OpenAI's H1 2025: $4.3B in income, $13.5B in loss • Hacker News Board

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

#681

Post by chermi » Fri, Oct 03, 2025, 8:29 PM UTC

oskarkk wrote: post #483

hxorr wrote: post #331

Earlier quoted context omitted.

I watch a GPU repair guy and its interesting to see the different failure modes... * memory IC failure * power delivery component failure * dead core * cracked BGA solder joints on core * damaged PCB due to sag These issues are compounded by * huge power consumption and heat output of core and memory, compared to system CPU/memory * physical size of core leads to more potential for solder joint fracture due to therma…

These failures of consumer GPUs may be not applicable to datacenter GPUs, as the datacenter ones are used differently, in a controlled environment, have completely different PCBs, different cooling, different power delivery, and are designed for reliability under constant max load.

So, if anything,maybe were underestimating the lifetime of these datacenter GPUs?

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

#682

Post by buu700 » Fri, Oct 03, 2025, 8:41 PM UTC

chermi wrote: post #680

buu700 wrote: post #495

Earlier quoted context omitted.

Could AI providers follow the same strategy? Just throw any spare inference capacity at something to make sure the GPUs are running 24/7, whether that's model training, crypto mining, protein folding, a "spot market" for non-time-sensitive/async inference workloads, or something else entirely.

I have to imagine some of them try this. I know you can schedule non-urgent work loads with some providers that run when compute space is available. With enough work loads like that, assuming they have well-defined or relatively predictable load/length, it would be a hard but approximately solvable optimization problem.

I've seen things like that, but I haven't heard of any provider with a bidding mechanic for allocation of spare compute (like the EC2 spot market).

I could imagine scenarios where someone wants a relatively prompt response but is okay with waiting in exchange for a small discount and bids close to the standard rate, where someone wants an overnight response and bids even less, and where someone is okay with waiting much longer (e.g. a month) and bids whatever the minimum is (which could be $0, or some very small rate that matches the expected value from mining).

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

#683

Post by 1vuio0pswjnm7 » Fri, Oct 03, 2025, 9:09 PM UTC

No Javascript, text-only:

   curl -K/dev/stdin.*"|sed '1s/^//;s/\\n//g;s/\\//g' >0.htm
   url https://www.techinasia.com/gateway-api-express/techinasia/1.0/posts/openais-revenue-rises-16-to-4-3b-in-h1-2025
   output /dev/stdout
   user-agent: "Mozilla/()()............../......................./.... ....."
   header accept:
   eof
   firefox ./0.htm

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

#684

Post by danaris » Fri, Oct 03, 2025, 9:09 PM UTC

rajaman0 wrote: post #371

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…

Consumer brand quality is so massively underrated by tech people. ChatGPT has a phenomenal brand. That's worth 100x more than "product stickiness". They have 700 million weekly users and growing much faster than Google. I think your points on Google being well positioned are apt for capitalization reasons, but only one company has consumer mindshare on "AI" and its the one with "ai" in its name.

Sure, "ChatGPT" has entered the common consciousness as the name of LLM chatbots as a product.

But does that mean that all of the people who talk about "asking ChatGPT" are actually asking ChatGPT, from OpenAI?

How many of them are actually asking Claude? Or Gemini? Or some other LLM?

That's the trouble when your brand name gets genericized.

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

#685

Post by jrflowers » Fri, Oct 03, 2025, 9:15 PM UTC

bobxmax wrote: post #673

jrflowers wrote: post #292

Earlier quoted context omitted.

[flagged]

[flagged]

That’s a good point. Pets.com raised $82 million from its IPO pre-revenue (bad) and Anthropic raised $500 million from Sam Bankman-Fried pre-revenue (good)

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

#686

Post by typpilol » Fri, Oct 03, 2025, 11:14 PM UTC

6031769 wrote: post #544

typpilol wrote: post #363

Earlier quoted context omitted.

I actually find bing better now for more technical searches. Most technical Google searches end up at win fourms or official Microsoft support site which is basically just telling you that running sfc scannow for everything is the fix.

