The question sitting at the center of institutional investment committee rooms this week is not whether AI infrastructure spending is real. It is whether the companies funding that buildout with debt can survive long enough to collect. Oracle’s investor day on Wednesday, October 28, is where that argument gets stress-tested in public.
Two Markets, Opposite Conclusions
Equity analysts are broadly constructive on ORCL. Oracle has flagged remaining performance obligations north of $500 billion, with growing demand from major customers. Analyst sentiment holds a “Moderate Buy” consensus with a street target around $252. The cloud business delivered 121% IaaS growth in the most recent quarter.
The bond market is not buying it. Oracle’s balance sheet has gotten considerably heavier, with total debt now sitting at approximately $167 billion, having nearly doubled in recent years. The CDS spread range is more in line with BB and B credit, high-yield territory, and the five-year spread has been hovering around 295-300 basis points. Oracle’s S&P rating is BBB-, the lowest rung of investment grade.
CoreWeave Makes the Bear Case Concrete
If anyone doubted that credit markets were running a separate scoring system on AI infrastructure, CoreWeave has settled the argument. The yield on its $2.75 billion bond maturing in 2031 has been around the mid-15% range this week. The stock dropped to the mid-$50s, down roughly 70% from its all-time high. The spread on CoreWeave’s five-year credit default swaps has at times traded above 1,000 basis points.
The business model that bothers bondholders is straightforward: CoreWeave borrows money to buy chips and build data centers, rents that capacity to AI developers on multi-year contracts, and uses the rent to repay the debt. At close to 15% yield, every new data center has to work much harder just to cover its interest bill. Oracle is a more complex company, but its creditors are applying the same lens.
Barclays Turns, But With Conditions
The analyst who called Oracle’s credit deterioration earliest is now the most important voice in the room going into Wednesday. Almost exactly a year ago, Barclays’ IG TMT credit analyst Andrew Keches slapped an Underweight on Oracle’s debt, told clients to buy the five-year CDS at roughly 80 basis points, and warned Oracle could run out of cash by the November 2026 quarter. In a note published this week, Keches and Remy Neuhaus turned tactically bullish, recommending clients buy Oracle’s lowest-dollar long bonds into next week’s investor day. The bank is turning tactically constructive after one of the most punishing selloffs in the investment-grade technology complex, but is far from declaring the all-clear.
The distinction between “tactically bullish on the bonds” and “the credit is fine” matters. Barclays is arguing that the selloff has overshot, not that the leverage is comfortable.
The gap between equity and credit readings is not unique to Oracle. Investment-grade bond spreads have been widening across the technology sector as record AI-driven issuance competes for the same wallets.
What Wednesday Actually Has to Deliver
The equity market wants revenue guidance and backlog conversion timelines. The credit market wants something narrower: a credible path to positive free cash flow. Free cash flow was negative $23.7 billion for fiscal year 2026 as Oracle continued to execute on investments to support its Cloud Infrastructure business. The deciding detail investors are waiting for is a named fiscal year for positive free cash flow.
There is one potential wildcard. TD Cowen analyst Michael Elias wrote on October 20 that industry checks suggest Nvidia may be providing credit support for some of Oracle’s data center leases. Confirmed supplier financing would change the leverage math materially. Unconfirmed, it is noise.
Stocks to Watch
ORCL: Oracle closed at $126.16 on October 23, down more than 50% from its peak. Wednesday is the first opportunity management has had to speak directly to credit market concerns since the CDS widening accelerated. A free cash flow timeline moves the stock; an absence of one extends the selloff into December earnings.
CRWV: CoreWeave is the canary. Its bond and CDS moves have consistently led Oracle’s by days. If the investor day stabilizes Oracle credit sentiment, expect a reflexive relief trade in CRWV. If Oracle disappoints, the 15%-plus yield on CoreWeave’s 2031 notes looks less like a bottom and more like a waypoint.
NVDA: Nvidia sits on both sides of this argument. No company collects more of the hyperscaler AI buildout spending than Nvidia, the dominant supplier of the GPUs those data centers are built around. If Oracle’s investor day signals a capex slowdown, the primary transmission mechanism hits Nvidia first. TD Cowen’s lease-support thesis, if confirmed Wednesday, would make Nvidia a creditor to the buildout as well as its largest supplier.
META: Meta Platforms plans $135 billion to $145 billion in capital spending for 2026. Meta’s balance sheet is structurally different from Oracle’s, funded from operating cash flow rather than debt. A credit-driven AI spending slowdown at Oracle would actually tighten supply for Meta’s compute purchases, potentially accelerating its own infrastructure advantage.
SoftBank: SoftBank is a co-founder of Project Stargate alongside Oracle and OpenAI. The $500 billion AI infrastructure program is framed as an ambition to invest $500 billion over the next four years building new AI infrastructure for OpenAI in the United States. Any credibility Oracle’s investor day lends to that commitment directly affects Stargate’s financing assumptions. SoftBank’s exposure to the credit story is underappreciated by most equity-focused analysts covering the AI buildout.
