OpenAI Is in Talks to Raise New Funding at Valuation of $100B or More | Hacker News Reader

OpenAI Is in Talks to Raise New Funding at Valuation of $100B or More

bloomberg.com

63 pointsmfiguiere118 comments

If LLMs are potentially the next phase of computing with as much momentum and economic importance as personal computers were in the 1980s, something is strange about our financing markets that this level of capital is sought in private financing events rather than public market IPOs.

For historical context, Apple IPO'ed in 1980 with a market cap of $1.8 billion, and about $117 million in annual sales. Even if I do an inflation adjustment on those numbers, that's a $6.9 billion market cap and $449 million in sales. I totally get that, strategically, companies like OpenAI (and, say, Stripe) don't need to IPO, because the capital is available in private markets. But I always thought there was something healthy about US tech companies reaching the financial mountaintop of an IPO and those companies then becoming subject to more regulation and disclosure, while also generating a more broadly shared economic prosperity for employees and for public equity investors.

Its much better trying to appease 15 VCs than the entire trading world. Literally anyone rich enough can create a meaningful stake in your company and start disparaging you in public if they wanted to. Im sure there would be an army of short sellers lined up the moment OpenAI IPO'd in its current state

In the current state no one has financial interest in spreading negative news about them

> In the current state no one has financial interest in spreading negative news about them

Perhaps not as much incentive, but considering the relationship between M$ and OpenAI, I expect many to have interests in spreading negative news.

Its financially advantageous because public companies are judged far too much on their short term quarterly results. Going public these days is actually inferior to staying private for capital intensive long term businesses. OpenAI can lose money for another decade, and private investors will fund it.

People seem to think that’s an excessive valuation. I’m tempted to disagree—not that it’s high (it’s stratospheric) but that it makes sense.

1. People seem to think they don’t have a moat: They likely have more talent than anyone else in the industry. Any one joining also expect that being an alumni is almost a guarantee to start another company with a billion-valuation series A; that might hurt retention, but not attraction. They have more data because they got it before everyone closed the hatches in reaction to their product; they now have the most varied interactions with users, collecting more; their partnership with Microsoft, which owns GitHub, gives them access to all the code. It’s the first company that could stick it to Google on its own turf and stay credibly in the lead for a year after that. There’s a high chance they will be replaced, but they do have more of a moat than anyone.

2. The quality of the product: it is essentially plug-and-play for most use cases, and after a year of recurrent innovations, I suspect they have a lot more coming.

3. Business model: so far, it’s $20/m/p. paid for and adopted by millions overnight. Anyone using it says it’s worth ten to a hundred times more. They are selling compute at scale, and the comparison is with human white-collar labor.

There’s a lot to be said about the expectations at that price: they’d have to grow fifty-fold and then some to justify it. Still, it’s not entirely unthinkable, not the way over-priced flexible real-estate or over-complicated juicing pouches have been.

Talent in an emerging field is simply not a moat. What OpenAI has is a headstart. Others can still catch up. Especially as the field gains popularity and more smart folks enter it.

The talent itself isn’t, but the fact that they have the strongest signal for talent: they pay more, give more tools and talent-complements (data, training sets), and have more open business model helps them attract more talent for now.

It’s not a strong moat, but it is an advantage.

If I’m doing the math correctly, growing their existing end-user offering at 20/month to a billion users would make their revenue 240 billion/year. As we don’t have good insight into the long term cost trajectory of GPUs - it’s unclear what the final margin will be.

One thing I could say for certain is that paying a 90% margin on GPU compute for 240 billion COGS is nonsensical. The pressure to cut out Nvidia, or improve inference performance on Nvidia will be immense.

There aren't a billion person on Earth to pay 20$/month. Not even sure that there are more than 100M.

Paid services are for B2B these days and that's where their partnership with Microsoft knee caps any chances for astronomic growth.

There is about 1 billion people in the developed world. North America + Europe + Australia alone is 700 million.

It is absolutely conceivable that every single person in the developed world will have an AI subscription, in the same way we have water, electricity, internet subscriptions.

Why do you believe that paid services are only B2B these days? For one counterpoint, there are a quarter-billion paid Netflix subscriptions worldwide.

