ChatGPT Users’ Politeness Is Draining OpenAI’s Wallet. Sam Altman Says THIS In Response

OpenAI CEO Sam Altman has shared that users’ polite habits — specifically, saying “please” and “thank you” while interacting with ChatGPT — are racking up millions in operational costs for the company. Altman’s comment came in response to a curious user on X (formerly Twitter) who mused about the hidden cost of civility in the AI age.

“How much money OpenAI has lost in electricity costs from people saying ‘please’ and ‘thank you’ to their models,” the user wrote.

Altman’s reply? “Tens of millions of dollars well spent.” He followed up with a cryptic: “You never know.”

X Users Offer Sarcastic Solutions

Altman’s tongue-in-cheek response sparked a range of reactions online, many of which were equally cheeky. One user suggested a simple fix: “I feel this can be solved incredibly easily with client side code answering you’re welcome lol.” Another quipped, “If they really wanted to save on electricity, they'd stop having it end every answer with a question.”

While clearly amused, the online community’s reaction also highlighted a deeper point — how even small, seemingly insignificant user behaviors can have large-scale implications when multiplied across millions of interactions per day.

AI Usage Surge Driving Energy Bills Up

ChatGPT’s popularity has surged in recent weeks, thanks in part to viral trends like the Ghibli-style AI art craze. This has pushed the platform’s weekly active users past 150 million — the highest so far this year.

This rise in usage comes with hefty energy demands. A Goldman Sachs report revealed that each ChatGPT-4 query uses around 2.9 watt-hours of electricity — roughly ten times what a regular Google search consumes. Given OpenAI processes over a billion queries a day, that amounts to an eye-watering 2.9 million kilowatt-hours of daily energy usage.

New Models, Bigger Ambitions

Even as ChatGPT’s politeness tax makes headlines, OpenAI is pushing forward on the innovation front. This week, the company unveiled two new reasoning models — o3 and o4-mini — aimed at outpacing global competitors like Google, Meta, and xAI.

According to OpenAI, the o3 model delivered a stellar performance, scoring 69.1 per cent on the SWE-bench verified test — a benchmark for coding abilities. The o4-mini wasn’t far behind, clocking in at 68.1 per cent.

In the fast-evolving AI race, even manners come with metrics — and, apparently, a multimillion-dollar power bill.

technology