Share.

    14 Kommentare

    1. That really doesnt feel like all that much with the numbers that are being thrown around.

      Sounds equivalent to me turning off the lights when I go to bed.

    2. RobotIcHead on

      Going to use a work buzzword to describe this: that sounds like scalability problem.

      Basically it means that the product team was happy with functionality but actually running the platform/ product is impossibly expensive. It is also sounds lie if they added more users and projects to the platform the costs would escalate.

      It was sadly a common problem in an old job, overdesigned software that is unsellable to customers. It prompted a saying in the company: you don’t need a cluster of Kafka running to crack a nut.

    3. RoyalCities on

      That is too expensive. What were they running inference on?

      The open source community has optimized these models down so much that you can get Sora quality at home.

      https://www.reddit.com/r/StableDiffusion/s/0T80eHUdyW

      I feel like they saw the writing on the wall here. Their bread and butter seems to be chat / coding but generative video / audio has been largely solved at home / on prem and any business will probably eventually go local deployment as they realize they can save way more and also control the whole pipeline internally (and even gen video is a niche within a niche from a business perspective – meme machines aren’t really something corps rush for)

      The local chat and coding models are also getting there – I haven’t seen anything as good as say Claude but outside of really bespoke software or large codebases I think that will also eventually be moved to in house as well.

      How are they burning so much cash? Serving APIs for tech that’s getting smaller and easier to run isn’t a long term healthy business plan.

    4. monospaceman on

      That’s it? People were generating an astounding amount of slop on there. I would have expected this to be much much higher

    5. These AI companies have no path to the black. They will always operate with a deficit. Hence the bubble that will eventually bust. None of these companies will be considered too big to fail. They will all eventually collapse or get consolidated and become state owned.

    6. It also helped raise their valuation by over $200 billion in the months leading up to a $110 billion round of funding which is what the VCs care about over revenue or losses, at least for now.

    7. Then how much does it cost xai. I think grok is the most used video generator other than local ones

    8. iaNCURdehunedoara on

      It’s surprisingly low at the rate they’re plowing through cash.

    9. One estimate I saw was closer to $15 million a day, that’s why they shut it down. Just a huge money incinerator

    Leave A Reply