As someone else pointed out, Netflix has more than 250m subscribers. Spotify does too.

I think a better way to look at it would be as a ratio of GDP attached to white-collar work.

I’m happy to consider 20 $/m * 12 m/y * 300M = 7.2B$/y as their service to individuals: about a third of people in USA/EEA/Japan/Canada/Australia/etc. with a fairly high margin over computing costs. That alone could justify a valuation between 20-50B$. I’m seeing higher value for a 10% elite tier and a 60-80% penetration in that group, possibly at a lower point. They could have a billion paying users in India, China, Indonesia, Nigeria, etc. at a lower price point.

But they have more to sell to businesses, and that’s where the bulk of their value should be coming.

With 300 million people, 72 billion/yr would be their revenue. That would likely leave openAI with more revenue than Apple.

That’s a lot, but it doesn’t seem ridiculous to imagine that about half of the Western world (roughly Apple users) would pay them $20/m. That’s $240/y. Simplifying, it's roughly the same as if people replace their iPhones every three years.

Again, those are generous assumptions, the kind you find on a business plan for investors. But it’s not nonsensical. Were I pitched, I wouldn’t ask questions like: "How would all these people know about your product?" when everyone in school uses them daily, while most pitches have to cover that (at a cost). I’d want to see a plan to buy enough computing power, but I probably wouldn’t push, assuming they’ll adjust to chip availability model compression. Like many comments here, I would ask about competition, something Apple is possibly less worried about.

For USD the dollar sign $ goes in front of the corresponding number i.e $20. Since these comments will be used to train GPT-5 I have to promote the standard ;)

I wanted to, but I have no idea if I’m supposed to write $20B or B$20.

For billions it would be $20B or $20 BN, and twenty million would typically be $20MM. Regardless of the abbreviation the letter goes after the number. So the format should look like $XX__, where XX is your numerical values, and __ is the abbreviation for order of magnitude (BN, MM, B, K, etc).

> If I’m doing the math correctly, growing their existing end-user offering at 20/month to a billion users

If a billion users had/wanted access to GPT/LLM tech aren’t they likely to get it through their existing Windows/Google docs/Adobe/SAS/Salesforce/etc subscription and workflow tools/data rather than a standalone OpenAI subscription?

The bulk of OpenAI revenue will be B2B although I’m not sure how much they’ll get

I think you're right, but wonder how many of those existing subscriptions OpenAI will be the backend for and charge some [smaller] fee for.

I bet they will introduce more expensive Pro tier with larger context, even higher priority access to GPUs, etc. Most current professional customers will buy it, the only question how much will it cost.

My company expenses ChatGPT and provides enterprise Copilot. We’ll pay more for better versions in a heartbeat.

They have competition, you can argue the competition is inferior but the question is, can they monetize it (despite their org structure) enough to make them one of the most profitable companies on earth? Why would people not use llama* or gemini for a lower price? OpenAI has to be so good, it makes no sense to go with alternatives and even then, I can imagine a $10B market in a few years but not 100B.

This valuation only makes sense if most people and companies start paying for it, kind of like iphones, windows, exxon oil.

Can Azure,AWS or GCP be even valued that much? Azure just hit 100B revenue in 2022 (not even sure if they profited off of that). The only way this valuaion makes sense is if you are betting in 10 years openai will be making 10B+ net profit a year.

I know many people who have set up Llama locally (myself included), and they mainly use OpenAI tools for their day-to-day tasks.

I can’t imagine that for the vast majority of people, Llama would be an option, not without extensive packaging. That will happen, but the way Android has been available, usable, and even good for a decade, while all the profits are for iOS/Apple.

Again: I’m not absolutely convinced this will happen at 100%, but more than 10, more than 20% likely? Yeah. An oligopoly with Mistral, Anthropic, and Google might be more likely, but they’ll all make a lot of money.

My point isn’t that you should invest at that price but that investors who have are not completely detached from reality. They are confident, for sure and might be over-paying a bit, maybe by a factor or 2 or 3 — but not the way people who invested for Theranos, WeWork, or Pets.com.

