Anthropic is set to begin trading on the Nasdaq on Thursday after pricing its IPO at $83 per share, valuing the artificial intelligence giant at $1.5 trillion and surpassing SpaceX as the largest initial public offering on record.

The higher-than-anticipated pricing signals broad investor enthusiasm for AI stocks even amid some market uncertainty stemming from the ouster of Nvidia’s CEO and worries about an AI bubble. But the bigger test will be how Anthropic fares on the stock market.

Anthropic said the offering consists of 980 million shares, roughly 956 million sold by the company and the rest by selling stockholders. Proceeds totaling $81 billion put Anthropic’s IPO ahead of SpaceX, which raised about $75 billion in June. SpaceX, however, holds a higher valuation, about $1.9 trillion as of Wednesday afternoon.

you are viewing a single comment's thread
view the rest of the comments
[–] 9 points 2 days ago (4 children)

I just still don't understand how the math is mathing.

Even if they decide to start trying to hook companies on this and ratchet up prices, how many companies will go along with that? With things like openrouter.ai and local models that you can set up and are probably "good enough" for an awful lot of corporate uses...I don't see anything else but the prices per 1M token going down, even for "frontier" models, since it is not like any one company has a lock on this.

Unless there is an expectation of a huge drop on their operating expenses?

  • source
  • parent
  • hideshow 4 child comments
  • [–] 3 points 2 days ago

    Yep. If they took enough money to be profitable, they'd probably lose 90+% of their customers, and the rest will scale back big time.

    The only chance they have is to hook as many people and companies to their "services" and make them so dependent on it that they'll have to eat the bills. Basically the digital crystal meth dealers.

  • source
  • parent
  • [–] 1 point 2 days ago (1 child)

    Their higher models are a lot better, it's like a different ball game

  • source
  • parent
  • hideshow 1 child comment
  • [–] 4 points 2 days ago

    I have to use Anthropic at work, but also keep an eye on other ones and try things out on what I can run locally (on a laptop). I've used Fable, but honestly, for a lot of tasks, Opus or Sonnet are just fine for many tasks.

    My point is....do all consumers even need some of these higher models? How did they get by with the other models before? And even if they do need them, models from other routes are not far behind, maybe months in some cases, maybe as much as 2 years in others.

    I just don't see how it works out for these guys. Americans would have to spend ~9% on AI - which is roughly the same as food - in order for these guys to be profitable. And this stuff should only become cheaper with time, unless the expectation is that everyone is going to pay for the most expensive model for the most mundane of tasks, and Anthropic, et al, can continue to charge what they are charging now, or higher for those upper end models.

  • source
  • parent