The AI Thread

[Gori the Grey]

it and claude, with access to a zillion chats about my job, are suggesting lots of ways I can make my boss a small amount of money.

Quelle suprise!

 

[Thorgalaeg]

Never have, never will.

This boy never pays to play.

Yeah, stealing is much better.

 

[Samson]

Yeah, stealing is much better.

You are not stealing by using the free sample from either slop generators or heroin dealers.

Spoiler Actually :

I do not believe that heroin dealers giving free samples to generate customers is a thing. It makes no economic sense, the market is too fragmented.

 

[Samson]

Oracle is already underwater on its ‘astonishing’ $300bn OpenAI deal (FT, not currently paywalled)

It’s too soon to be talking about the Curse of OpenAI, but we’re going to anyway.

Since September 10, when Oracle announced a $300bn deal with the chatbot maker, its stock has shed $315bn* in market value:



OK, yes, it’s a gross simplification to just look at market cap. But equivalents to Oracle shares are little changed over the same period (Nasdaq Composite, Microsoft, Dow Jones US Software Index), so the $60bn loss figure is not entirely wrong. Oracle’s “astonishing quarter” really has cost it nearly as much as one General Motors, or two Kraft Heinz.

Spoiler Long article :

Investor unease stems from Big Red betting a debt-financed data farm on OpenAI, as MainFT reported last week. We’ve nothing much to add to that report other than the below charts showing how much Oracle has, in effect, become OpenAI’s US public market proxy:



The theory goes that OpenAI is in a rush to define discover AGI, and Oracle is uniquely able to scale the compute capacity it needs. Oracle promises the lowest upfront costs and fastest path to income generation among the hyperscalers because it’s a data centre tenant rather than the landlord.

Alternatively, Oracle doesn’t have as much operating profit to burn as its competitors, so is throwing everything it can at supporting its one big customer in exchange for an IOU:



At an analyst day last month in Las Vegas, Oracle said it was aiming for cloud computing revenue of $166bn by 2030:

To get there, Oracle’s capex budget for the current financial year ending May is $35bn. The consensus has annual capex levelling out at around $80bn a year in 2029, after which revenues continue to ramp:



And from 2027, the majority of revenue would be coming from OpenAI:



But Oracle’s net debt is already at 2.5 times ebitda, having more than doubled since 2021, and it’s expected to nearly double again by 2030. Cash flow is forecast to remain negative for five straight years:



So while the OpenAI agreement has been more than written off the equity, the risk of unfunded expansion remains and the cost of hedging Oracle debt is at a three-year high.



We need to add the usual warnings: Credit-default-swap liquidity isn’t great; the increased demand for Oracle CDS comes after $18bn of bond sales in September; a CDS premium in the low 100 basis points isn’t that exciting; and some firms taking the other side of the trade are no mugs. Still, pointy.

Beyond the charts, a broader question relates to whether an OpenAI deal is still worth announcing.

A few months ago, any kind of agreement with OpenAI could make a share price go up. OpenAI did very nicely out of its power to reflect glory, most notably in October when it took AMD warrants as part of a chip deal that bumped share price by 24 per cent.

But Oracle is not the only laggard. Broadcom and Amazon are both down following OpenAI deal news, while Nvidia’s barely changed since its investment agreement in September. Without a share price lift, what’s the point? A combined trillion dollars of AI capex might look like commitment, but investment fashions are fickle.

 

[Akka]

Yeah, stealing is much better.

TBH, any existing AI has been developped through stealing and should be barred by law to be elective to any sort of paying service.

 

[Thorgalaeg]

Maybe, but that is applyable to anyone using any non free for commercial use data found in internet for earning money. Bassically any proffesional who uses google without filtering results by license (so basically any proffesional).

Anyway we were talking about playing non-free games for free.

 

[Samson]

Anyway we were talking about playing non-free games for free.

Joij21 was talking about using the free versions of American chatbots for free. As I understand it the contention was that without using the non-free versions you cannot effectively compare them to the Chinese chatbots. With possibly the allusion towards not using prostitutes.