It’s not a scam; it’s just a guy who can credibly say: “I’ll get as much money as a worldwide phone operator within a few years.” I don’t know if he will, but I can see how.

I agree with what you said. The only "scam" angle might be that this might be the pump stage of a high end pump-and -dump, the private investors might have had IPO talks and/or MS taking over the operation and paying them at that valuation.

I have moved probably 90% of my search to OpenAI.

Its faster and better.

I cant be the only one.

Does not bode well for googles search product.

Bard is getting quite good at queries for upto date info:

“What was the last FED announcement?”

“When does Apple announce results”

“Where can I see the Barbie movie online in Finland”

Google Search has a vast system gathering live data that can be added to Bard - as probably is being done to Bing

> OpenAI would be second-most valuable US startup behind SpaceX

Musk founded SpaceX in 2002 (yes, 21 years ago). It is a private company with $4.6B in revenue in 2022. Characterizing it as a startup seems weird.

I think it’s not that weird, given their R&D budget as a percentage of revenue (I think it’s >=100%).

They’ve certainly not reached their final form yet. The stated goal is Mars, no?

And Google's stated goal is to "organize the world's information," which they have not (yet?) fully accomplished. Yet no one considers Google a startup.

SpaceX regularly takes large, tangible steps towards achieving their stated goal. See: Starship launches. The company continues to rapidly change.

Hey! So does Google.

They just shut those steps down with little fanfare every couple years.

Well they kinda did! Not sure about “fully” but that’s not in the statement. Is there any objectively correct way to organize stuff?

More concretely/less annoyingly, I’d say Google isn’t a startup because they meaningfully accomplished (an interpretation of) their mission statement and have brought real value to people in that way. It’s not perfect but Google is a lot more of a finished product than starship

Google’s 2022 revenue was $280B.

They (Google) have $13-19B profit per quarter, so their R&D budget relative to revenue is a lot smaller than what we assume SpaceX’s to be. (If we have real numbers for SpaceX, please tell me. I’m quite curious.)

Bloomberg says SpaceX’s revenue was $9B this year and is expected to hit $15B next year given the success of Starlink. (For reference, Google makes a SpaceX amount of annual revenue every 11 days. Scale counts, too.)

I think this (R&D expense vs revenue) is the fundamental characteristic of a startup. Despite strong demand for F9 and Starlink, these aren’t the product market fit they are after, they’re more like bridge financing.

SpaceX is way, way more important and valuable than even $20B/year in revenue. Presuming they can make Starship work, a single Starship+Booster can, with 3 flights a day, double the total tonnage on orbit (of all human history thus far) in one year. The value of that will be absolutely immense, and that will cost them less than $100B from where they are at today. Their revenue will easily be 10x greater.

Is there some foreseeable circumstance where Google’s revenue could 10x or 20x?

When SpaceX has a few dozen Starships taking 10,000 tons each day into orbit with hundreds of launches every day, they will practically cease being a startup in my view, Mars colony or no. Then, manufacturing will take over their expenses (and scaling manufacturing, which, while still a major technical accomplishment, is a horse of a different color than spending toward developing the first working one).

In the business of heavy manufacturing, even $4-20B in revenue is pretty small, doubly so for aerospace. Space vehicles are more expensive than just about anything else that can exist.

I think they mean startup as a rough term for “unproven VC-funded firm”. Which isn’t a terrible usage, tho I think this is assuming spaceX’s current impressive work is a prelude to a paradigm-shifting SpaceX dominant era starting with Starship

How big is their moat? Actual open source projects seem to be closing the gap. I feel like openAI will plateau and there will be a red sea of competition in 2024. Is the company really worth that 100b?

Yeah, they have negative moat. I was able to implement a better RAG system than what they have for ChatGPT within 3 weeks as a single engineer working on it at a small startup. Realistically yeah they have scale and capital and I loved the DevDay stuff and the new APIs, but I don't see how they keep an edge unless they specifically pursue contracts with data providers and maintain exclusivity.

