251–260 of 443 posts
I've been very disappointed in recent model releases, to be honest. It seems that o1 is their venture into reasoning, llms lack so much, but it's unclear if their approach does actually works towards robust reasoning. I do cheer for them and hope they can come up with something. Ai research is advancing too slowly!
Earlier quoted context omitted.
> if they don't make the numbers make sense soon then I don't see things ending well for OpenAI. This is pretty much obvious just from the valuations. The wild bull case where they invent a revolutionary superintelligence would clearly value them in the trillions, so the fact that they're presently valued an order of magnitude less implies that it is viewed as an unlikely scenario (and reasonably so, in my opinion).
You don't need science fiction to find the bull case for OpenAI. You just have to think it stands to be the "next" Google, which feels increasingly plausible. Google's current market capitalization is in the trillions.
Earlier quoted context omitted.
This is a company at a $4 bill annual run rate. In times gone by this would be a public company already. It's just an investment in a company with almost 2000 employees, revenue, products, brand etc. It's not an early stage VC investment, they aren't looking for a 10x. The legal and compliance regime + depth of private capital + fear of reported vol in the US has made private investing the new public investing.
Another POV is that, how many new household names familiar to adults can you think of? Since 2015? Basically TikTok and ChatGPT. If you include kids you get Fortnite, Snapchat and Roblox. Do you see why this is such a big deal?
Earlier quoted context omitted.
Model training is what costs so much. I would expect OpenAI makes a profit on inference services.
The hardware required is the same, just in different amounts. It’s less (gross) expensive for inference, since it takes less time, but the cost of that time (per second) is the same as training.
We can do the math. GPT-4o can emit about 70 tokens a second. API pricing is $10/million for output tokens and $2.5/million for input tokens.
Assuming a workload where inputs tokens are 10:1 with output tokens, and that I can generate continuous load (constantly generating tokens). I'll end up paying $210/day in API fees, or $76,650 in a year.
Let's assume the hardware required to service this load is a rack of 8 H100s (probably not accurate, but likely in the ballpark.). That cost $240k.
So the hardware would pay for itself in 3 years. It probably has a service life of about double that.
Of course we have to consider energy too. Each H100 is 700watts, meaning our rack is 5.6 kilowatts, so we're looking at about 49 megawatt-hours to operate for the year. Let's assume they pay wholesale electricity prices of $50/mwh (not unreasonable), and you're looking at a ~$2,500 annual energy bill.
So there's no reason to think that inference alone isn't a profitable business.
Earlier quoted context omitted.
Same thing was said of Netflix, of Uber, etc... Venture capital loses money to win marketshare, tale as old as time
Were Microsoft, Google, Amazon and Tesla/Twitter(x), and a whole bunch of other gigantic (Chinese aswell) corporations trying to compete for the same market back then? I don't think Netflix and Uber had even a fraction of the competition that this field will have.
Uber had significantly more competition than all of these companies combined. Including from China which they were forced out of.
Earlier quoted context omitted.
There's no lock in but there's also no alternative. Bing is vastly inferior. ChatGPT meanwhile can be directly challenged by Claude, Gemini and several others.
I guess you mean free alternative. Kagi at least provides very competitive results.
Earlier quoted context omitted.
Difference is is that I can just switch to another LLM. I cant really do the same for Google.
I haven't used Google for year, and I find better results than ever.
Earlier quoted context omitted.
Google is a digital advertising company. OpenAI hasn't even entered the ads business. In the absolute best case they can take over a large chunk of Google's search market share, sure, but that still doesn't make it anything similar to Google in terms of finances. How do they start making the queries profitable? What do they do when their competitors (Claude, Gemini, Llama, Mistral, Grok and several others) undercut t…
Google didn’t start as an ads company. It started as a blank text box that gave you a bunch of good answers from the internet in a list of links. Were there competitors that did the same thing? AltaVista? Yahoo? Did they undercut on cost? Google was free, I guess. But Google won because it maintained its quality, kept its interface clean and simple, and kept all the eyeballs as a result. Now Google is essentially the…
I already use ChatGPT as my first go-to stop for certain search queries.
Earlier quoted context omitted.
> if they don't make the numbers make sense soon then I don't see things ending well for OpenAI. This is pretty much obvious just from the valuations. The wild bull case where they invent a revolutionary superintelligence would clearly value them in the trillions, so the fact that they're presently valued an order of magnitude less implies that it is viewed as an unlikely scenario (and reasonably so, in my opinion).
That assumes that the revolutionary superintelligence is willing to give away its economic value by the trillions. (Revolutionary superintelligences are known to be supergenerous too)
Earlier quoted context omitted.
Uber was a literally life-changing product with an obvious value for anyone. LLMs have neither benefit.
Every cafe, airport, school I've been to has people using ChatGPT or its competitors. Its obviously valuable for almost anyone. Just like how people cant imagine life before smartphones, people wont be able to imagine life before LLMs became ubiquitous. Its everywhere.