Maybe it's not the LLMs nor the weights nor the data. But there are a great many things that can make a moat around a company: culture, talent, deals, investors, brand, press attention, willingness to boil the oceans. For the moment at least, OpenAI seems to have quite many of those.
It looks like Ilya Sutskever wasn't as indispensable as he thought. SSI, Inc. is nowhere near the scale of OpenAI.
was his goal ever to become the next OpenAI? I figured it was to get a bunch of money to do his own research with friends, and that seems to be working.
I don't find it convincing that the tech isn't the moat. If the tech wasn't a moat, you'd see Microsoft spinning up its own competitor, you'd see Amazon, Apple, Meta, Oracle all have SoTA frontier models as well.
We don't see that, we see three established players in the frontier model space, and a lot of folks fighting in the second tier category.
> If the tech wasn't a moat, you'd see Microsoft spinning up its own competitor, you'd see Amazon, Apple, Meta, Oracle all have SoTA frontier models as well.
Rather: these companies consider it to be a really bad business idea to spend lots of billions for building a new state-of-the-art model that will be obsolete half a year later.
Are you under the impression they aren't burning money trying to make their own foundational models?
> Are you under the impression they aren't burning money trying to make their own foundational models?
Indeed, I think they burn money with that, but not as much as they would if they were putting all of their eggs into one casket, as the "AI companies" like OpenAI and Anthropic do (or do much more).
There exist multiple (not mutually exclusive) explanations for that:
- Limititing the "money burn rate" on AI is a political compromise that was made at the companies between various decision makers
- The companies hope that at the end even a not state-of-the-art AI model might offer business opportunities
- Perhaps such a model might give you a better "bang per buck" rate (cost of training, cost of running)
- These companies want to get experience with AI, so they currently burn a lot of money of it, but will pivot when their AI models have been out-competed
- Such a pivot could be getting from "state-of-the-art models" to "models that are insanely cheaply to run, while still being powerful"
- Perhaps the decision makers of the respective companies believe in the (not implausible) scenario that AI could from a technical perspective continue improving a lot, but these models will get disproportionally expensive to run, i.e. in the upcoming future AI models won't improve so much anymore because no one will be able to pay for it. In such a case having a slightly worse model is much less of a disadvantage.
What is interesting about the AI/LLM hype-cycle/bubble is how it can support so many entrants and players, like Cursor AI, Open AI, Cline, Gemini, Claude, Grok, CoPilot etc., and each carving a niche and a high valuation. THis is is in contrast to social networking, in which Facebook was the overwhelming dominant player, followed a distant second by LinkedIn. (Instagram was bought out.) Or Yahoo vs. Google. But now it's like a dozen companies, and each worth a lot and not as interchangeable, as seen with search engines or social networking. Facebook was the clear and dominant winner and superior platform, and there was little reason to use an alternative, except maybe LinkedIn for job searches.
They don't have a network effect, the cost of switching is almost null, and no model seem to have a clear and lasting edge.
It made sense to be on Facebook, because everyone was on Facebook. It made sense to use google, because for a long while they were head and shoulders above the rest.
Not licensing deals: https://jperla.com/blog/licensing-is-all-you-need
So was WeWork
I think people can excuse the Pied Piper's in market bubbles if they earnestly believe that what they are pushing is worth it / revolutionary even if they realize it is absurdly valued. What OpenAI and Nvidia are doing is insultingly in everyone's face. It's just a so obvious that it's become a bit of a punchline.
>"What OpenAI and Nvidia are doing is insultingly in everyone's face. It's just a so obvious that it's become a bit of a punchline."
From your post, it seems that you resent OpenAI for something, but I don't really understand what. Could you please clarify?
I personally believe that LLM "AI" is a 'bubble', but I am not sure it will end soon enough to be worth shorting players' stocks. If you believe that this 'craze' will end suddenly and soon, would it not be better to just invest accordingly, and profit from their 'absurd' behaviour?
I do resent that it is propping up its valuation by creating circular money flows. This illusion keeps them viable for just a bit longer. OpenAI promised AGI and we get SDKs. And the amount of the economy these lies and exaggerations are consuming is mind boggling. They don't seem to care that they magnify the damage through these shenanigans as long as they keep the music going just a little bit longer.
I want to say that to everyone who predicts impending doom from the sidelines.
Show me your positions. If you’re not short (or at the very least, hedged) you don’t really believe what you’re saying.
It’s a bit like saying you think the rapture is tomorrow, while you’re still contributing to your 401k…
I think requiring everyone to disclose their positions would make these conversations much more interesting.
There are some interesting points, but a few cringe things.
1) (referring to the Antifraud Company) "...they call themselves “a private-sector DOGE” -- Not exactly putting your best foot forward.
2) OpenAI paid $6.5B for Jony Ive's year old startup. That's a 'B'. Sure, stock is funny money, but it's difficult to defend the overpaying here. A lot of money seems to have been lit on fire.
Buy OTM AMD call options before the deal and then book profit to cover part of the purchases.