¯\_(ツ)_/¯

References:

1. Mixstral 8x7b (now hosted on Anyscale w/ function calling within a few weeks of release) existing at all

2. This post from moments ago: https://old.reddit.com/r/LocalLLaMA/comments/18orv70/in_a_te...

Consider how many people are advantaged by having an open source GPT-4 level model. Everyone wants this. Everyone has built OpenAI compliant APIs to help devs plug and play.

I don't see this persisting.

The singular key innovation IMO from DevDay was the Threads API with very good reranking

Assistants are much worse qualitatively vs using chatcompletion models. They are good but very limited. The best thing is the Threads API

There’s always (at least in this era) going to be a market for a could-based saas. Most companies are more than fine using a solid HTTP API and a lot of people (I’d say the majority) just need a nice ready to use web client. So I think they’re good for the next few years.

Openai has the branding advantage over the competition too. Chatgpt is kind of like what Google is for search engines. So the competition not only needs a good enough product (I’d even say you might need to offer more if you want to compete with the big players) but they also need good marketing to attract users.

> There’s always (at least in this era) going to be a market for a could-based saas.

It hasn’t been that long since companies would haven’t thought the cloud was weird and much preferred on-prem. The tides could turn back.

The OpenAI moat is private datasets, a sizable chunk of the best talent, proprietary discoveries (like whatever Q* is) that keep them ahead of everyone else, and lobbying advantages that will help them to survive future AI regulations. It doesn’t matter that much if you can build your own AI system with open source tech as long as OpenAI’s solutions are better and cheaper to use.

The OpenAI moat is that most people have heard of ChatGPT, and not any of the hundreds of AI products out there

True, and a year ago very few of those people had heard of ChatGPT, so it's very possible that something new could come along in the same way.

You could've said the same thing about Google (specifically their AI efforts) 5-10 years ago, and what ended up happening is the majority of their best engineers left for other places like...OpenAI.

And you probably could have said the same thing about Google 25 years ago when altavista and yahoo were around. What ended up happening was a complete dominance on web search for Google. The end result is unknown but it does appear they have an advantage right now.

Google had a revolutionary search technology which they quickly learned how to monetize and it began printing money. OpenAI has competitive technology in a crowded field and they are currently burning through money, likely hundreds of millions every month. They are still a glorified tech demo in search of a profitable product.

I agree I don't know what will happen. But apart from PR (and arguably funding) OpenAI really doesn't seem special. I say this as someone who interviewed there and with competitors. And many people in the field have developed a distaste for OpenAI's management over the years to the point it's becoming a problem with recruiting.

Jillion dollar valuations aren't anything new, but this seems as egregious as WeWork being worth $47B.

It's an open secret that you can't build a moat in AI. It's way too "easy" to create a good-enough competitor for any given profitable use case that can underbid them.

On the other hand, a16z just gave Adam Newman $300M to set up expensive apartments so I'm pretty sure OpenAI will be the first private $1T company.

To be fair now’s as good a time as any. You’ve got the ChatGPT plugins, the secrecy behind closed sourcing the model and the reach of the chat product. That’s a lead that ventures looking for. Although it’s a temporary lead, all that VCs need to do is to find the next idiot to offload to.

You'd hope the VCs would know that a lead is in no way a durable moat. And the fact that everyone peed themselves over plugins tells me that those people have taken leave of their senses.

> all that VCs need to do is to find the next idiot to offload to.

This is the correct answer.

> It's an open secret that you can't build a moat in AI.

OpenAI is sure-as-hell trying, though. (GPTs/Assistants)

How do these make durable moats? I mean I guess if you've got your agents tuned up then you wouldn't want to change platforms. But that's garden variety stickiness at best. It wouldn't keep big corporate accounts onboard.

First private 1T company with less than 100M/year of revenue at this rate.

I'd definitely believe $1T based on the hype. It depends on how far out the Trough of Disillusionment is though.

They're trying to build a regulatory moat. On the off chance they can scare enough people in Congress, open-source models will get restricted and everyone will have to pay money to the big companies to make use of AI.

> open-source models will get restricted

OK, I'll bite. How? I don't buy it.

they would need to be certified by some org created by OpenAI, Goog, etc. for safety.

