With all the uncertainty about AI regulation, I don't think now is the time.
How do you even value a company when we don't even know if GPT-6 will be made available to the general public?
It's over. Open models and chinese models will make fast progress and that nvidia+ms 128gb monster is what everyone will end up buying. sama can go back to running scams.
AI exits in America probably have a political cliff approaching fast as populist backlash will hit them, or perhaps they see political winds favorable to regulatory capture in the future and are waiting for that?
> AI exits in America probably have a political cliff approaching fast as populist backlash will hit them
The populist backlash is coming for datacenters. I'm unconvinced that's truly problematic to these companies given data travels close to the speed of light and plenty of countries have energy, data interconnects and governments unresponsive to locals' concerns.
Data centers are the next Dark Fiber from 2000. After VCs and private Equity fund them there will not be sufficient demand because AI will inevitably not entirely live up to all the hype. The fire sale will eventually begin. Then Google, M$, Apple and Amazon will buy them at a discount just like Google snatched up the dark fiber after 2000.
OpenAI is in deep trouble is what I am reading into this
Why?
Bad idea, the AI hype train still has some gas, when it settles in that it's just another tool it's all going to fizzle out.
I can’t wait. It’s gonna be an absolute blast to watch.
> up from the company’s last private valuation of $730 million
typo
For now.. ;)
[dead]
They just want to see Anthropic crash first and then be the last survivor.
If Anthropic tanks in the public markets, that will cause a revaluation of OpenAI in the private markets. If they delay IPO to try another private round, they also want to sign that round early.
Perhaps that's Anthropic's plan, is they believe OpenAI is weak. If the IPO is good they win. If it's bad OpenAI loses.
> The A.I. company’s advisers are pushing its chief executive, Sam Altman, to move slowly after SpaceX’s stock has been volatile and as the start-up grapples with financial challenges.
Surely if your company isn't just blowing smoke then you have nothing to worry about. Or is this an admission that the insane valuation for these companies is currently just bullshit?
> if your company isn't just blowing smoke then you have nothing to worry about
Not really. Plenty of solid companies have to wring their hands around IPO timing based on market conditions. Sometimes, this is due to valuation multiples. Sometimes it's due to fads, e.g. investors preferring capital-structure efficiency versus low leverage.
I mean, my comment wasn't necessarily meant to be some insightful analysis. But I do find it weird that OpenAI has seemingly gone from racing Anthropic to "maybe in 6 months" in the span of a week.
> OpenAI has seemingly gone from racing Anthropic to "maybe in 6 months" in the span of a week
When was the last time someone seriously asked if OpenAI was going to go public before Anthropic? For me, it's been at least months, maybe closer to a year. The corporate-governance complexity drove half of that, momentum the other half, and messaging from both companies having been consistent with that timeline for months sealed the deal.
I mean, they did the "confidential S1 filing" thing a few days apart.
Anthropic on 1st June: https://www.anthropic.com/news/confidential-draft-s1-sec
OpenAI shortly before June 8: https://openai.com/index/openai-submits-confidential-s-1/
That was less than a month ago. They seemed to be on a similar pace at least from my point of view.
I was really hoping that they Ipoed this year, so we can see their stock shoot up and down in flames, and we're really done with them and Sam Altman, once and for all.
After the IPO Altman will be even more insanely rich, and that time in a more liquid manner. I don’t think he will go away
While spcx has room to go up or down from where it is today, the reality is it that didn't drop like a rock on IPO day, so wall street bets vibes-based online "analysis" investing is only good for paper money.
The investment bankers were in there manipulating, but that's over, and gravity is here.
It is funny to mention “gravity” in reference to company that makes rockets that escape gravity
It might even elicit a mild chuckle when you realize nothing "escapes" gravity. You might be thinking of escape velocity?
The window has basically closed for them for the time being. The business math just isn’t there.
The best option at this point is kick the can down the road and hope market sentiment improves next year. Not much signal that it will, and quite a lot of signal the sentiment only declines, but pumping the brakes is the least worst option on the table.
I doubt that anyone at OpenAI would let their payday decrease. If anything, they got assurances that everyone would keep the bubble going until 2028 no matter what.
> window has basically closed for them for the time being. The business math just isn’t there
Unless Anthropic also cancels its IPO, this probably isn't it.
