https://x.com/lmarena_ai/status/1891706264800936307
It's been said before but it is great news for consumers that there's so much competition in the LLM space. If it's hard for any one player to get daylight between them & the 2nd best alternative, hopefully that means one monopolistic firm isn't going to be sucking up all the value created by these things
It passed every goofy test I have for writing articles which involves trying to surface arcane obscure details. (it certainly means however they are scraping the Web they are doing a good job here)
It made the database code I wrote over the last week with o3/o1/GPT4o/Claude3.5 look like a joke.
It fills me with rage over who owns this thing.
Even if people tank Tesla's car business and run Twitter into the ground, I think our new Galactic Edgelord is going to win his first trillion on xAI and Teslabots anyway.
btw: it tried to charge me $40/mo for this thing: https://imgur.com/a/QXslgBo
Also, the "Deep Search" button was not available when it first went live, so I'm retesting everything again with this feature enabled, which I assume is a reasoning version of the same model.
Personally, I wouldn’t use it though. What’s going on with Elon Musk right now is completely insane. I hope to see OpenAI’s GPT-4.5 & GPT-5 releases to catch up soon, if nothing else. Announced for this year.
This hype will burst sooner than later and will trigger yet another global recession. This is untenable.
This lame HN trope of LLMs having no business model needs to die.
It’s losing more billions than what it’s generating. Revenue does not equate profit.
https://www.cnbc.com/2024/09/27/openai-sees-5-billion-loss-t...
If started selling 5 dollar bills for 1 dollar, I could generate a lot of revenue with $150B. You wouldn't believe the demand we would see for $5 bills.
That is only true is your primary concern in life is personal wealth and you’re burning other people’s money.
The bandwidth costs made it deeply loss making for a long time despite having loads of engagement and ad revenue. However over time they became more cost efficient at sending video over the internet and became profitable.
This strategy obviously doesn't always work, with WeWork being the canonical example. But it's not guaranteed to fail either.
The capabilities of LLMs are impressive, but none of them have published an idea I consider to have the same potential for a trillion $ monopoly that the current hype looks like.
There are far more similarities with the dot-com hype.
No critical first mover advantage in sight. All parts are replaceable for the cheapest variant with little to no down side to users.
All the LLM providers are - extremely useful - tools. Currently I can only see the 'non-monopoly' proportional improvement when their userbase grows from 100 to 1000.
But I might be wrong, and I wouldn't be surprised if in hindsight it will be obvious what the real disproportionate advantages there were to be found.
There were LOTS of funded competitors to YouTube between 2006 and 2009, including Viddler (who paid Gary Vaynerchuk a small fortune to host his WineLibraryTV show there exclusively), DailyMotion (which is still alive today, although no longer a threatening contender), etc.
In 2009 I had a coaching business and was buying marketing courses and software which would deploy your videos across 40+ different video websites (including Google Video which was a separate thing until they acquired YouTube and merged those), and YouTube wasn't yet amounting to 50% of our video traffic.
I think you might be mistaken with the bold statement above.
Remember when everyone on HN was sure Uber would never be profitable? Or Tesla? Or Amazon?
Are we really still doing this nonsense? If Open AI wanted to become profitable they could do it inside of 12 months - growing companies don't care about profitability (nor should they)
All currently known profitable use-cases are competing on price. All the unicorn examples you're biased for had in their pitch deck the network effect of being the largest.
OpenAI, Grok, etc, have shown no unique value prop or an idea with monopoly potential.
OpenAI is currently in an explicit non-profit seeking mode using a technology that we have demonstrated 10-100x or greater decreases in compute to achieve the same outcomes.
This is not a declaration that OpenAI will become wildly profitable. This is just me saying that these aren't comparable companies.
[0]: https://www.theregister.com/2025/01/06/altman_gpt_profits/
Of course, they are clearly cooking something or they wouldn't have just published a benchmark in which they do badly.
https://sacra.com/research/cursor-at-100m-arr/
Sounds legit.
I assume you're referring to this:
https://sacra.com/research/cursor-at-100m-arr/
It went from 10M MRR to 100M
Google gives everyone free access to a good spreadsheet tool, even though Microsoft Office exists.
Web browsers are free, despite the value of the entire internet.
Compilers are free, despite the value of all software collectively.
LLMs being really valuable doesn't automatically mean anyone can get rich from them.
I think everyone last year parroting "moat!" was cringe (like Altman of all people wouldn't know about this already, c'mon), but you do actually need something that other people don't have. I expect Altman's already got stuff in mind, but he's hardly the only one, and that means it's a game of "which multibillionaire with lots of experience building companies will win?", and that's hard for non-skilled normies (in this case incluing me) to guess.
Re: Dropbox, from a well known user. It didn't age well and we've been asked not to repeat it because it makes the author with connections to this site's operator look bad.