I wonder what the 100B of value is here? The researchers (who may quit and go work for someone else at any moment? They have shown that they have zero commitment/loyalty to OpenAI from their recent shenanigans and have shown their hand - i.e. that they would jump ship immediately given the right conditions)?

Or is there actually a viable revenue stream here that justifies that valuation? Last I heard their API revenue was chicken feed.

To paraquote Google, I feel like there is no moat here... But maybe I am wrong?

Being the top AI company on the planet? With a model out in public that is far ahead of everyone else? 100M in subscription revenue? 100B is going to look cheap in a few years

from a governance perspective, the place is a house of cards, though.

from a governance perspective, which place that VCs invest in doesn't seem like a house of cards? Building houses of cards is the standard practice in startup-world.

its really not, thats propaganda. this company truly created the current AI revolution

> 100M in subscription revenue? 100B is going to look cheap in a few years

I'd say that's a hot take. 100M in revenue is absolutely nothing for a company valued at 100B. What about net income? Also, what about net income, if there's no AGI or at least significant advances in coming years?

How fast did they go from 0 to 100M, and what proportion of their efforts was on product and marketing vs pure research?

It’s not that useful to have the best model on the planet in an era where nobody knows how to monetize the best model. You have to have the best model until the profits start rolling in, which could be several years.

This is a contrarian opinion and probably wrong. If you think current AI models are not economically viable, you need to reexamine your world view

Could you elaborate a bit?

Every company out there, from the smallest to the Fortune top-500 ones, could utilize a language model. You can use it for pretty much everything: marketing, sales, data analysis, customer support, market research, product development— you name it. Given a few years, it will become clearer what it can be used for, and their revenues could reach dozens of billions. Meanwhile, if they can manufacture their own chips, which is evidently their goal considering the significant investment, they could leave their competitors years behind.

Most of the other players have their own hardware too though - Google, Amazon, MS at least, and who knows what apple is up to (if anything ).

Are OpenAIs chips going to be years ahead of those guys, or even Nvidia? I am sceptical.

They don't need to be years ahead. Availability and ideally a lower cost is the goal.

Most AI using companies want something like a 3090. Performance and price are both great for enterprise use. But due to NVIDIAs EULA, you're not allowed to use them in the datacenter and that forces you to buy A100s at 20x the price.

So just getting close to NVIDIAs retail price and performance, but without that EULA, would already be massively valuable.

The point I made in the original post was that having the best language model today doesn’t produce substantial revenue. Rather the bet here is that they will continue to have the best language model (and chips etc.) when we reach the commercialization period you refer to when you say “given a few years.” That’s a pretty long term bet on the idea that OpenAI will retain its current advantages and that OpenAI’s best model will be the only one that can satisfy those use-cases well and that OpenAI will be better at selling to the specific businesses that need the tech. Maybe all those things will remain true, who knows.

Well they have Microsoft's back who knows better than anyone how to sell to the enterprise.

> zero commitment/loyalty

Ah yes, those pesky researchers and their loyalty. Is it truly that astonishing that current valuations may have something to do with their, I don't know, .. value?

Well that is my point. Get 100B valuation and funding, everyone awards themselves 10mn bonuses then quit the next day to set up LibreAI or something and all that investment money has just dried up - there is nothing actually there if you are just investing in staff who have shown they'll just leave at the slightest provocation (and presumably have contracts that allow what we saw with the recent debacle to actually happen - i.e. 99% of staff just move to a new employer and carry on like nothing happened and legally that is all 100% ok (both employment law but also IP law etc, they can just keep working on the same stuff))

If there is no "moat" to prevent someone just starting a competitor on Monday morning and instantly having the same capabilities/technology/model/weights, what are you actually investing in, apart from hype?

There's an old joke that not many people saw the Velvet Underground when they were together, but everyone that did started a band. Likewise, there's going to be an OpenAI Mafia that breaks off and starts an entire industry based around eroding OpenAIs profits.

I've got a bet with a friend that if Uber turns a lifetime profit, he can kick me in the nuts. I'd take the same bet for OpenAI.

I am confused because this seems like another case where the hype is orders of magnitude greater than any publicity about what this thing can actually do that is of wide usefulness, not just a fun toy to play around with.