The math doesn’t help Anthropic either but the market views these two companies very differently at the moment. Anthropic is seen as having momentum. Open AI is seen as having likely peaked. That makes a huge difference when pitching an IPO.
> but the market views these two companies very differently at the moment. Anthropic is seen as having momentum. Open AI is seen as having likely peaked
What are you basing this on? Both are currently doing rounds/tenders that are placing without problems.
The media treats these two differently, as do financial influencers. But I'd be careful about conflating either of them with the market.
> But I'd be careful about conflating either of them with the market.
The finance market and the market for these products are two different things. Anthropic has definitely been stealing market share to OpenAI in the past few month on many segments (be it enterprise or even consumers).
it's not just media. or rather media is reporting based on fund interests.
> media is reporting based on fund interests
Can you give an example that shows funds actually souring on OpenAI? (Like, not less enthusiastic than before. Actually souring. Selling.)
ChatGPT is the 5th most visited website in the world and gets a ridiculous amount of user data. They're going to be an advertising powerhouse.
Useless stat without breakdown of time spent at the site and bot activity.
It will draw a non-zero number just curious people who never use ChatGPT.
And bots are exfiltrating model knowledge for the benefit of competition.
There are only so many ad dollars to go around. ChatGPT getting traffic doesn't mean they can convince ad buyers that their dollar goes further than at established players like Google and Meta. AI search is a commodity at this point, even DuckDuckGo has it.
Reddit has been one of the world's top websites for over a decade, yet they are totally irrelevant in terms of ad product market share.
I think a lot of money and effort is spent on advertising on reddit. It just goes to astroturfing instead of paying Reddit.
Because reddit users are low quality users. Meta and Google can peak into your data and segment you to sell.
ChatGPT ads are low quality placed at the bottom barely noticed.
ChatGPT ads will get better. Meta's social userbase is drying up. Google keeps introducing and removing things from chrome to keep access to spy on you for them alone. We'll see how things pan out but in ten more years reddit ads will still be worthless.
The advertising angle is significantly overrated. Theres only so many ad dollars to go around and so every dollar they make they have to take from so other player like Google. With Google having decent AI search now for free OpenAI is already well behind here.
Users coming from Chat or Claude or the likes would be very high intent if done correctly, advertisers would pay a lot of money. I'm certain people would pull budget from Google.
But Google has chat too now plus a ton more data about users OpenAI doesn’t have. There really isn’t anything OpenAI can do that Google isn’t also already doing.
Agree but Anthropic momentum is fading too.
Open source is starting to slowly become a source of frustration for frontier labs In the discussion around value for money.
Yeh local models will continue to gain performance before they can IPO
I only use free Gemini Pro to plan then scrape the log in Google Drive into local Qwen/Gemma+pi set up
I can plan and architect with Gemini on my phone or wherever and a cron job + custom JSON parser at home updates context in local model setup
How is momentum fading when their headline product is so good it’s illegal?
That is a good marketing headline, but for it to work the model has to become available again in a reasonable timeframe.
Otherwise people try other cheaper models, and they find out those models work perfectly for what they need.
I think they may have overplayed their hand so to speak. The end consequence is that their best model isn’t available right now, people are exploring alternatives, and realizing they work fine.
It’s such a fast paced and competitive industry, anyone who takes even a short break is going to have a hard time coming back from it, and that’s basically what they’ve done.
Thankfully, many tech companies have shortened parental leave to 6 weeks!
How do you grow your business when your flagship product is illegal?
easy - lobbying
Hasn’t worked for them so far
is no longer being able to sell it to half of your market good, financially?
Expected cash flows, growth and risk.
Go ahead and incorporate that in those 3 variables... lets see what you know before I bother replying.
Maybe they want a Mythos level model first.
Good news: GPT-5.6 has been export restricted.
This is patently false, don't spread rumors. Voluntarily delaying release at the request of the government is not the same as imposing export controls.
I hesitate to call anything "voluntary" when a competitor company was declared a domestic supply chain risk for refusing to do everything the administration requested.
By saying it’s false you are also spreading a rumor that OpenAI is goated by us defense. The reality is we don’t know the truth. But since we are spouting off rumors: Government could have given them a national security letter that says, “send all of your prompts and response data to a mirror run by NSA”
I would categorize the op as "technically false" but if they don't voluntarily delay, they would be export controlled.