> 1. For a Linux user, you can already build such a system yourself quite trivially by getting an FTP account, mounting it locally with curlftpfs, and then using SVN or CVS on the mounted filesystem. From Windows or Mac, this FTP account could be accessed through built-in software.
https://news.ycombinator.com/item?id=8863
the controversy: https://news.ycombinator.com/item?id=27067281
>This hype will burst sooner than later and will trigger yet another global recession.
It seems to small of bubble for global recession. I mean if it is a bubble at all, there is all the reasons to believe that the strategy will work with a significant probability.
There’s a whole class of educational apps that could open up for people learning in the physical world. Whether it’s building physical things, sports or reading books or notes written in non-latin scripts... the impact will be enormous!
OpenAI's business model was literally "we don't have one: we'll make AGI and we'll let AGI tell us how to make money". This is so idiotic it's not even a scam. xAI will compete on the same plane field. Not sure about Anthropic: they seem a bit more sane.
What does the scenario look like where everyone else gets eaten alive?
Sounds incredibly valuable, but in reality collapses into Butlerian Jihad fairly quickly when you have 90% unemployment.
Edit: if the claims are true, then this will be far more destabilizing than social media. What do elections mean when the AI-guided political parties are putting out AI press releases for the AI press summaries, which voters have read out to them through their AI? What happens when >50% of the voters ask the AI who they should vote for? You end up with an AI dictatorship where the levers of discontent are fake.
But nobody really knows if that happens or not as a consequence, let alone quickly, because the transition itself only happened at all so far in fiction.
Whatever does happen, I think it's going to be a surprise to at a minimum 80% of the population.
What AI and robotics does is actually create a machine that has no use for humans at all. Ñ
The dot com bubble also gave us the most valuable companies in history, like Google, Apple, Amazon, Facebook, etc.
I quite like the idea of a future where the AI job holocaust largely never happened because license costs ate up most of the innovation benefit. It's just the kind of regressive greed that keeps the world ticking along and wouldn't be surprised if we ended up with something very close to this
Also, the pool of public domain data is always increasing, so the AI will eventually win in any case, even if we have to wait 100 years
But one thing this reminds me of is the idea of a "trap street", something mapmakers used to do was put in false locations on their maps to prove that other mapmakers were copying them: https://en.wikipedia.org/wiki/Trap_street . I figure you could do something similarly adversarial with AI to pollute the public training data on the internet. IDK like adversarial attacks on image classifiers https://www.youtube.com/watch?v=AOZw1tgD8dA . With an LLM you could try to make them into a manchurian candidate.
Not quite realistic. You are talking about very huge benefits, in favor of which licenses will be abandoned. And who don't abandoned them... I mean you can look at the Amish settlements.
Not everyone has to know about, understand, or use open source solutions for it to open the field.
You need to pay energy bill, do the update/upgrade and you need to build a LLM rig.
Nvidias Digits Project could be very interesting, but this box will cost 3k.
We are a lot closer to running it at home than i assumed we would but plenty of people prefer SaaS over doing stuff themselves.
This is just a weird dichotomy you're introducing. Open source will introduce price pressure as any competition will - that doesn't mean you won't have a monopoly.
Idk what you mean by saying this doesn't preclude a monopoly - having your pricing power eroded by competition is kinda one of the key features of what a monopolistic market isn't
A monopoly means a company has enough leverage to corner and disproportionately own the market. This is entirely possible (and usually the case) even with significant pricing pressure.
Feels like I won't be paying for anything that isn't real-time. And that any time delay I can introduce in my process will come with massive savings. Picture hiding the loading of loot info behind a treasure chest opening animation in a game, except that time difference means you can pull all the work in-house.
Openrouter.ai seems like a step in the right direction but I'd want to do all their calculations myself as well as factor in local/existing gear in a way they don't.
Right now the average person has to go through a vendor with a web app, there's not a lot of room for the public to explore.
Things could change in a hurry.
I don't think we expect a company to exist solely making a proprietary web server anymore and be a behemoth of 300B. OpenAi might end up at the same model as Nginx or Docker if they don't pivot or find a different model.
These things still cost me time because of hallucinations.
How many times do you think I've heard that over the past three decades? And you know what? They've been right every time, except for this one little fact:
The machine cannot make you give a shit about the problem space.
It matters if it is better than what you have.
If it breaks a cup but is 10x cheaper than a human, go figer
I wouldn't bet on that, given the undemocratic influence Grok's owner has in government.
That being said it's my understanding that these companies don't have many huge contracts at all -- you can audit this in like 10 minutes on FPDS. Companies need a LOT of capital, time, and expertise to break into the industry and just compliance audit timelines are 1-4 years right now, so this could definitely change in the next couple years.
That's what I mean. One example are PV panels, they are making energy production cheaper, so their producers should be good investment right? No they go bust all the time because prices are falling and margins are thin even if the volume grows. Of course the economies of scale here are different but still.
Is it? Because it seems like a bunch of megacorps pirating every single copyrighted work available in digital format, spending an enormous amount of electricity (that is probably not 100% clean) to churn through them, and the end result we have a bunch of parrots that may or may not produce accurate results so that spammers can more effectively fill the Internet with crap.