Gsus4@mander.xyz to World News@lemmy.worldEnglish · 2 months ago‘The cost of compute is far beyond the costs of the employees’: Nvidia exec says right now AI is more expensive than paying human workersfortune.comexternal-linkmessage-square108linkfedilinkarrow-up1728arrow-down16cross-posted to: [email protected]
arrow-up1722arrow-down1external-link‘The cost of compute is far beyond the costs of the employees’: Nvidia exec says right now AI is more expensive than paying human workersfortune.comGsus4@mander.xyz to World News@lemmy.worldEnglish · 2 months agomessage-square108linkfedilinkcross-posted to: [email protected]
minus-squareKnock_Knock_Lemmy_In@lemmy.worldlinkfedilinkEnglisharrow-up2·2 months ago AI is strictly more expensive the more users there are since the cost is per use. Isn’t most of the cost in the model training?
minus-squareporous_grey_matter@lemmy.mllinkfedilinkEnglisharrow-up3·2 months agoThe cost of inference has passed the cost of training quite a while ago and it’s by far the majority of expenses
minus-squareKnock_Knock_Lemmy_In@lemmy.worldlinkfedilinkEnglisharrow-up1·2 months agoThanks for the link, but disagree with your summary. The Spend on inference has passed the Spend on training. Training is the monopoly product and even the cheapest known model (deepseek) cost 10s of millions to train.
minus-squareporous_grey_matter@lemmy.mllinkfedilinkEnglisharrow-up1·2 months agoThat’s got nothing to do with how expensive it is to serve models to customers which is what was being discussed.
minus-squareKnock_Knock_Lemmy_In@lemmy.worldlinkfedilinkEnglisharrow-up1·2 months agoWe are discussing cost. There is fixed cost (training) and variable cost (inference). Both components being discussed.
Isn’t most of the cost in the model training?
The cost of inference has passed the cost of training quite a while ago and it’s by far the majority of expenses
Thanks for the link, but disagree with your summary.
The Spend on inference has passed the Spend on training.
Training is the monopoly product and even the cheapest known model (deepseek) cost 10s of millions to train.
That’s got nothing to do with how expensive it is to serve models to customers which is what was being discussed.
We are discussing cost. There is fixed cost (training) and variable cost (inference). Both components being discussed.