Hacker News

> And when the cash dries up this whole thing comes crashing down like a house of cards.

The problem in this market is that too many players are trying to play a winner-takes-all angle.

For the companies that pull it off, it could be very lucrative.

In a real market we’ll get a couple of big winners rather than one, but there isn’t enough room for all of these moonshot efforts to land.

I don’t see the whole thing coming crashing down, but I do see a consolidation coming that leaves some companies in a very bad state.

> Market signals on an impending AI bust are broader than just Oracle’s woes.

It's worse than that - I believe that Oracle is one of the (many) companies right now that, if their AI experimentation fails, will stop the music, and everyone will be running for a chair.

Oracle is one of a few foundational components in the circular-investing group of AI companies. If they fail to make their commitments they're the first domino to fall.

This is a pretty Oracle-specific situation, isn't it? They bet the company on an AI infrastructure buildout and levered hard to do it. Google, Amazon, and Microsoft aren't in comparable situations. Oracle is transforming itself into a value-added CoreWeave (not just in terms of product packaging but also the financial structure of the company), in a way the other hyperscalers aren't.

This story has been playing out for years now, and reads to me like the market simply recognizing that Oracle is not in the same business as it once was. It could succeed, wildly, at this new thing, but its risk isn't going to be valued based on the business it was 10 years ago.

And none of the major model makers (not counting SpaceX) have IPO'd yet

I was at the ophthalmologist for the second time in two weeks - my new prescription wasn't quite right, new lenses should be here this week.

All that to say: I had to move my focus around a bit and re-read "...pumped into AI buildout." several times, because I thought I was reading Ed Zitron :D

Hi there, how do you know Amazon's bond offering was "challenging"? Curious to learn more. Thank you.

Nothing says “full of shit” like someone saying “market is signaling an impending X”. Why not make a huge levered bet and get wildly rich if you think so?

IMHO these signals have more to do with the market than AI. They aren't finding AI to be have less ROI than before - they are requiring higher ROI than before, because there is less money remaining to be invested.

Managing the total amount of money so that investment bubbles peter out before they get excessively big is supposed to be the central bank's job.

When we tried to do a pilot with their cloud we couldn't even sign-up. None of the corporate credit cards were accepted.

In addition to that the form basically only worked in Edge. We emailed support, they changed something on the backend. It still did not work. We gave up.

In retrospective that was a very clear warning sign that their priorities were misguided. I'm glad we did not waste any further time and effort on them.

True. The linked article's title says that. I wonder if that was a typo by the OP or one of those HN quirks where the title was automatically changed when it shouldn't have been.

I think there's an errant space in between the BBB and the - but yes, the title is wrong with that space

I would say that the more a company still has plenty of old-fashioned intangible positive corporate goodwill, the bigger the notch.

Wouldn't want to be negative at a time like this.

I hope not, that would further weigh them down.

This shady site is an established business created in 1949.[1]

[1] https://en.wikipedia.org/wiki/Heise_Group

Bond rating is about financial solvency, not goodness.

he is correct on most counts and for the rest I lack the competence to vouch for or denounce his research. a rare sight!

I've been thinking for a while, there's not real winners here except the incumbent technology providers. Hear me out: all models are converging towards the same level, gains are getting smaller and harder to come by. The models are commodities nothing more.

This is the leap, nobody really wants to front a model for someone else. If i build an agent, or a service that requires a model, I'd prefer to push the model onto someone else, preferably at no cost. This is a leap as I'm sure right now, most people / businesses are thinking actually i do want to own / front the model.

However, if you accept the leap the easiest way to do this is to make the model the users problem.

From a business point of view that makes things really easy, from a customer point of view, they simply have to accept whatever their vendor of choice is pushing down their throats.

So as a business I build for whatever model Google makes available to android, and whatever model windows bundles, and whatever model Apple bundles, and, excluding the long tail of Chinese vendors and Linux (sorry, its always left out) and that's it, problem solved, and the customer picks up the tab for the tokens

I think anthropic with its enterprise strategy and google with its integration in everything have a bit of a moat.

But I switched from ChatGPT to Claude 3 months ago because my account was down for like 6 hours. I haven’t used it since. It’s too easy to switch away from chatbots on a whim. There is no moat for that.

AWS and Google at least own their own hardware (Trainium and TPUs, respectively). It's a moat in the sense that designing, building, and deploying your own chips at scale is quite a feat and not easily replicated. The vertical integration will allow them to continue to be profitable once the models get good enough and competitors' prices race to the bottom. Google has Gemini; AWS may not deploy its own models (yet?), but that's not necessarily a losing position, as long as the market is able to run models sourced elsewhere on Trainium and the price is right.

