416 points by surprisetalk 1 day ago | 782 comments | View on ycombinator
throwa356262 1 day ago |
blfr 1 day ago |
Meanwhile Claude is getting really good. It's a decent harness now, even the remote control over a couple servers is quite OK. They're adding artifacts left and right, new features. And all without making the product seem bloated, they stay focused.
The prediction markets don't show anything on the horizon either: https://releaseoracle.xyz/
I am not a big AGI believer but business- and product-wise it seems like ANT has something that others are missing.
purple_ferret 1 day ago |
6thbit 1 day ago |
The offering consists of 980,000,000 shares of Class A common stock
That 980M is more than they had previously offered (825M), so they did expand the IPO.And they're taking around ~105B in cash.
81.34B from the 980M shares at $83 and 26B from Google/Nvidia/etc, minus 1.9B the existing holders selling 23.6M shares.
ozozozd 1 day ago |
Be careful out there.
wbsun 1 day ago |
testfoobar 1 day ago |
jfrbfbreudh 1 day ago |
(I will be buying shares.)
plaidfuji about 20 hours ago |
The recent cybersecurity incidents will only increase that perception. It appears to be weaponizable, whether or not that perception is grounded in any kind of reality. Whether it exposes vulns, fixes them, or all of the above, it just translates to more demand.
I think this will go up for quite a while before the world realizes the limitations. At least until their first earnings report. That said, I’m not a gambler.
enahs-sf 1 day ago |
consumer451 about 23 hours ago |
timmg 1 day ago |
Like 1.5T could make as many as 1,500 billionaires. It's certainly much more concentrated than that. But it is still insane to think about.
rdwrrr about 23 hours ago |
JKCalhoun about 20 hours ago |
rubyn00bie 1 day ago |
That’s not to say there aren’t a lot of winners even 1-2 years later but I’ll happily be sitting this one out (and wait for a drop or two). The market gives some pretty sweet deals for those who are patient.
mococa about 21 hours ago |
wmf 1 day ago |
purpleidea 1 day ago |
As a result, I think this stock (which has no voting right btw) is a major gamble and unlikely to pay off before they run out of money.
More importantly, I think Dario is dishonest, and trying to thwart open models which are soon going to be just as good as his models, but without the spying on what you're doing.
seanjhkwon about 18 hours ago |
Beijinger about 18 hours ago |
It would not really be a problem if Anthropic is gone tomorrow.
thibsrn 1 day ago |
Retail investors will be left holding the bag. This is the ultimate cuck tranche.
senordevnyc about 17 hours ago |
Anthropic has raised about $141B.
But their S-1 also says they have $132B in cash on the books. So they've basically not spent hardly any of what they've raised.
Yes, they have huge spending obligations over the coming decade, but I don't understand how this fits the narrative that they're massively unprofitable and just shoveling investor cash into a furnace to stay afloat.
nemo44x about 19 hours ago |
Lots of opportunity to make money with this until then, but it’s coming.
dominotw about 19 hours ago |
shevy-java about 23 hours ago |
I think in a few years many of these AI companies will no longer exist; and the survivors will be bigger than they are as of right now.
rdiddly 1 day ago |
elbistec24 about 18 hours ago |
jonathanstrange 1 day ago |
Sorry but if a company insists over months that their products are unsafe and even hint they could lead to the destruction of humanity, then I don't understand why authorities don't act and stop the company until they can demonstrate their products are safe.
torlok 1 day ago |
gonzalohm 1 day ago |
nicce 1 day ago |
justauserid about 24 hours ago |
Because if that doesn't happen then I have a feeling the world will proceed something like this: AIs paying for things to other AIs (as a circular gesture; possibly different from the circular finance of today) while some rich folks will sit in their safe island citadels or in NZ (etc) nuclear bunkers controlling the arms, ammunition, surveillance, medicine, finances, along with pretty much everything else of the world while the rest of humanity, the majority – who couldn't tap into that abundance, will finish each other off one by one – or maybe worse, they will be live to witness it all. I try really hard not to picture this post apocalyptic scenario but either all this IPO, moat, and whatnot are hogwash, or that is happening :)
mikewarot about 21 hours ago |
moneycantbuy 1 day ago |
apexalpha 1 day ago |
Companies used to IPO when small and needing money for growth.
jhallenworld about 17 hours ago |
Netflix back in 2010: you look around, everyone was using it.
Anthropic today: look around..
hmokiguess about 19 hours ago |
troglodytetrain about 11 hours ago |
Codex is completely unusable at all for anything but small tasks when handheld by a person. Orchestration is simply unusable. Anthropic falls apart and refuses a critical set of things that are small enough to block you from getting it to build out real work. Like basic ci/cd server setups and using login creds provided on web, etc. Conveniently setup in a way that doesn't at all prevent a hacker group from at scale pwning people/corps but does prevent normal workflows. Want it to build a full stealth toolset for control of android phone over adb bridge? No problem. Now can you use it for anything? No I won't log in for you. Pressing the fucking login button is just something I'm morally against. etc. Basic room temp IQ computer tasks. Like they are trying to prevent someone from automating your average corporate workflow while it can do everything else trivially. Why? Because 80% of their revenue is from Corps letting employees burn $$$ tokens for every day tasks like proofreading an email.
I know this for a fact because of custom harness work in a corp with Microsoft EntraID SSO over a year ago where all of these tasks were easily performed by the old Sonnet model and I trivially automated every workflow.
And, strangely, when I do something that should make everything much worse, simply just tell Claude to use Codex as sub agents, performance gets drastically better, presumably because teams from both corps didn't have the chance to specifically fine tune their models to be hamstrung when run in this configuration.
It is obviously not about safety and has never been. These models still will easily find vulnerabilities. Just get some code you want to exploit and throw $$$$ inference at it to 'find all bugs confirmed via test case suite' etc. Then throw inference at targeted exploit scripts.
1-more about 17 hours ago |
liquid_oil 1 day ago |
tt_dev 1 day ago |
hollerith about 21 hours ago |
DeathArrow 1 day ago |
Some people will be left without a part of their hard earned money.
Danox 1 day ago |
Varelion 1 day ago |
anthonyc 1 day ago |
Farooqiainews about 14 hours ago |
jlk34t about 18 hours ago |
jlk34t about 18 hours ago |
logifail 1 day ago |
jlk34t about 18 hours ago |
jlk34t about 18 hours ago |
jlk34t about 18 hours ago |
sick_of_slop 1 day ago |
thesurlydev 1 day ago |
simianwords 1 day ago |
Anthropic and OpenAI are the AltaVista of the AI era. With first starter advantage but no real moat and very good at spending money.
I am still waiting for the Google of this era to show its face.
Edit: every time I question these crazy valuations I get down voted really quickly.