Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

OpenAI's H1 2025: $4.3B in income, $13.5B in loss

techinasia.com

Post reply on HN
Print view View original

571–580 of 720 posts

Re: OpenAI's H1 2025: $4.3B in income, $13.5B in loss

#571

Post by Ianjit »

necovek wrote: post #460
chermi wrote: post #183

Earlier quoted context omitted.

It's pretty well accepted now that for pre-training LLMs the curve is S not an exponential, right? Maybe it's all in RL post-training now, but my understanding(?) is that it's not nearly as expensive as pre-training. I don't think 3-6 months is the time to 10X improvement anymore (however that's measured), it seems closer to a year and growing assuming the plateau is real. I'd love to know if there are solid estimate…

Someone brought up an interesting point: to get the latest data (news, scientific breakthroughs...) into the model, you need to constantly retrain it.

The incremental compute costs will scale with the incremental data added, therefore training costs will grow at a much slower rate compared to when training was GPU limited.

Re: OpenAI's H1 2025: $4.3B in income, $13.5B in loss

#572

Post by vinhcognito »

wltr wrote: post #462
chii wrote: post #435

Earlier quoted context omitted.

when people started referring to searching the internet as googling, they know their brand has made it. It is the same with chatGPT.

Even I often tell I chatgeepeeteed the result, in the same fashion when I continue saying I googled the result, while actually I used Duck Duck Go. I could ask another LLM provider, but I have no idea how to communicate that properly to a non-technical folks. Heck, I don’t want to communicate that _properly_ to tech peers either. I don’t like these pedantic phrases ‘well, actually … that wasn’t Google, I used DDG for…

I like to go with 'I asked my bot|chatbot'

Re: OpenAI's H1 2025: $4.3B in income, $13.5B in loss

#573

Post by davedx »

raw_anon_1111 wrote: post #467
nojs wrote: post #231

I think people are massively underestimating the money they will come from ads in the future. They generated $4.3B in revenue without any advertising program to monetise their 700 million weekly active users, most of whom use the free product. Google earns essentially all of its revenue from ads, $264B in 2024. ChatGPT has more consumer trust than Google at this point, and numerous ways of inserting sponsored results…

Why do people always think that just because you have a lot of users it automatically translates to ad revenue? Yahoo has been one of the most trafficked site for decades and could never generate any reasonable amount of ad revenue. The other side of the coin is that running an LLM will never be as cheap as search engine.

> running an LLM will never be as cheap as search engine.

Complete and unfounded speculation.

Re: OpenAI's H1 2025: $4.3B in income, $13.5B in loss

#574

Post by Ianjit »

cs702 wrote: post #45

Correction: 4.3B in revenues . Other than Nvidia and the cloud providers (AWS, Azure, GCP, Oracle, etc.), no one is earning a profit with AI, so far. Nvidia and the cloud providers will do well only if capital spending on AI, per year, remains at current rates.

Are Azure/GCP making a profit with AI? ORCL definatly isn't, FCF will go heavily negative.

Re: OpenAI's H1 2025: $4.3B in income, $13.5B in loss

#575

Post by amanaplanacanal »

onion2k wrote: post #506
asa400 wrote: post #362

Earlier quoted context omitted.

What is OpenAI's competitive moat? There's no product stickiness here. What prevents people from just using Google, who can build AI stuff into their existing massive search/ads/video/email/browser infrastructure? Normal, non-technical users can't tell the difference between these models at all, so their usage numbers are highly dependent on marketing. Google has massive distribution with world-wide brands that peopl…

What prevents people from just using Google, who can build AI stuff into their existing massive search/ads/video/email/browser infrastructure? Google have never had a viable competitor. Their moat on Search and Ads has been so incredibly hard to beat that no one has even come close. That has given them an immense amount of money from search ads. That means they've appeared to be impossible to beat, but if you look at…

Aren't Chrome and Gmail also pretty much number one in their areas? I don't really use either one, but that's my impression. Also Android.

Re: OpenAI's H1 2025: $4.3B in income, $13.5B in loss

#576

Post by bcrl »

muskyFelon wrote: post #313
bcrl wrote: post #271

Earlier quoted context omitted.

This! The cost of training models inevitably goes down over time as FLOPS/$ and PB/$ increases relentlessly thanks to the exponential gains of Moore's law. Eventually we will end up with laptops and phones being Good Enough to run models locally. Once that happens, any competitor in the space that decides to actively support running locally will have operating costs that are a mere fraction of OpenAI's current busine…

Bad news on the Moore's Law front. https://cap.csail.mit.edu/death-moores-law-what-it-means-and...

The number of transistors per unit area is still increasing, it's just a little slower than it was and more expensive than it was.