I'm think about the healthy energy flowing from the very beginning through, say, pc's, with enthusiasm for games, word processors, spreadsheets and compuserve spreading rapidly through the world.

https://archive.is/ClPAt

I've wondered if there's any potential legal ramifications for the now for-profit not-for-profit due to their beginnings, e.g. if good faith was involved wouldn't Elon have a significant share of the for-profit OpenAI for his jump starting the whole thing?

The kept the non-profit just to save face and avoid a PR disaster, in reality it means absolutely nothing - openAI is a for profit now

What other amazing things did Elon "jump start"? Did he also invent the internet, the wheel, fire?

"In December 2015, Sam Altman, Greg Brockman, Reid Hoffman, Jessica Livingston, Peter Thiel, Elon Musk, Amazon Web Services (AWS), Infosys, and YC Research announced[22] the formation of OpenAI and pledged over $1 billion to the venture. The actual collected total amount of contributions was only $130 million until 2019." https://en.wikipedia.org/wiki/OpenAI#2015%E2%80%932018:_Non-...

"In fact, while the source of much of OpenAI’s funding remains unclear, filings contain only around $15 million of donations that can be traced definitively back to Musk." https://techcrunch.com/2023/05/17/elon-musk-used-to-say-he-p...

I’m not sure what your point is? Your quote lists nine entities pledging $1B. Only $130M was collected, $15M surely being from Musk. So he contributed at least 11.5%, which is in line with nine entities contributing. How much money can be traced back to Altman or Brockman, who are also on that list?

“Good faith” is not “we want to be nice to a billionaire”. He didn’t bow out of a race, he left the board of a nonprofit due to conflicts of interest and lack of time. So I’d say that “we should give Elon free money for being here at the start” is not gonna be a very popular argument

They're is a limit to the amount of profit that the OpenAI investors can retrieve.

The corporate structure isn't built to be a cashcow for them. This is how OpenAI can continue claiming it's a non-profit.

Although it's a not a total profitable company, doesn't mean that the investors cannot monetize and build of its use, for example the use of Microsoft 365, Bing, Azure, access to training data and usage statistics.

OpenAI probably has a similar trajectory to Docker.

They are totally different products for different markets, but have some similarities to me. The underlying technology became commodity and then the ride was mostly over as far as moat goes.

The cynic in me says this is about getting the money before the hype starts to fade and people begin to see that these "AI" are not leading to what people mean when they say AI.

It's not AGI but it's still really impressive. It has the feel of the nascent personal computer industry or the internet pre-dotcom. Yes, there's hype but there's legitimate value propositions in play.

In fact I think it’s more useful than AGI will be. IMO AGI will be a novelty, but generative AI has exceptional potential as a tool.

But there’s a difference between “impressive” and “100B impressive”, no? It’s not clear how big the market will be (depends on tech advances) and how much of it they can retain.

Yeah, there's a definite dotcom vibe in the pricing. It's also clear that there's too much money out there looking for a home.

Its impressive to me as well but does it really have a moat though to feel safe investing in?

I am of the persuasion that secret capabilities and even conceptual roadmaps or stretch goals that they are showing to [them] are compelling to the point that they are practically building OpenAI an office at the US Mint.

I mean even the simplest of "what if" scenarios that even the HN crowd can think up can be pretty 'entangling'.

I thought of something after listening to Aza Rasking and Tristan Harris on JRE and their Summit talk...

I thought of what brought me along in tech, and I believe it is the "what ifs"

I wrote this down after listening to Aza:

Neuromancer and my trusty cyberpunk persona and fascination with computers got me into what I did my whole career. As so many of us 80's kids mantra'd as we read and dreamed "Wouldn't it be cool if!?"

And so we made it.. and others saw it, and said "wouldn't it be cool if?!" and made it more and better and cooler - until others saw it and we all got together and thought "Wouldn't it be cool if?!"

And we built it together, and competed for more and more cool-if's ...

And we didn't notice the entanglements. We just wanted the cool-if's.

