Plenty was written on their "transition" in the past.
Long story short, the board briefly fired the CEO Sam Altman because they were starting to eye that kill-switch and they didn't have confidence that his leadership aligned with the goals of the non-profit's mission. A few days later OpenAI and its employees decided that no, actually they like the money that Altman seems to bring in and they hired him back. If one were being charitable maybe there's an argument that the money is important to implement the ideals, or maybe now that there's a competitive LLM marketplace bowing out wouldn't accomplish anything. Regardless, now it's just a bog standard Valley company which is partially why they can IPO. So it goes.
I could be distorting some things. I'm only passingly familiar with the details of the matter. Feel free to correct me if this seems substantially inaccurate. Cunningham's Law and all that. ;-)
Perhaps both your rendition and the MSFT thing are sympatico but it's confusing to me at the moment.
Having two wrong answers surely will kick Cunningham's Law into overdrive here... one can hope, anyway.
Like imagine if the dogs trust just got a judge to let the management turn it private. Think of how much money they could steal.
Strictly speaking? No. But the people who shoveled money into it want an offramp, and that's what the IPO is. So for all intents and purposes, yes they need to IPO.
You think the circumstances still suggest they have to do it?
Yes. They need to find a bagholder before the bubble pops.
Employees would want to cash out.
If there is no exit in horizon, employees look for better ventures, and new funding may be hard to come. New funding (debt or equity) will be needed if cash flow is negative.
> Employees would want to cash out.
Oh where are the times when you invested because you expected dividends and employees were being paid out for performance in yearly bonuses...
Not if the dividend is a return-of-capital distribution.
They aren't. They depend on investment infusions to continue to function because they don't have a product that they can sell profitably.
Normally, this is seen as an impediment to gaining future investment. Rational people still think that way.
It’s uncharted territory to have companies of this size unable to produce positive cash flow and seeking continual infusions of cash just to avoid going bankrupt.
No? What comes after a pump is why
/s
But that rarely (never?) happens, especially for high-growth companies like this because they're investing their revenues in growth.
I'd say HN is split on whether ANT's growth will lead to massive profits in the long run.
A lot of folks point to Amazon and say “but they weren’t profitable for ages and look at them!” But the difference is during that time Amazon focused religiously on free cash flow. It was printing money during the growth years and just choosing to reinvest it. AI companies are black holes for cash and they keep having to raise cash from the outside, which is far riskier proposition.
Earlier this summer, I asked ChatGPT if it could justify the ~$1 trillion valuations for OpenAI and Anthropic.
Its response was that maybe Anthropic could grow enough to eventually justify $1T, but it was highly skeptical of OpenAI.
Take that with a massive grain of salt, but I thought it was interesting.
I'm not here to defend companies running a loss and asking for investors/handouts, that's for sure. It is worth noting, though, that its common for companies to raise funding before turning a profit.
Remember recent headlines that OpenAI was struggling to raise cash at valuations above where ANT is now trading.
Net here it’s is a good day for those looking to cash out of ANT and a very worrying signal for other elsewhere that were hoping to cash out too.
The reality is that most consumer AI is going to be on devices, and Apple managed to buy theirs for literally $1B per year (plus say a couple of billion for the private cloud data centre rent), from Google.
Apple Intelligence and Google's Android AI offerings need to be only incrementally better over time for ChatGPT to end up struggling and enshittified with ads, and both Apple and Google have the advantage of being able to design the device around the product and roll out deeply embedded features.
OpenAI has a slight advantage in the EU, because there they are an app consumers can install, and Apple is still hamstrung by legislation. But European sentiment towards US AI companies is not exactly positive!
Might be the last hurrah for tech IPOs for a while.
For comparison, Google's IPO raised $1.6 billion.
And that just so happens to be the rough prediction of their Q4 2026 operating loss.
Adjusted for inflation, Anthropic raised 28 times more than Google.
It's a logical fallacy to suggest it's silly to compare them.
Anthropic will burn through the money they raised by selling between 5 and 7.5% of the company in twelve months — on their immediate commitments for next year and their projected losses for Q4 2026 alone.
They are loss-making, they have hundreds of billions of dollars of future commitments.
Google was profitable at IPO and had been for at least the previous year, they raised more than 3.5 times their yearly operating costs, or roughly about three times their total outstanding commitments at that point according to their S-1.
Google was in a healthy position. Anthropic is in a historically huge debt and obligation hole.
If this is the end of AI trade (a ludicrous idea for me), then Anthropic absolutely didn’t make it.
Yesterday you could explain away the sell-off by saying people were selling stocks to buy ANT. Not sure what the explanation can be today when ANT is falling and has twice got within touching distance of its IPO price.
Re: a small number of their shares: worth bearing in mind that Amazon and Google between them own 28% of Anthropic.
It looks to me like the market thinks the music is about to stop.
That said I'm in the camp of "how does it matter?"
2) Many companies have this situation where there will likely be future shares issued. But the math on market cap is simple… number of current shares times price. Thats how it works.
This nonsense of “but but if you include this other thing and then assume X it’s higher!” is just made up nonsense math said by folks that wanted the number to be higher. You could do the same funky math for most companies but again that’s just not how “market cap” works.
If you're a buyer, you'd think people who get issued those options will hold them instead of selling. And you'll also assume that for the options that do get sold, there will be buyers at the same price you bought, if not more.
The current market cap of the company in so far as the collective value of all actual shares is around ~$1.27T. Thats the generally accepted value of the company now that its publicly traded and yes that is much lower than they were reportedly originally hoping for.
It’s more of a problem for OpenAI though as that’s lower than where OpenAI is trying to value itself now.
The chart of the stock price is so ridiculous, it has a straight 45 degree-ish angle line from a fake “prev-close” for some number from “yesterday”, all the way up to the opening price, and right after that, it drops, then flat but sliding
And so, even though it’s been steadily dropping since open, because there’s a much lower “previous close”, then it’s still shown as a big gain for today (+17.7% !)
The steadily decline once trading started isn’t great but it will take a week or two for a realistic price to settle. Theres so few shares actually trading right now that it’s just the same shares flying around in circles at the moment.
Trading Point Fully Diluted Market Cap Opening $2.022T $1.474T Intraday Peak $2.088T $1.522T Closing $1.762T $1.285T
To be clear, market cap and full value closed DOWN on the day. However both figures remain comfortably within the desired and forecasted range. Most analysis discussed the fully diluted figure prior to IPO. There is a large and unusual gap between diluted and market cap.
And with this now below where OpenAI has been trying to value itself on the private markets that doesn’t bode well for OpenAI.