The moat is shifting from technology to access to proprietary training data. It doesn't matter how good your LLM platform is if you don't have good data to feed the training run. Public Internet data and published media is already mined out. Now the frontier LLM vendors have shifted to licensing proprietary data that's locked up behind corporate firewalls, and even hiring human domain experts specifically to create new training content in target verticals. You'll see the effects of this next year, although it might not be obvious to those who mostly only use LLMs for coding tasks in popular programming languages for which there was already a lot of training data.

You may not care, but a lot of people I know care what brand chat bot they use personally,. usually it's tied to trust and reputation more than anything else. People are fickle.

Google has a bit of a Network Effect going... my vehicle got an OTA update to use Gemini. Between that, search, storage, and the YT Premium bundle it was enough to convince me to float a subscription.

The adoption of standards like skills and agent setup helps a ton. Nobody wants to be locked into an AI vendor like with cloud systems in general. And companies can't hold on to the #1 spot across multiple areas for very long, so users are even more motivated to move their process and stack between coding tools and AI companies behind them like Claude code.

Vendor lock in cannot happen, or you're bankrupt.

Amazon Bedrock is probably middlemanning an insane amount of token consumption these days for the same reasons.

> I use their services, but I frankly don't care who provides it. I'll chase the chepest/best and have no issue switching from one to another.

For the hyperscalers, there is an ease of remaining in the Azure/AWS/GCP fabric from a data provenance perspective, particularly for regulated industries or large, risk-averse enterprises. There's also, of course, a certain network egress tax in most cases.

Nvidia has a moat. Hardware is hard. No one really competes with them for general compute

Uhh. I actively and vocally avoid all things Microsoft. I see Microsoft and I immediately think buggy software with zero security.

Most sensible Tesla valuation.

The KOSPI went up already 125% in the past year, so some sort of correction was inevitable, even if the underlying companies are healthy. The crash has been exacerbated by South Koreans levering up heavily in the past few months and now getting wiped out.

Well it seems like he bought the “AGI is 2 years away” line. As did… pretty much everyone in Silicon Valley.

> I don't know what possessed Ellison to ruin a functioning company,

Same thing that drives all these execs of large companies - naked greed!

"If only we can fire all workers, imagine how profitable we'll be!"

They are attempting to set civilisation on fire with the intention of being on top when they no longer need humans.

They'll use their purchases of TikTok and Paramount to campaign for a bailout.

They can sell the software business to broadcom.

They will ask tax payers for a bailout?

Their competitors eat them. I would not be surprised to see Oracle's cloud business get absorbed by IBM or Microsoft. Maybe Amazon. The extra DC capacity is valuable to a couple companies right now.

The lenders will then just report missed payments as revenue on their books.

Whatever happens, I can assure you that the Ellisons will remain multi-billionaires and the American taxpayer will manage to end up poorer, courtesy of their friend in the White House.

The general fallacy of the “but inference is profitable” argument is that it tends to ignore all the costs of building and training the model. Given the fact that 1) that’s not trivial, and 2) the arms race underway means one can’t stop training, then it ruins the financial picture.

It’s like saying a new apartment building is “profitable” because the monthly income covers the monthly running costs, but ignoring the giant mortgage that covers the cost of building the building. That thinking is a good way to go bankrupt in real estate and a good way to go bankrupt in AI.

Good question.

Given what happened with xAI’s excess capacity lease to Anthropic, and Meta’s noises about doing the same, seems likely that the demand for inference will continue to slope upwards for a while. If I’m Oracle, I’m not worried about being able to utilize the data centers I’ve built for some price, almost certainly a profitable one.

I’m guessing, though, that Oracle made their capital investments on assumptions of a higher price & return. Possibly because it wasn’t clear when these decisions were made how much competition OpenAI would have at the frontier.

I don’t think this math is all that hard. Capital markets have everything they need to start to figure it out, most especially a year or two of history to project forward.

HN has been split on this question, with both pro and con strongly and vigorously argued.

Inference might be profitable, but it does not mean the profits of AI datacenters will rise in future. Open weight models and local AI already put the pressure on the AI datacenter profit margins, and local AI is set to become much more efficient in the future.

I think those are just the loud minority. I wouldn't be surprised if they're like 20-30% if a poll were made here

That sentiment only seems to pop up in Anthropic / OAI threads, wonder why

Wait, they don't own US TikTok? Who does?