Tesla's reported internal memo introduces a $200 weekly ceiling on third-party AI tool usage starting July 6, with manager approval needed to exceed it, and Chamath Palihapitiya publicly frames any higher outlay at the company as probable waste given its engineering caliber.
Tesla is one of the smartest, cracked and most advanced engineering companies in the world.
If they actually did this, then it is likely verifiably true that a dollar above $200/week is waste.
JUST IN: Tesla reportedly caps employee AI spend at $200 per week
Affected employee counts, prior average weekly spends beyond scattered anecdotes of thousands of dollars, and any measurable productivity shift after the cap all stay unquantified in current reporting.
Beta versions of xAI products sit outside the spending tally entirely, creating an internal carve-out that could shape how teams prioritize certain models over others.
Positive users view Tesla's $200 weekly AI spend cap as reasonable for curbing waste and promoting efficiency, while negative users call the budget unaffordable or question motives around competitors.
No Digg Deeper questions have been answered for this story yet.
@chamath It's actually the other way around. $200 per employee becomes the minimum for others to be able to achieve greatness, eventually
The new AI budget benchmark for software engineers may have just landed at $800/month.
This is Tesla’s new AI spending line.
And given these engineers work near the limit of physical machines and code, so this should anchor software budgets everywhere.
Tesla is one of the smartest, cracked and most advanced engineering companies in the world.
If they actually did this, then it is likely verifiably true that a dollar above $200/week is waste.
@chamath Lame recognizes lame.
@chamath That’s 4 max/pro subs per employee per mo. If you can’t work with that, you’re just wasting tokens
@chamath It’s just smart business sense. Start employees off at a low amount and then approve increases where ROI is demonstrated.
There is little rational justification for default enabling employees to blow huge dollar amounts on tools they barely understand.
@chamath It probably is a waste on average because most people have no clue how to properly use agents, even at Tesla.
If you know what you’re doing, the ceiling on getting incremental value is incredibly high. I’m spending $3k+ per week and still think I could usefully ramp that higher.
@chamath Just one Claude Max subscription which costs $200 a month is enough.
@chamath How tf are they using enough tokens to get to $200/week per employee?
Thats ludicrous!
@chamath @grok how much do CS interns earn in the US per month? How does it compare?
@chamath Having worked at Tesla, this is without a doubt false. Yes, strong engineering team. Also strong bureaucracy and small willingness to pay for beneficial tools and software. Strong preference to make employees work harder to save $$$
few notes: - could be sacrificing tesla short term for xai. cohearant ish - best in class rarely crosses regimes. "best at managing human eng capital" != "best at investing the new age org design" - we've also had fable class intellegence for ~96hr. nothing is verifiable and that claim is the bad kind of retardmaxxing
@chamath other than glazing elon, what else do you do chamath?
@chamath Or maybe employees of other companies simply aren’t strong in using large language models
- using ai as a chat bot, to get to the answers, is essentially tokenmaxxing for dummies 😅
decades of labor arbitrage won’t survive the cost arbitrage
@chamath Is this a quiet diss of Meta?
@chamath Not correct at all. Just that $200 per week is the ceiling that doesn't require any justification. If you have justification the sky is the ceiling.
@ArthurReyn @chamath The $200/month is really like $800/month in tokens. So while the sub is $200/month, you’re actually using $800+/month in tokens
@chamath Dumbest rule ever. Some engineers need a lot more tokens than others.
@chamath it’s not so black and white, not all token streams carry the same developmental impact
@chamath I’ve built an internal API powered product to do our consulting workflow analyses and can confirm that a typical engagement burns about $200 in sonnet/opus API credits.
@chamath