If you are ending up at win forums or Microsoft's support site then the chances are that you were searching for something Microsofty in the first place. And if that's the case then it's hardly surprising that Microsoft's own search engine is better for promoting Microsoft-related responses than any other. Try searching for something technical which isn't MS-specific. That should be a more neutral test.

You're not wrong. I've been doing all of windows server (eww) work lately

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

#687

Post by typpilol » Fri, Oct 03, 2025, 11:14 PM UTC

exasperaited wrote: post #664

typpilol wrote: post #357

Earlier quoted context omitted.

Also what effect will these extra instructions have on output? Every token of context can drastically change the output. That's a big issue right now with Claude and their long conversation reminders.

You mean, for instance, if you ask it to insert an advert into content, can it do so, based on its training set, without changing the wider content into advertorial? It's a really good point. And it has some horrifying potential outcomes for advertisers.

Exactly.

Am I going to come back from coding with a function named BurgerKing or something? Lol

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

#688

Post by hn_throwaway_99 » Sat, Oct 04, 2025, 12:35 AM UTC

MountDoom wrote: post #626

hn_throwaway_99 wrote: post #486

Earlier quoted context omitted.

> All of these teens use Google Docs instead of OpenAI Docs, Google Meet instead of OpenAI Meet, Gmail instead of OpenAI Mail, etc. Google Docs, Google Meet and Gmail provide a tiny fraction of Google's overall revenue. And they're hardly integrated in with Google's humongous money maker, search, in a way that matters (Gmail has ads but my guess is that its direct revenue is tiny compared to search - the bigger value…

> Google Docs, Google Meet and Gmail provide a tiny fraction of Google's overall revenue. I don't understand your argument here. Like Chrome and Android, these products exist to establish foothold, precisely so that Microsoft or OpenAI can't take Google's lunch. My point is that brand recognition doesn't matter: if you can get equivalent functionality the easy way (a click of a button in Docs), you're not going to op…

> My point is that brand recognition doesn't matter: if you can get equivalent functionality the easy way (a click of a button in Docs), you're not going to open a separate app and copy-and-paste stuff.

I couldn't disagree more with this statement. So far I've seen companies trying to shoehorn AI into all these existing apps and lots of us hate it. I want Docs to be Docs - even if I'm writing some sort of research paper on a topic, I still don't want to do my research in Docs, because they're two completely separate mental tasks for me. There have been legions of failed attempts to make "everything and the kitchen sink" apps, and they usually suck.

> Even for search, right now, I'm pretty sure there are orders of magnitude more people relying on Google Search AI snippets than on ChatGPT. As these snippets get better and cover more queries, the reasons to talk to a chatbot disappear.

I'm sure that's true for older people, where Google is "the default", but just look at all the comments in this thread about where younger people/teenagers go first for information. For a lot of these folks ChatGPT is "the default", as as that is Google's big fear, that they will lose a generation of folks who associate "ChatGPT" with "AI" just like a previous generation associated "Google" with "search".

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

#689

Post by ipaddr » Sat, Oct 04, 2025, 1:47 AM UTC

glenneroo wrote: post #645

ipaddr wrote: post #399

Earlier quoted context omitted.

Normal people ignore ads. It gets easier with time. Television with ads means conversation about what we just watched.

How do you ignore ads when you ask for a recipe and it suggests using ingredients, due to their superior flavors, textures, ability to mesh with the other ingredients, etc? Sure, you can decide to go with another brand, but over time, that stuff has the ability to stick in your brain. There have been many billions sunk into how to psychologically manipulate humans to get them to come buy your products, for instance M…

You mind starts filtering out those extras or you start using other sources.

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

#690

Post by catlifeonmars » Sat, Oct 04, 2025, 2:24 AM UTC

dns_snek wrote: post #539

catlifeonmars wrote: post #433

Earlier quoted context omitted.

Anecdata, but my nontechnical friends have never heard of uBlock origin. They all know about ad-free youtube.

uBlock Origin specifically or ad blockers in general?

Surprisingly, ad blockers in general. And everyone (I know) hates ads.