Or company announces plan to buy Bitcoin, but doesn't actually do so. Buys call options before the announcement. Sells call options after the stock surges and uses proceeds and inflated stock to issue secondary offering to buy the Bitcoin, and buys puts before the secondary offering, also funding the BTC purchase. Stock returns to original price. Net result is free BTC.
There are tons of financial tricks by taking advantage of materially important news and the ability to 'create money out of thin air' by exploiting price swings.
They are backed by crypto bubble experts. It is all horse trading and coercion:
https://www.ainvest.com/news/trump-ai-crypto-czar-sacks-nvid...
In today's episode Lutnick puts a UAE deal on hold until the UAE invests in the US. Things like this are going on daily. Lutnick's Cantor & Fitzgerald also handles Tether (located in El Salvador) collateral.
I'd be very surprised if "Open" "AI" did not get help from their new best friends in all those magical deals. Watch out if AMD gets any favors in the next month.
What will happen if it's a bubble and they can't buy agreed amount? I worry more about impact on AMD than about OpenAI tanking.
Related today,
AMD signs AI chip-supply deal with OpenAI, gives it option to take a 10% stake
I have an ignorant question about insider trading. Say I'm a company thinking about making a giant, billion dollar purchase. I will probably buy from one of a few companies. Can I legally buy a bunch of stock in one of those companies, then arrange a giant purchase, then announce it, and then sell the stock? Can I have insider knowledge about a company that the company in question doesn't itself know yet?
No that's not allowed
> Can I have insider knowledge about a company that the company in question doesn't itself know yet?
Yes you can
It's useful to think of this not as "insider knowledge" but the actual term: material non-public information.
Is the information you have material (i.e. will move the stock price)? Is it non-public? Then there you go: it would be ill-advised to trade on it.
So, if it's not okay for me to buy their stock, is it okay to make them giving me a bunch of their stock part of the deal? Like "I will buy a billion dollars worth of your merchandise for $1.5 billion, and you will give me the merchandise plus $500 million of your stock?"
Yes because in that case you’re acquiring the stock directly from the company as part of a planned transaction. In one case you’d be dealing with individual uninformed shareholders which would cause a lack of confidence in markets (the entire reason insider trading rules were created), in the other you aren’t.
You would get in trouble for trading on information about the company you own. Buying a bunch of stock in companies you think your company might acquire is insider trading against the company that you own.
>Can I have insider knowledge about a company that the company in question doesn't itself know yet?
Yes. However you can get around it by simply back filing a disclosure (whoops im sorry)after the fact since everyone in congress does it and it has literally never been legally pursued and never will be. I'm not being sarcastic something like 80% of congress disclosures are 2 years late or more. I don't remember the exact numbers but its insane. Just another spin on too big to fail, they can't arrest us because they would have to arrest all of congress and collapse the country.
Trading on non-public information is prohibited. Doesn't matter if you are or are not actually an "insider" in the company whose stock you are trading.
Scaling Laws are the ultimate VC ponzi scheme vehicle.
Keep raising 10x more for each round of scaling and very quickly you get large enough to be able to bully anyone into playing with you.
Sama got big enough to be able to twist any arm he wants and soon OpenAI will be too large to fail.
The Dutch East India Company wasn't too big to fail. There is nothing too big to fail and if anything too big is guaranteed to fail.
Too big to fail means that as long as government backing it is stable enough to support it, it likely will.
Sure Bank of America could fail and US Government can’t back it but that’s likely massive upheaval in United States.
Yes US can fail first and take OpenAI down with it
If by deals you mean all kinds of money shuffling nonsense that enables taking on more debt than they can ever payoff then yes. IF a person did it it would be called a financial crime.
Even a couple of years ago it was mindboggling to look at their corporate structure:
A non-profit parent entity with a wholly owned for-profit subsidiary (actually, the non-profit wholly owns and controls a third "manager" entity which in turn is the controlling but not majority shareholder of the for-profit subsidiary) where the profit of the for-profit entity is "capped" and oh by the way, that for-profit subsidiary has a 49% shareholder who happens to be Microsoft (but despite a 49% stake gets 75% of their profit..........until their initial investment is recouped). Employees get Profit Participation Units (PPUs) which are NOT equity but which entitle employees to a percentage of profits, again up to the "cap". Then of course, when OpenAI achieves AGI, whatever the hell that is is, then everything from that point forward reverts back to the non-profit entity.
I know it's changed since then, but it has always been a sort of financial Rube Goldberg contraption setup.
Corporations are persons though. Can you point to a crime they are committing?
I said if it was a person.
If I give my brother/neighbor 100k and tell him to spend it at my business then I go to a bank citing huge demand for my business and get a loan for 10 million based on that it is called fraud and the bank would spare no expense to send me to jail and get their money back. The Judge would with 100% certainty send me to jail and give me a lecture about how we live in a society and I broke that trust to take advantage of that society. That they are absolutely disgusted by my lack of ethics and morality.
When these companies do it apparently its just good business.