+ change) depending on capital raising activity from OpenAI. If OpenAI files for an oversubscribed IPO in 2026 at high valuations and it's new GPT models beats out Gemini/Claude, we can see this change. However, many other players are isolated from OpenAI. The original thesis of the Neocloud sector was Mag7 capex funndel from their cash cows segments (Azure, AWS, GCP) down into: $NBIS, $IREN, $CIFR, $WULF, and others. But as the largest players ( $ORCL, $CRWV) fall, these algorithmically bring down the whole sector. If you look at the companies individually, companies like $CIFR and $WULF are being backstopped by $GOOGL, and $IREN / $NBIS are funded by $MSFT. These are locked in contract backlogs from Hyperscalers/Mag7, not OpenAI. This irrational selloff due to misunderstanding of risks presents an amazing buying opportunity for the Necoloud sector, but not companies tied to OpenAI like $ORCL and $CRWV.

在 X 上查看原推 ↗