Unlock the Editor’s Digest for free
Roula Khalaf, Editor of the FT, selects her favourite stories in this weekly newsletter.
Oracle has emerged as the weakest member of the AI hyperscaler herd, with hedge funds now shorting both its stock and debt. Morgan Stanley’s credit analysts think things are going to get even worse.
The investment bank’s analysts Lindsay Tyler and David Hamburger had previously recommended investors buy credit-default swaps on Oracle, but hedging the exposure with long positions in its bonds. In other words, a “basis trade” on the view that the CDS would see the sharpest move on Oracle’s deteriorating creditworthiness.
They’re now recommending that investors ditch the bonds and simply go long Oracle CDS, because of its “funding gap, growing balance sheet, capex & obsolescence risk, ratings pressure, counterparty risk, and more”.
Mainly, though, it’s due to Oracle’s ballooning financial liabilities — both in the form of bonds and data centre leases — which Morgan Stanley predicts will almost treble over the next three years.
These adjusted net debt forecasts were actually made back in September, and already “may even be too low”, Morgan Stanley glumly noted.
Then there are the broader growing concern over all the debt issuance by the likes of Meta, Amazon and Alphabet, where Oracle gets the brunt of the attention simply because it is the weakest of the lot:
. . . we think bondholder sentiment has shifted this fall as a result of the hyperscaler mega bond deals, which offer investment alternatives though at a spread premium, and raise questions on long-term capex and debt needs in the broader ecosystem.
As a result, Morgan Stanley’s analysts predict that Oracle’s five-year CDS contracts will probably soon smash past the 150 basis point mark and “with additional widening possible in the long term”.
A reminder in case it’s handy: This means that it would cost about $150,000 a year for five years to insure against the default of $10mn of Oracle bonds, up from below $40,000 earlier this summer.
In fact, they warn that the price of Oracle CDS could approach 200 bps in the new year — a level it last hit at the depths of the 2008 financial crisis — “if investor communication around the financing strategy remains limited”.
And even if Oracle does feel compelled to give a bit more detail then it might still unnerve investors, Tyler and Hamburger argued.
. . . Concerns have started to weigh on the equity, which may incentivize management to outline a financing plan on the upcoming earnings call, with details on Stargate, data centers, capex, and RPO. While such communication might potentially support bonds at current levels, we also can’t rule out that it could underwhelm, excitement may fade (similar to the past couple months), or updates may be delayed.
It should be noted that a CDS price of 150-200 bps isn’t great, but it’s far from the level that would signal real fears that the company will go belly up.
Morgan Stanley’s analysts noted that the recent jump in Oracle CDS prices was probably also driven partly by banks hedging their loan exposure, so prices could fall as they gradually distribute these loans on to investors.
However, Oracle’s status as the runt of the hyperscaler litter means that its CDS market has become unusually large and vibrant. It therefore serves as a good proxy for anyone who wants to bet against AI. Which is starting to look like it might become A Thing in 2026.
Further reading:
— Oracle is already underwater on its ‘astonishing’ $300bn OpenAI deal (FTAV)
— ‘Best way to describe the market is bonkers’ (FTAV)
