Oracle faces double jeopardy, viewed both as a legacy software business and as an overlevered infrastructure business that depends on OpenAI being good for its capex commitments. OpenAI is already so aggressively valued that even catching Anthropic in private valuation leaves limited upside, which also limits how much SoftBank, which @goodalexander calls leveraged long OpenAI, could benefit. x.com 

Oracle's year-long evaluation of Cerebras' costs, performance, customer uses and equipment life strengthens the case for its inference technology. Oracle chose to bring the equipment into its data centers; that real-world buyer decision conflicts with the short seller report's criticism. x.com 

Oracle should in theory benefit more than anyone if data-center returns come in high, because it applied the most leverage to the buildout. The stock is down 55% in a year despite continued earnings growth and accelerating revenue growth led by cloud. The software business is stagnating as the company pours its resources into data centers, and the continued selloff has been a surprise. x.com 

Oracle offers no reason to buy over the other hyperscalers. It carries a ton of debt and a weaker business model than Alphabet or Microsoft, meaning much more risk for similar potential reward. x.com 

Oracle looks set to burn many more billions in cash at least into the first half of fiscal 2028. That is the less reassuring implication of Reuters' headline that Oracle expects cash flow to turn positive in the second half of that fiscal year, even though the market liked the news. x.com 

Oracle now looks bullish given strong upside momentum and accelerating sentiment, reversing an earlier refusal to buy the dip in it. Recent tightening in credit default swaps and neocloud GPU deals help the case. Oracle's planned provider-managed OCI-to-AWS networking layer, starting later in 2026 in us-east-1, could reduce deployment friction and deepen Oracle-AWS multicloud economics, but it stays financially indirect until rollout and customer adoption are proven. x.com  x.com 

Oracle's big rally on its RPO jump was fake, and the only data since show OpenAI writing checks its business can't cash. The pop came on an earnings miss and took 57 days to undo; the stock then fell another 46% and has dropped after every report since. Within two weeks, a force majeure dismissed as a nothing burger gave way to word that its cash-incinerating key customer has $20B less revenue than due diligence assumed. Shorting in size is still hard. x.com  x.com  x.com 

Oracle is being repriced from a high-margin, low-capex software firm into a capital-heavy cloud infrastructure builder, and the market doubts shareholders will capture the profits. Q4 revenue of $19.18 billion, EPS of $2.11 versus $1.96 expected and $638 billion of RPO all beat, yet shares fell on plans to raise nearly $40 billion of debt and equity in FY2027, including a $20 billion ATM. Mid-chain, Oracle pays Nvidia and Broadcom first while bearing financing and execution risk. x.com 

Oracle could get more room to bring capacity online if separate financing covers hardware costs before cloud revenue flows. Per WSJ via @KobeissiLetter, Broadcom is seeking more than $50 billion in financing, with lenders such as Apollo and Blackstone, for the OpenAI custom chip the two are developing, intended to help Oracle bridge the gap between paying for hardware and collecting cloud revenue. That would help Oracle monetize demand. x.com 

Oracle looks like the Lehman of the hyperscalers in a subprime-style debt bust around the AI data-center build-out funded by private credit, private equity and bank lending. Per HedgieMarkets, Oracle sent a force majeure notice on Project Jupiter, its New Mexico data center for OpenAI's Stargate, after the state twice denied permits for a gas pipeline the site needs. Oracle wants the right to delay payments if it misses its 2028 opening, and $18 billion of project loans trade at 89 to 91 cents. … Show more