Like being export controlled if you don't track everyone's identities is a form of imposing export restrictions. That is a true statement.
Where are the anecdotes about it hacking the NSA though?
It's not even out yet. Give it a second.
They need to do something new like create a more efficient battery formula out of recyclable abundant components. Hydrogen + Oxygen and Water are a good abundant recyclable example but also so is Gravity. Nightly / seasonal / long term energy storage needs another compact and efficient solution as elegant as these but better. (saying superconductors is cheating)
> The A.I. company’s advisers are pushing its chief executive, Sam Altman, to move slowly after SpaceX’s stock has been volatile and as the start-up grapples with financial challenges.
SpaceX's stock volatile? It's a shame nobody saw that coming.
Yes, it's actually the first volatile high-profile IPO so you can see why some people need to be reminded of the possibility.
Launched in the same way they launch Starship, full of ambition, promising a bit too much, but might explode at any moment. Either way it will be a spectacular show regardless of what happen.
> shame nobody saw that coming
Seeing something coming is very different from having it not only confirmed but also quantified.
It's actually remained about 14% or more above the IPO price which is roughly what you'd want but gone up and down a bit.
It's funny with stock prices - they all go up and down a bit in kind of random ways but people project all sorts of stories onto them that often don't relate much to reality.
It peaked at around +60% from IPO price and swung daily around 10-15%. It’s possible it’s starting to stabilize but that first week was basically the definition of volatile.
Just googling the ticker shows it at 149, which is below its opening.
Is it still being prematurely included in the major index funds?
Only the Nasdaq, which is an intentionally aggressive index. The S&P rejected all proposals.
yes, in few weeks.unfortunately the stock will be back from this slump
Ummm probably not. Lock ups are going to dump far more stock into the market.
But they are going to coincide lockups with the release of additional stock float from 5% up to 20% of the total "valuation" with a 3x QQQ multiplier so that stock indexes will treat them as 60% float even though 2/3rds of those shares are unavailable. Thus they guarantee that even more shares must be bought by tracking ETFs and institutional buyers. Everybody (that already owns pre-IPO shares) wins!
But that's not a secret, and therefore already priced in, right?
It's also a tiny effect given the total-market funds buy small amounts of each company, and the NASDAQ 100 isn't particularly big.
If S&P had changed its rules for the S&P 500, there would have been an effect. In the end, the drama was almost entirely a spectacle for finance influencers and their viewers.
Huh. I wonder if everyone breathlessly defending OAI and disparaging Ed Zitron on here a couple weeks ago is ready to admit they were wrong?
https://news.ycombinator.com/item?id=48550465
This comment is just great, starting off with:
“To be honest I almost think the numbers are irrelevant...”
Here’s another gem:
“My takeaway from this is that it's incredibly validating as a business model. Inference is _highly_ profitable...”
Thanks for the laughs. It’s a small compensation for the immense damage you’ve all done to the industry and more importantly the economy (which you will deny until the very end of the cycle like the cowards and frauds that you are).
You know, Ed actually did Scam a favor by leaking those numbers and saving him the embarrassment of filing an S1 (something Wario still hasn’t gotten up the nerve to do yet by the way).
... you think this is vindicating Ed Zitron? The dude is on a spree claiming the bubble will burst any time soon [1]. In fact Ed Zitron predicted that OpenAI will IPO sooner and not later [2]! This whole post is yet again another thing that he got wrong.
[1] https://www.youtube.com/results?search_query=ed+zitron+bubbl...
[2] https://www.wheresyoured.at/openai-cfo-news
> It's clear that both OpenAI and Anthropic are rushing toward a public offering so that their CEOs can cash out, and that their underlying economics are equal parts problematic and worrying.
I like that people will post stuff like “Ed Zitron is always wrong! Look at this wrong claim he made!” and then link to him not making that claim at all.
“Rushing toward a public offering so that their CEOs can cash out” is not a prediction of a specific time to IPO, and is supported by OpenAI’s own public statement two months after that was published
I don't even know what you are trying to say, I opened your link
> We have not decided on timing yet; it may be a while because there are things we want to do that are likely easier as a private company. But it’s a complicated set of tradeoffs and this gives us the option to go public sooner if that ends up being best.
So... OpenAI has specifically said that they have not decided on the timing and it may be a while. And now we have news that they are waiting till next year.
What do you think is supported by whom? Being more clear and concrete helps the discussion.