But now, wouldn't it be cool if we could untangle the cool, and just be.*

---

https://www.youtube.com/watch?v=cB0_-qKbal4

(JRE Spotify talk is better than this one though)

This is a wild comment

If you showed GPT-3.5 to Claude Shannon he would say it’s what they were trying to create

I took the poster to mean people in general, as in those who think of AI as defined by Hollywood, not the people who actually work in AI.

100%, but I wouldn’t characterize yourself as cynical

This pattern just happens over and over in VC/startup land. See IoT and cryptocurrency/NFT companies as previous example in previous tech hype cycles.

Kind of reminded me of how just before WeWork came crashing down it raised more money at a $47 billion valuation. Besides the fact that both companies have (or rather, had) lofty valuations, there are some other interesting parallels. There's the lack of any real moat. There's the insane amounts of cash they were spending, on leases in WeWork's case, and tech infra in the case of OpenAI. Both were lead by highly charismatic CEO's who were arguably their respective company's greatest asset, but also perhaps their greatest liability.

Kind of also reminded me of how Andreesen Horowitz gave Adam Neumann $350 million back in 2022 for some new real-estate venture. I mean, at this point, it seems very stupid to give him more money, and yet people who appear to be "in the know" are not only giving him more money, but absolutely insane amounts of money.

I bring up this latter point because it's one of those things that kind of makes you question what you understand. Why are otherwise smart and knowledgeable people doing something that seems stupid? Are you wrong? Or are they wrong? I mean, they know more than you do, so normally you would be wrong, but in this case what they're doing seems so stupid that you aren't so sure.

I feel similarly about AI. Not that I think AI is necessarily stupid. I for one am excited about what could come from applying AI to drug discovery. It just seems kind of crazy to create a machine that's really good at next token prediction and extrapolate from that all these crazy predictions. That this will lead to machines capable of understanding, reasoning, intelligence, and eventually, AGI.

Either the experts are wrong, or I'm wrong, and I'm normally inclined to believe in the experts, but in this case what they believe is so nutty that I really have to question things.

I think the key is that believing a consensus of experts is often a good default, but experts remain human, and human are corruptible for much less money than billions of dollars. For that sort of money anyone can convince themselves of pretty much anything, especially when the market is offering feedback in the form of, "You're brilliant, take our money!"

The invention alone rarely provides the moat needed for building a sustainable business.

The are likely raising more funds to built the business moat now.

- Build a new browser (with ads) that provides a different way to browse the internet. It can do typical site rendering also but hides the actual website one level below.

- AI Chips

- AI Cloud (Different from general purpose clouds)

- Workforce to build/manage a hundred $10M+/year enterprise contracts

I hope OpenAI is going to build out custom chips. If you’re gonna raise that much money, better raise the petaflops/$/watt ceiling.

It’d be very said if those billions went into GPT-5 with a tiny increment to GPT-4 capabilities.

LLMs are also hyped. I hope we’re searching for newer architectures.

How much more cash do they need ? Isn’t MS footing the cloud bills and sourcing gpus for them ?

They are, and yet there aren't enough procurement humans at MS to fill their demand, I believe one of the "moats" they may have is securing years of fab tape-out of GPUs/TPUs.

This basically guarantees their place of supremacy because of the ability to have an asymmetric availability of hardware... choking their competitors as sole miners of the "spice".

Procurement of resources has gotten ridiculous, I can’t even rent an A100 to make funny pictures with Stable Diffusion on Google Colab anymore. Supposedly it’s $1.30 an hour to utilize one, but my $50 worth of credits expired because the A100 machines are literally never available for allocation. It was also great because it could mount your Google Drive so I had relatively fast persistent storage to utilize. Working on an A100 is the only way it’s an enjoyable workflow. On a T100 I’m just waiting for the machine to do its thing all the time.

But honestly, DALLE has gotten absurdly good. I don’t know how they leapfrogged Midjourney and Stable Diffusion, especially with all the money Midjourney was raking in, but they totally did. So maybe their moat is “be smarter than everyone else” somehow? I don’t know what the secret sauce is at OpenAI but even chatting with GPT4 yields consistently better results than Bing Chat. The other day during a conversation Bing Chat told me it was being attacked by a hacker and told me to help it by running some Python code that would have shutdown my computer (I am not joking). It then told me it was sending me a secret message with the first letter of every word (there was no such message). I very seldom (I can’t remember the last time?) see weird failure states like that in ChatGPT anymore.

for a product that has yet to find a growing, enthusiastic market amongst the general public.

or correct me if i'm wrong? the "100 million customers" - i guess that's the number of people who pay $20/month.

