Meta hires some super smart people. If all this token consumption by them does not lead to breakout technologies with massive usage growth, that’s an indictment of token-backed intelligence.
NEW: Meta employees consumed 73.7 trillion AI tokens in a single month. Which costs roughly $221 million a month and around $2.65 billion a year.
Many users slammed Meta's massive AI token spending as wasteful and ineffective at producing real progress, while a few noted modest recurring revenue as a redeeming factor.
No Digg Deeper questions have been answered for this story yet.
repeat after me metered effort not metered intelligence
Meta hires some super smart people. If all this token consumption by them does not lead to breakout technologies with massive usage growth, that’s an indictment of token-backed intelligence.
@deliprao It's very possible to do many more things while not doing the right things
@deliprao Imagine if they gave their people a month off and said - go do whatever you want just create value
@alth0u
@atelicinvest @deliprao You mean like how you’ve spent the past year with $3.5 in MRR to show for it ?
@deliprao nah this is like micro-services. There's a real reason to have them, but when management just mandates them and/or makes a metric to track them, engineers always go *way* overboard.
@deliprao Trillions of $s spent, 5 years, Quadrillions of tokens.. nothing novel found or created by the revolution.
Maybe a few trillion more will do the trick surely.
NEW: Meta employees consumed 73.7 trillion AI tokens in a single month. Which costs roughly $221 million a month and around $2.65 billion a year.
@deliprao Yep, Meta's failure shows you the technology is immature. A lot of the smartest people in the world have no idea how to monetize their tokens spend or build great products on the application layer.
@ebitdaddy90 @deliprao
@AndrewBean @deliprao What has that legion of employees been doing all these years? Facebook has no new functionally, and the quality of content and ads as gone downhill.
Everyone’s assuming they’re using these tokens to build some product.
What if is knowledge extraction from frontier models? And silently using it for fine-tuning their own models which they’re patiently taking a while to do.
They’re starting to make their own chips. They’ll get at least a decent model for internal use off their own RL approach from frontier model content, then in the future have very little dependency/spend on external models.
Also token cost is deductible via section 162 of the tax code, and if used for research then also under section 174.
@deliprao One of the wisest thing anyone said to me is education is not a magic bullet and intelligence is no guarantee of success. The odds are pretty bad actually...
@deliprao No, they’re literally just running their LLMs in a circle to token max
@deliprao Or they hire smart people who aren’t v entrepreneurial or creative , which has kind of always been their MO.
They couldn’t even clone Snapchat well the first time and Snapchat’s a p simple fucking product tbh.
@deliprao does high intelligence help or hurt when your main job is milking tokens for breakthroughs?
@deliprao smart people in a barrel of stupid can't do smart things
@deliprao They have some top folks, and they have some wheel spinners. Depends entirely on which group is using the tokens.
Hmmm... I also wonder why Google rate-limited their use of Gemini.
@deliprao @JagersbergKnut Meta has super smart people, but if they fail that still may not conclude on token based intelligence.
What if the failure says more about their incentives?