>it may be a while
Or sooner. It says sooner or later. It only means “later” if you don’t read the “sooner” part. Choosing to read selectively isn’t One Weird Trick To Own A Blogger. “This post about submitting a draft S1 to the SEC is proof that they don’t want to go public” lmao
> Choosing to read selectively isn’t One Weird Trick To Own A Blogger.
I mean it’s pretty obvious that the pro AI as an industry wide replacement people think it is One Weird Trick. I’ve still yet to see any of them articulate how they are going to reach profitability without resorting to the meme of projecting that their baby has doubled size in 6 months so is projected to reach 10.5 trillion pounds by the age of 10.
OAI canceling an IPO this year a week after he released their dogshit financials is not a coincidence and yes it does vindicate him.
He’s not shorting the market or calling a top. He’s saying that the bubble will pop because the underlying business model is and always will have a NEGATIVE ROI. Unless you’re speculating on semiconductor stocks the difference is irrelevant.
Do me a favor and tell me how much of the 1,000,000,000,000 spent / committed to a datacenter buildout has been returned to shareholders / creditors?
1. they didn't cancel their IPO and they were deliberate about having the option to time their IPO
2. he has tried over and over again to predict the bubble and peak [1] [2] [3]
3. that OpenAI is filing for an IPO next year is no vindication of Ed's claim when he specifically predicted the opposite (as I showed in the above comment)
4. OpenAI filing for an IPO next year has no bearing on its fundamentals
5. on Datacenters: Anthropic had to lease it from Elon's datacenter because they were too short on capacity and every one was complaining that their limits were too low
[1] Ed on 2024, "threaten to begin a collapse that I’ve been predicting since March" https://www.wheresyoured.at/burst-damage/
[2] Ed on 2024, "three quarters to prove itself before the apocalypse comes" https://www.wheresyoured.at/peakai/
[3] Ed on 2024, "things are beginning to collapse" https://www.wheresyoured.at/subprimeai/
Edit: the quality of discussion in this website is annoying sometimes.get downvoted for good faith discussion
> Edit: the quality of discussion in this website is annoying sometimes.get downvoted for good faith discussion
Ignore it and don't do this: "Please don't comment about the voting on comments. It never does any good, and it makes boring reading."
> OAI canceling an IPO this year a week after he released their dogshit financials
There is zero evidence of any causal link between him and this. The obvious one, instead, is SpaceX's volatility.
> Do me a favor and tell me how much of the 1,000,000,000,000 spent / committed to a datacenter buildout has been returned to shareholders / investors?
If Anthropic also delays its IPO, you'll have a point.
> There is zero evidence of any causal link between him and this.
How would there be? He's a blogger and a youtuber, they are a private company with secretive financials, bankers involved in pre-IPO work don't talk openly. Nobody is going to say "we only decided not to do this because of a youtuber" because that would make them look like the ill-informed, over-eager idiots they've been to let this nonsense get this far.
Why would there be any concrete evidence it was down to him specifically, and not, say, dozens of finance people saying "what the *fuck* is that marketing budget about — that's so large it looks like something's been hidden in it" after reading about it from him and the FT and everyone else who was involved.
> bankers involved in pre-IPO work don't talk openly
We're an obnoxiously chatty bunch.
SpaceX demonstrated that the public markets have a limited tolerance for a multi-trillion dollar company that doesn’t make any money.
Ed’s leaks demonstrated that OAI doesn’t make money (even on inference).
Put these two together and I think the conclusion is pretty obvious.
> SpaceX demonstrated that the public markets have a limited tolerance for a multi-trillion dollar company that doesn’t make any money
What? How? SpaceX loses oodles of money. It's trading above its IPO, and just filled an oversubscribed bond deal.
> Put these two together and I think the conclusion is pretty obvious
Zitron has a faithful following. He isn't a broadly-influential analyst.
You said: “ There is zero evidence of any causal link between him and this. The obvious one, instead, is SpaceX's volatility.”
What exactly do you think “volatility” means in this context?
> Ed’s leaks demonstrated that OAI doesn’t make money (even on inference).
They did not.
Oh? They’re profitable now? Or anywhere close to it?
No, but the GP wasn't satisfied with that, and had to put in a snide "even on inference" parenthetical. The leaks showed inference having positive margins.
