311–320 of 720 posts
However the revenue generation aspect for llms is still in its infancy. The most obvious path for OpenAI is to become a search competitor to google, which is what perplexity states it is. So they will try to out do perplexity. All these companies will go vertical and become all encompassing.
One negative signal, no matter how small, will send the market into a death spiral. That will happen in a matter of hours.
There’s been loads of these signals and the market keeps ignoring them.
Not saying that will happen, but it's always good to rewatch just as a reminder how bad things can get.
Earlier quoted context omitted.
What is OpenAI's moat? There's plenty of competitors running their own models and tools. Sure, they have the ChatGPT name, but I don't see them massively out-competing the entire market unless the future model changes drastically improve over the 3->4->5 trajectory.
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…
https://cap.csail.mit.edu/death-moores-law-what-it-means-and...
Earlier quoted context omitted.
Exactly: when was the last time you used ChatGPT-3.5? Its value deprecated to zero after, what, two-and-a-half years? (And the Nvidia chips used to train it have barely retained any value either) The financials here are so ugly: you have to light truckloads of money on fire forever just to jog in place.
> money on fire forever just to jog in place. Why? I don't see why these companies can't just stop training at some point. Unless you're saying the cost of inference is unsustainable? I can envision a future where ChatGPT stops getting new SOTA models, and all future models are built for enterprise or people willing to pay a lot of money for high ROI use cases. We don't need better models for the vast majority of cha…
Because training isn't just about making brand new models with better capabilities, it's also about updating old models to stay current with new information. Even the most sophisticated present-day model with a knowledge cutoff date of 2025 would be severely crippled by 2027 and utterly useless by 2030.
Unless there is some breakthrough that lets existing models cheaply incrementally update their weights to add new information, I don't see any way around this.
Earlier quoted context omitted.
Yeah, I can rent an A100 server for roughly the same price as what the electricity would cost me.
Where?
Edit: https://getdeploying.com/reference/cloud-gpu/nvidia-a100
Earlier quoted context omitted.
I really don't trust either. Google because of what they've already done, OpenAI because it has a guy at the helm who doesn't know how to spell the word 'ethics'.
That's mostly because LLMs think in terms of tokens not letters, so spelling is hard.
OpenAI’s era-defining money furnace
https://www.ft.com/content/908dc05b-5fcd-456a-88a3-eba1f77d3...
Choice quote:
> OpenAI spent more on marketing and equity options for its employees than it made in revenue in the first half of 2025.Earlier quoted context omitted.
> Effectively every single H100 in existence now will be e-waste in 5 years or less. This remains to be seen. H100 is 3 years old now, and is still the workhorse of all the major AI shops. When there's something that is obviously better for training, these are still going to be used for inference. If what you say is true, you could find a A100 for cheap/free right now. But check out the prices.
Yeah, I can rent an A100 server for roughly the same price as what the electricity would cost me.
These numbers are pretty ugly. You always expect new tech to operate at a loss initially but the structure of their losses is not something one easily scales out of. In fact it gets more painful as they scale. Unless something fundamentally changes and fast this is gonna get ugly real quick.
The real answer is in advertising/referral revenue. My life insurance broker got £1k in commission, I think my mortgage broker got roughly the same. I’d gladly let OpenAI take the commission if ChatGPT could get me better deals.
t. perplexity ai
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…
> They generated $4.3B in revenue without any advertising program
Also, their press release said:
> $2 billion spent on sales and marketing