/AI2h ago

OpenAI's Gabriel Petersson says AI token consumption will replace headcount as the largest operating expense for some companies

He warns that slow-adapting executives face rapid replacement

30377175918.8K
Original post
gabriel@gabriel1#712inAI

the hardest task for CEOs for 300 years have been scaling companies with more people

but tokens will quickly rise to be some companies largest cost. this will happen at the speed of CEOs learning how to adapt ai or being replaced

12:36 PM · Jun 9, 2026 · 11.8K Views
Sentiment

Positive users highlight rapid AI scaling benefits and strong ROI from rising token costs, while negative users question whether inference will replace labor at scale and criticize company leaders.

Pos
50.0%
Neg
50.0%
10 comments with sentiment.
Cluster Engagement
Posts from X
Most Activity
Most Activity
VIEWS3.4KBOOKMARKS5REPLIES2
gabriel@gabriel1

denier ceos will have a hard time when their board asks why their competitor can do their accouting, repetitive operations work, and customer calls 10x faster and 10x cheaper. and it would take at least a year to intuitively understand and implement

gabriel@gabriel1

the hardest task for CEOs for 300 years have been scaling companies with more people

but tokens will quickly rise to be some companies largest cost. this will happen at the speed of CEOs learning how to adapt ai or being replaced

2hViews 3.4KLikes 61Bookmarks 5
LIKES70RETWEETS2
gabriel@gabriel1

that's why companies scramble to use "ai", people really want to learn

when you have intelligence on a tap that improves 4x per year, and intelligence is the most important ingredient to economic productivity, it's becoming increasingly obvious for them where we're heading

gabriel@gabriel1

denier ceos will have a hard time when their board asks why their competitor can do their accouting, repetitive operations work, and customer calls 10x faster and 10x cheaper. and it would take at least a year to intuitively understand and implement

2hViews 3.1KLikes 70Bookmarks 5
Kevin Kwok@kevinakwok

@gabriel1 It is very funny when you look at corporate history with the reframe of it being forced scaling of people despite no one wanting that

gabriel@gabriel1

the hardest task for CEOs for 300 years have been scaling companies with more people

but tokens will quickly rise to be some companies largest cost. this will happen at the speed of CEOs learning how to adapt ai or being replaced

2hViews 1.1KLikes 7Bookmarks 1
Priyanka Ojha@OjhaO39356

@gabriel1 Recently started reading about the Science of Scaling, and one idea stood out that every era has its scaling constraint. For 300 years it was people. In the AI era, it may be how effectively organizations scale intelligence through tokens, systems, and automation.

1hViews 85Likes 2Bookmarks 1
Rv@InvestorRVD

@gabriel1 COBALT gonna be the next robotics platform by @animesh_garg

2hViews 636
Tony Rost@raspberryman

@gabriel1 I’ve been telling clients to expect 5-10% of net sales will be AI compute, and this will be expected in 10k and investor decks. The ROI will be easily justified by huge reductions in COGS.

1hViews 67Likes 1
Nicholas Mugalli@RealNickMugalli

@gabriel1 So as a result CEOs will be scaling companies worth billions of dollars with maybe 1-2 people at best…what a life

2hViews 36
Weeb@CryptoWeeb808

@gabriel1 One of the recurring patterns in technology is that solving a constraint often reveals a new one.

If AI dramatically reduces the cost of certain types of labor, attention may shift toward efficiently allocating intelligence rather than simply accessing it.

1hViews 22
Akhil Ivaturi@GordianKnot256

@gabriel1 A big part of the CEO-experience is imposing power over a lot of humans. Controlling AI agents instead of humans deprives them of this experience.

1hViews 20
not me@notme7920

@gabriel1 Then what do u think is the most important thing for an Ai pioneer that has been exposed to the field for years. Ofc it’s building and shipping. But what else would u keep in mind or hold ur self with ?

2hViews 15
REAZ@LLMathematician

@gabriel1 Token maxing is the way, Mr. CEO. Thanks for understanding.

2hViews 13

@gabriel1 tokens are the new payroll.

1hViews 9
Paras@buildwithparas

@gabriel1 per-token price keeps dropping but people just run it on more stuff, so the bill doesn't actually go down

1hViews 5
FinanceAI Labs@financeailabs

Tokens as the dominant cost center assumes that inference actually replaces headcount at scale, which five years of AI adoption data does not yet support. Productivity at the margins is real. Wholesale organizational transformation is mostly a conference talking point. Boards are still asking about the ROI that almost no one can clearly quantify.

1hViews 5
iloccorb@dotnet_enjoyer

@gabriel1 can you guys please model-mog anthropic asap, im losing my fucking mind being gaslit that Fable isnt a waste of everyones time

1hViews 4
retto@rettooooo

@gabriel1 Give me more tokens.

1hViews 2
Ferbin@Ferbin08

@gabriel1 tokens are becoming the new cloud costs. but most companies aren't measuring how much each AI task actually costs them. that gap is where the real competitive edge is.

1hViews 2
Load more posts