The Zitronites will say that the data is fraudulent, and OpenAI must have classified some of their inference as marketing, or R&D, or some other wacky theory of the week. But the actual data does not show that. It is made up.
If you want to cherry-pick the worst parts from the leak and disbelieve the more positive ones, it feels like you're not in a great place epistemically...
How do you explain OAI spending $6B on “sales and marketing” in a year. More than Coca Cola?
I think it’s reasonable to draw the conclusion that they are folding inference subsidies (for both paying and non-paying users) under this category. Frankly I think occam’s razor demands it because where else would all that money have gone? Fancy trips for enterprise clients?
> Zitronites will say that the data is fraudulent, and OpenAI must have classified some of their inference as marketing, or R&D, or some other wacky theory of the week
Which, look, could be true! But it's currently speculation only.
> The leaks showed inference having positive margins.
They don't. They show that OpenAI need to people to draw that conclusion, because of course they do.
> The Zitronites will say that the data is fraudulent, and OpenAI must have classified some of their inference as marketing, or R&D, or some other wacky theory of the week. But the actual data does not show that.
It doesn't?
It shows a marketing budget so absolutely mahoosive that it's almost completely implausible, which does make you think — have a percentage of marketing-driven free plan tokens been hidden in there? If not, what the hell is in there? Because it's an insane figure for a company that has benefited from a level of word of mouth that makes "ChatGPT" broadly synonymous with "AI".
Fraudulent is a big claim, of course. I didn't say it.
The inference part requires training the models to keep up to date as far as I am aware(I am open to being convinced otherwise if you have data showing so). Everything I’ve read about it implies that within a few years of reality updating or even just software if we constrain it to models for coding, won’t be able to handle all of the updates within the context windows available.
Given that I don’t know how you(royal you, not you specifically) can constrain an analysis of inference profitability to just the literal running of inference and not include training costs. That’s where it all breaks down for profitability.
In what way is SpaceX's volatility an obvious cause? It would be one thing if SpaceX was down from its IPO price, but it's not, it's just down from a post-IPO peak. To me this has all the hallmarks of a backfilled rationalization.
> OpenAI’s advisers presented company executives with the option of waiting until 2027 to go public with a $1 trillion valuation, or lower the targeted valuation for a quicker I.P.O. Mr. Altman, said one person in contact with him on the topic, responded that any change to the trillion-dollar valuation was a nonstarter.
I really don't know how to read this and reach any conclusion other than, OpenAI leadership won't accept what financial analysts consider to be a rational valuation of its stock.
> In what way is SpaceX's volatility an obvious cause?
Let me amend: it's a more-obvious cause given it's pertinent new information in a way Zitron partially leaking financials many institutional investors have already seen is not.
> don't know how to read this and reach any conclusion other than, OpenAI leadership won't accept what financial analysts consider to be a rational valuation of its stock
Neither did SpaceX and, as you say, it's trading above its IPO price and placing tens of billions of dollars of debt.
I think Zitron's analysis was on the balance good, though it didn't say a lot of what folks on here seem to have taken away (e.g., about OpenAI's inference being marginally unprofitable). It seems he's got a bit of a cult of personality around him, which makes me inherently sceptical. But it's a pretty ridiculous reach to claim OpenAI had to delay its IPO because of him versus the much-more visible and talked about thing.
Perhaps we're really on the same page. I don't think OpenAI executives read the Zitron article and said "oh my god now we can't IPO!"; I think they're both downstream of the underlying bad financials, which SpaceX only managed to mitigate due to the Elon Musk personality trade. (And I agree that this theory falsifiably predicts Anthropic will also find a reason to delay.)
> they're both downstream of the underlying bad financials, which SpaceX only managed to mitigate due to the Elon Musk personality trade
Idk, I'm still sceptical how someone could look at the market right now and conclude that it's suddenly hyperaware of financial metrics.
For whatever reason–maybe it's corporate-structure complexity, maybe it's lawsuits–OpenAI was always at the end of the pack of AI IPOs. If SK Hynix take August and Anthropic September or October, that would mean a 2026 OpenAI IPO would have to (a) coincide with one of those or (b) go to market during an election/post-election fiasco and/or the holidays. The realistic options were July or October, the latter being between a likely Anthropic IPO and the midterms. The timing just doesn't make sense and maybe someone realise that.
