- OpenAI CEO says users’ polite habits like saying “please” and “thank you” cost millions in server expenses.
- Each extra word adds to processing time and cloud computing costs.
- With billions of daily queries, even small additions snowball into huge bills.
- The issue highlights the rising operational cost of large language models like ChatGPT.
- OpenAI now focuses on making AI responses more efficient to cut costs.
When Kind Words Cost Real Money
Politeness isn’t free—at least not for OpenAI.
In a recent public talk, OpenAI CEO Sam Altman shared a surprising detail about how small words like “please” and “thank you” are quietly racking up big expenses. These extra tokens—meaning units of text processed by models like ChatGPT—might feel small to users, but they add up quickly when billions of queries are involved.
“Those words are tokens. And every token costs us something to process,” said Altman.
Every Word Has a Price
OpenAI’s artificial intelligence, including its popular ChatGPT product, runs on massive computing infrastructure. Each prompt or reply passes through GPUs in cloud servers, which charge per token processed.
- A “token” can be as small as a word or part of a word.
- Each polite phrase adds 1–2 tokens at minimum.
- Multiply that by millions of users—and you get a hefty bill.
Altman estimated that simply saying “please” and “thank you” in prompts may cost the company millions of dollars annually in server time and compute power.
The Scale of AI Usage Makes It Expensive
ChatGPT now serves hundreds of millions of users globally. With people asking questions, generating content, and even just chatting casually, every interaction counts.
- Even a one-word input like “Hi” uses tokens for both the input and the AI’s response.
- When users send multi-part messages or polite expressions, the cost compounds.
- OpenAI uses Microsoft’s Azure for cloud services, which charges by compute time and memory usage.
What seems small from the user’s side becomes massive at scale.
Why OpenAI Is Talking About This Now
As ChatGPT becomes more integrated into daily life, from student essays to corporate work, OpenAI is working to balance user satisfaction with sustainability.
This isn’t about discouraging polite behavior. Instead, it’s a reminder of the hidden costs behind AI usage.
OpenAI is now:
- Investing in faster, more efficient models like GPT-4 Turbo
- Streamlining backend processes to reduce token handling
- Considering pricing models that reflect resource usage more accurately
So Should You Stop Being Polite?
Not at all. But being aware of how AI operates can lead to smarter usage. If you’re a heavy user, consider:
- Using fewer filler words when not needed
- Being concise in prompts when generating long texts
- Understanding that even small messages are part of a much larger system
The “please” problem opens a larger conversation—how to build AI that’s both powerful and cost-effective. As usage grows, so do expectations.
Can AI companies keep these tools affordable for users while covering operational costs? Will they limit free access or start charging per token?