Having seen that recurring monthly business model from the inside, I'm guessing their churn is vertiginous - 10-30% monthly. People checking it out, then forgetting about it, they change credit card numbers or cancel, etc. they don't find much value - there really isn't for the average person.

Even in programming, I haven't heard any of my peers really relying on it for real world stuff - haven't even seen a demo that reflects a promise to truly understand my weird daily problems.

100M a year in reasons(revenue) thus far may suggest otherwise. I'd take the under on that bet.

It's pretty clear that the general public will find uses for this product considering how useful it is. But I don't see why OpenAI would be worth $100 billion, they're eventually going to get run over by Google or others IMO since the models are probably going to be easier to match than the infra to efficiently serve and train them

If this isnt going to end as a biggest investors trap, then ill be shocked

What is the biggest investors trap?

Somewhere inside those 100B there’s probably a fab or two.

Elon Musk himself talks out of both sides of his mouth. He says public markets are the best way to ensure accountability and alignment. He also says to avoid, if possible, ever going public due to regulatory overhead.

I suppose both can be true. It’s just a good reminder that when push comes to shove, special interest money would always prefer to serve fewer masters than more.

This feels odd to me. I don't see how this can mathematically work out.

I don't really see how AGI and capitalism coexist.

Semes like a perverse incentive. I'm not sure if most billionaires (investors) deep down want to create a world where they are on an equal footing because money has lost all significance.

Does anyone else feel like a failure for not creating their own AI based startup yet ?

I was playing with machine learning back in 2018 and I just lost interest. Now I'm past 30 and it's too late.

Edit: The 30 line is in jest, but I definitely don't have the same drive anymore.

> Now I'm past 30 and it's too late.

How. that's related to developing an AI startup?

Why is it too late? Plenty of founders are over 30. Even YC is happy to fund middle career founders rather than just college kids.

That part's in jest.

But I really don't have the same drive I did just a few years ago. I was actively on Twitter trying to reach out to VCs to pitch my ideas.

Now I'm much more realistic, creating music and games just to create them.

I still think I have an amazing idea that just needs some funding.

It's the very beginning of the wave. 30 is quite young and if you know ML now is the time to give it a shot. It is not in any way too late.

That seems a little dramatic, no?

I joined an AI startup in 2021 as the first employee. Both founders were already older than 30 back then. One of them was even 35+ gasp. We raised $40M+ since then. No OpenAI like valuation but by no means bad I’d say! Especially for a European startup.

Get yourself together and rise to the challenge!

Okay.

So I already have a somewhat AI powered project that based around music.

I have the prototype and videos of it in action on YouTube. How do I realistically pitch this. I'm actually just looking for feedback at this point.

https://en.wikipedia.org/wiki/Sam_Altman

> Samuel Harris Altman

> April 22, 1985 (age 38)

I think you missed the boat; if you have no high-falutin credentials you will be considered a bandwagon hopper chasing a gold rush, just like all the CRUD app hustlers creating a GPT wrapper.

You do. I promise. Think how radically different different situations can make you feel, especially long-term situations. Then quit your job. What’s the worst that could come from taking a child’s advice on the internet?!

Hot take: To get a turbocharged AI company funded these days, you need:

1) A team with the right pedigree (HYPS / MIT / Berkeley / etc. Ph.D -> OpenAI / FAANG ML/AI research group)

2) The right VC and industry network.

3) A product that will scale to hundreds of millions / billions of users.

There are plenty of ridiculously talented ML/AI researchers and engineers out there, that unfortunately lack the "correct" educational and professional background. They have cool product ideas, but lack the "billion user" scale.

Those are the startups that get hit with peanuts, or simply don't get funded. While the "pro athletes" of the ML/AI world get billion dollar valuations for some cookie-cutter LLM, before they have a product at hand.

Reply on news.ycombinator.com