OpenAI did confidentially file their S-1, which costs a ton of money to put together for the bankers and regulators to review. They did test the market and it looks like either the banks or people directly around Altman told him not to move forward. That doesn't mean Zitron was wrong about OpenAI IPOing. They took steps in that direction and then decided not to move forward. That's not his fault.
As far as the bubble bursting soon, we are starting to see some pretty concerning signals. The South Korean stock market triggered trading circuit breakers twice earlier this week to stop a runaway selloff in the tech sector. For reference, circuit breakers have only been triggered in the Korean market 10 times in history, and only 5 times ever in the US markets.
https://finance.yahoo.com/markets/world-indices/articles/kos...
FWIW I think there's a "Y2K never happened" future here, where the bubble never bursts (in the sense of some insane market valuation proving to be lunatic) but everyone does what they can to make sure it doesn't burst on them, and they pull back and the bubble just deflates.
Take for example the action on SPACs that made the "SPAC everything" era end.
Maybe, it's certainly a possibility. I'm skeptical only because of the sheer amount of spending that Anthropic/OpenAI etc. are committed to. A soft deflation only makes sense to me if they start making a more profit-driven motion as opposed to growth at all costs, which I don't see happening.
Zitron's reporting shows that OpenAI's losses are scaling linearly with their growth, and their long-term valuation is based solely on the idea that they will eventually reverse this trend. Anthropic/OpenAI pumping the brakes on growth mode could have ripple effects throughout the entire market.
I think the more likely outcome is that these companies keep burning and demanding cash until investors call BS.
I'm also feeling like that, not a quick burst like dot-com but a deflating bubble over next 24 months. During which AI costs rise and IPO valuation comes down.
Maybe my memory of dot-com is fuzzy (likely) and perhaps I didn't see early warning (or even didn't know how to see it). I feel like it all transpired over 3 months. And for those of us in Seattle the "death-knell" was the Nisqually quake that made much of Pioneer Square unsafe to occupy for a while.
One of the things that really strikes me is that there are so many people talking about how great the frontier models are but there's nowhere near as many people talking about open weights and local models.
There is this huge knowledge gap here about what the local models could do in terms of consumer queries with a tiny bit of agentic support. If more people could tinker with local models and see what they can do, I think there would be far less belief that only the big two/three hold the keys to the kingdom and far more that the future is a bit more distributed.
On that basis I am always on the lookout for coverage that gets into this — Cal Newport touches on it a bit — because I think it is one of the deflationary factors, as is the quality of open weights models, which in turn is why Anthropic is now getting credulous journalists to write about "distillation attacks".
Re: the dot-com crash, I was in the UK working for an internet startup at the time though I was on a sort of health break/sabbatical at the time of the pets.com crash, so I was reading and researching, and while I do remember the coverage of their valuation being insane, and agreeing with that, I also think it did not have such a long run-up. We did know in 1998 and 1999 that Broadvision's valuation seemed absolutely insane, for example, and that some e-commerce developer and design startups were radically overvalued, but what is so different about this time is how much more broadly things were distributed then. So many players. Even the major duality (which was Microsoft vs ABM — a loose coalition of "Anyone But Microsoft") had many more players.
Whereas now… it feels like a prequel to Rollerball.
> If more people could tinker with local models and see what they can do, I think there would be far less belief that only the big two/three hold the keys to the kingdom and far more that the future is a bit more distributed
Eh, I've played with them. They take way more babysitting to do the work reliably, and I have to much-more closely monitor the outputs for hallucinations.
They'll get there. But it might take a decade or so for (a) consumer hardware and (b) SOTA distilled open-source models to converge with the closed-source stuff, and that's enough time for both a slow deflation and plenty of profits to be made.
I was specifically referring to consumer AI, not agentic coding, and I really think Apple have a good chance of demonstrating smaller on-device LLMs plus traditional symbolic AI work (even just rules systems like evolved forms of Automator and Shortcuts in harnesses) can do much of what consumers want.
The current industry obsession with agentic coding is a giant red herring, in a way. Or at best a shiny thing. The pitch from OpenAI for their extraordinary valuation has always been absolutely all-encompassing in terms of utility and market. But we've stopped talking about that.
I don’t think these vast LLMs will be the dominant approach in a decade, frankly, because the performance gap between a 35B MoE and these absolutely gigantic models just is not proportional to the scale difference, but at this point we are just pitching belief against belief.