And there are innovations that will continue the scaling that Moore's law predicts. Take die stacking as an example. Even Intel had internal studies 20 years ago that showed there are significant performance and power improvements to be had in CPU cores by using 2 layers of transistors. AMD's X3D CPUs are now using technology that can stack extra dies onto a base die, but they're using it in the most basic of ways (only for cache). Going beyond cache to logic, die stacking allows reductions of wire length because more transisters with more layers of metal fit in a smaller space. That in turn improves performance and reduces power consumption.

The semiconductor industry isn't out of tricks just yet. There are still plenty of improvements coming in the next decade, and those improvements will benefit AI workloads far more than traditional CPUs.

Re: OpenAI's H1 2025: $4.3B in income, $13.5B in loss

#577

Post by tobias3 »

netdevphoenix wrote: post #555
asa400 wrote: post #362

Earlier quoted context omitted.

What is OpenAI's competitive moat? There's no product stickiness here. What prevents people from just using Google, who can build AI stuff into their existing massive search/ads/video/email/browser infrastructure? Normal, non-technical users can't tell the difference between these models at all, so their usage numbers are highly dependent on marketing. Google has massive distribution with world-wide brands that peopl…

What is Google's competitive moat? There's no product stickiness here. What prevents people from just using Altavista/Yahoo/[any other search engine]. You vastly underestimate the power of habit and branding combined together. Just like then, the vast majority of people equate ChatGPT with AI chatbot, there is no concept of alternative AI chatbot. Sure people might have seen some AI looking thing called Copilot and s…

Google has defaults as their huge moat. They have Chrome and Android under their control and pay Apple and Mozilla to be the default search engine.

Here in Europe this is mitigated by them having to show a browser/search engine selection screen, but in the US you seem to be more accepting of the monopoly power. Or it seems the Judge in Calfornia seems to think that OpenAI actually has a change of winning this. It doesn't in my estimation.

On the other side Google has a monopoly on Ads. When OpenAI somehow starts displaying ads, they'd have to build their own Ad network and then entice companies and brands to use it. Good luck with that.

Re: OpenAI's H1 2025: $4.3B in income, $13.5B in loss

#578

Post by suspended_state »

gopheryourshelf wrote: post #496
jandy wrote: post #456

Earlier quoted context omitted.

Yep, this. I’ve switched to Claude for a while (because I can’t afford max plans for both) and nobody in the real world has any idea what it is I’m talking about. “Oh it’s like ChatGPT?”

Claude is also difficult to consistently pronounce for a non-English speaker. Sometimes people dont say that because it can get misinterpreted. ChatGPT is something easy on the the tongue and very difficult to mis-pronounce.

https://chat.mistral.ai/chat/12da2e83-f3f1-4a47-b432-753cac2...

I suspect they chose that name because of the proximity with the word "cloud".

Re: OpenAI's H1 2025: $4.3B in income, $13.5B in loss

#579

Post by davedx »

munk-a wrote: post #187
skywhopper wrote: post #147

Earlier quoted context omitted.

Unfortunately the chips themselves probably won’t physically last much longer than that under the workloads they are being put to. So, yes, they won’t be totally obsolete as technology in 2028, but they may still have to be replaced.

Yeah - I think that the extremely fast depreciation just due to wear and use on GPUs is pretty unappreciated right now. So you've spent 300 mil on a brand new data center - congrats - you'll need to pay off that loan and somehow raise another 100 mil to actually maintain that capacity for three years based on chip replacement alone. There is an absolute glut of cheap compute available right now due to VC and other fu…

There are some nuances there.

- Most big tech companies are investing in data centers using operating cash flow, not levering it

- The hyperscalers have in recent years been tweaking the depreciation schedules of regular cloud compute assets (extending them), so there's a push and a pull going on for CPU vs GPU depreciation

- I don't think anyone who knows how to do fundamental analysis expects any asset to "keep chugging for a decade without issue" unless it's explicitly rated to do so (like e.g. a solar panel). All assets have depreciation schedules, GPUs are just shorter than average, and I don't think this is a big mystery to big money on Wall St

Re: OpenAI's H1 2025: $4.3B in income, $13.5B in loss

#580

Post by silon42 »

gizajob wrote: post #155
adventured wrote: post #25

Earlier quoted context omitted.

There is an exceptionally obvious solution for OpenAI & ChatGPT: ads. In fact it's an unavoidable solution. There is no future for OpenAI that doesn't involve a gigantic, highly lucrative ad network attached to ChatGPT. One of the dumbest things in tech at present is OpenAI not having already deployed this. It's an attitude they can't actually afford to maintain much longer. Ads are a hyper margin product that are ve…

ChatGPT chatting ads halfway through its answer is going to be totally rad.

Imagine the emails / reports with copy-pasted ads.
Post reply on HN