Post
> The cost of compute is far beyond the costs of the employees: Nvidia exec
https://fortune.com/2026/04/28/nvidia-executive-cost-of-ai-is-greater-than-cost-of-employees/
If your AI use costs less than what it would take to pay a human to do it, then you're what's called a "mark" in the "pre rug-pull" phase of the grift
When they start passing those costs on to you, you will graduate to being the "bag holder"
So think hard about whether you want any "load bearing" tasks to be handled by an AI
@researchfairy "Cost of compute" is really hiding the ball here. The money they pay for "compute" is still going into human pockets. Just different (wealthier) humans.
@researchfairy
...
But if you ask for a rise
It's no surprise that they're giving none away
Away, away, away
Away, away, away
...
Pink Floyd
Money (1973)
@researchfairy And don’t forget that the employees we‘re talking about are at nVidia - making three to four times more than the industry median…
Not a good look for Klarna-type „efficiencies“ 🫠
@researchfairy but if you are rich enough you can still replace worker, who you hate, with AI, that you love.
@researchfairy Love this part too:
"While AI may cost more than human labor today, there will be warning signs of a tipping point toward AI’s economic viability. For one, Lee indicated, the cost of using AI will become significantly lower, with performing inference—how AI analyzes data—for a large language model with 1 trillion parameters plummeting by more than 90% over the next four years, according to a report last month from analyst firm Gartner."
Based on what? How are inference costs as part of total cost changing over the years? Not just the cost to infer one thing with different hardware (which also costs money to replace), but as the actual cost of overall use?
4 years ago, I bet the predictions about AI didn't have a damn thing to do with reality. So 4 years from now, assuming they can even build and power all this, is it going to be anything like what we see now? Lol no. Foh.
@researchfairy first AI couldn’t show incremental growth or productivity so it was about needing less headcount to save money and now it doesn’t even save money or headcount and in fact costs more but still doesn’t show any additional revenue resulting from its adoption
@researchfairy This is an incredible article. Big tech has lost its goddamn mind.
Feels like the type of stupid that we saw once people began to unpack the specifics of how the finance industry wrecked us all in '08.
And I'm sure they're not even thinking about externalized costs that we all have to bear, like the environmental impact, or the psychological damage to the content moderators or the destruction of democracy itself
I think this cost analysis is probably best-case for AI.
In addition to what you mentioned:
It doesn't take into account that what they're paying for AI is well below even a break-even price, sustained by a bubble that will eventually pop. Every company paying to us AI is getting a massive discount that cannot be sustained for long.
It doesn't take into account the same sort of downstream costs outsourcing has, what happens when you no longer have local talent or domain knowledge.
It doesn't take into account the currently known issues with AI, much less the costs to fix those mistakes down the line. In addition to hiring actual people to go through the mess and fix mistakes there seems to be a widespread belief that companies can avoid any form of liability by pointing to "the AI did it" which is optimistic at best.
It doesn't take into account the massive amount of power it's drawing, the hardware it's eating up, and the effects THAT has on the economy.
It doesn't take into account the economic costs of unemployment. It doesn't take into account the fact that Ai is basically replacing junior talent which means when all these costs come due, and it takes people to fix them -- they're not going to have them and so will have to pay a hefty premium.
This very much feels like all these tech bros looked at the 90s tech bubble, the 2008 mortgage bubble and the the worst examples of outsourcing and went "What if we did all these at once, but like 10 times as hard?"
And every CEO went "We don't have to pay for labor? HELL YEAH. END OF DISCUSSION"
@researchfairy Me, thinking about Cloudflare firing 10 or 20% of their staff: oh this is going to be not good.
Install bonfire.cafe
Get the full app experience
