Three people with 100 AI programmers burned through $1.3 million in a month! OpenAI: I'll cover the cost.

This article is machine translated
Show original

OpenClaw

Peter Steinberger

Edited by: Solomon

[New Intelligence Summary] 3 people, 100 AI agents, burning through $1.3 million a month—the father of OpenClaw turned software development into an "AI pipeline," with OpenAI footing the bill.

While others post their payslips, he posts his bill— $1.3 million a month!

That's nearly 9 million RMB per month. This completely stunned netizens.

Peter Steinberger, the creator of OpenClaw, casually tossed out a screenshot on X.

OpenClaw

Peter Steinberger

But the numbers in the screenshot are truly unsettling:

30-day expenditure: $1,305,088.81. 603 billion tokens were consumed. 7.6 million requests were made.

You read that right, it's $ 1.3 million. And that's not some big company's quarterly AI budget— it's just a three-person team's monthly usage.

Even more shocking is that OpenAI will reimburse this expense.

The comments section went wild instantly.

Some were amazed, some questioned, and some pulled out calculators to calculate "how many programmers this is equivalent to".

Steinberger himself responded calmly: "After turning off Fast Mode, my costs were less than the cost of an engineer, and it really helped a lot more."

In short— it's a really good deal!

Some netizens were amazed by the engineer's monthly salary of $400,000 – "San Francisco's job market is insane."

OpenClaw

Netizen comments

OpenClaw

Netizen comments

Some netizens are also curious about where all those tokens went.

Peter responded that most of it was used for the development of OpenClaw.

OpenClaw

Netizen comments

Cloud Programmer Army

The most outrageous thing about this is that Pete's small team only has 3 people.

They run approximately 100 Codex instances in the cloud for extended periods, performing the dirtiest, most tiring, and most maddening work in software engineering.

Reviewing PRs, finding security vulnerabilities, deduplicating issues, fixing bugs, monitoring benchmarks, discovering regressions and posting them to Discord, and even opening PRs directly after listening to meetings.

Therefore, AI is not just about "writing code for you," but about entering every gap in software collaboration.

That's terrifying.

Because the real expense in software development lies in communication, understanding, context switching, review, regression, fixing, waiting, and repetitive work.

In the past, a team would spend a lot of time every day on things that didn't seem like "creation" but would ruin the project without them.

Now, Peter has delegated all these steps to a group of AI agents.

This is where AI begins to maintain the nervous system of an organization for you.

OpenClaw

Schematic diagram

There's another important detail in this screenshot: it's not the OpenAI backend, but rather CodexBar created by Peter.

CodexBar is a macOS menu bar tool used to track usage windows, credits, costs, and reset times for various AI programming tools.

It supports a host of services including Codex, Claude, Cursor, Gemini, and Copilot.

What used to be in a programmer's menu bar? CPU, memory, battery, network speed.

Now there's something new: tokens. Tokens are becoming a new kind of "means of production".

OpenClaw

CodexBar

Finally, a couple of words.

$1.3 million a month, 3 people, 100 AI agents.

Think about these numbers: three living people, leading a hundred digital employees who don't eat, sleep, or ask for raises, did the work of an entire engineering team.

Some people were thrilled after watching: AI is finally not just a pretty face that can only chat! Others felt a chill down their spine: Wait, what are we coders going to do now?

But to be honest, what keeps me up the most is Steinberger's nonchalant remark: "I'm exploring what software development would look like if token costs weren't an issue."

OpenClaw

Peter Steinberger

Ladies and gentlemen, he meant "if".

The problem is that this "if" is turning into "when" at a visible speed.

Today's work cost $1.3 million. The model price was reduced once, to $130,000. Then it was reduced again, to $13,000.

On that day, having 100 AI agents working simultaneously will no longer be the exclusive domain of Silicon Valley giants, but rather a basic operation for any three-person startup team.

Three young men in a garage, commanding a hundred tireless AI programmers—the image is absurd to even contemplate.

Peter Steinberger revealed his hand.

The card on the table reads: The future is already knocking on your door, and it's not going to wait for you to be ready.

References:

https://the-decoder.com/for-1-3-million-a-month-openclaw-founder-peter-steinberger-runs-100-ai-agents-that-code-review-prs-and-find-bugs/

https://x.com/steipete/status/2055346265869721905

https://developers.openai.com/codex/speed

This article is from the WeChat official account "New Zhiyuan", author: New Zhiyuan.

Source
Disclaimer: The content above is only the author's opinion which does not represent any position of Followin, and is not intended as, and shall not be understood or construed as, investment advice from Followin.
Like
66
Add to Favorites
16
Comments