# If a $500K Engineer Isn't Burning $250K in Tokens a Year, an Alarm Should Go Off

Page: https://stenobird.com/podcast/yesterday-i-went-to-mars-7106663/if-a-500k-engineer-isn-t-burning-250k-in-tokens-a-year-an-alarm-should-go-off
Text version: https://stenobird.com/podcast/yesterday-i-went-to-mars-7106663/if-a-500k-engineer-isn-t-burning-250k-in-tokens-a-year-an-alarm-should-go-off.md
Podcast: [*“Yesterday, I Went to Mars ♡”*](https://stenobird.com/podcast/yesterday-i-went-to-mars-7106663)
Published: 2026-07-21T20:22:00+00:00
Episode link: https://podcasters.spotify.com/pod/show/makotowillolympusmons/episodes/If-a-500K-Engineer-Isnt-Burning-250K-in-Tokens-a-Year--an-Alarm-Should-Go-Off-e3mbkpd
Audio file: https://anchor.fm/s/fcce555c/podcast/play/123113709/https%3A%2F%2Fd3ctxlq1ktw2nl.cloudfront.net%2Fstaging%2F2026-6-21%2Fb5ea0853-6832-f14f-3dec-c6f67f2344d8.mp3
Processing state: not_requested
JSON: https://stenobird.com/v1/public/podcasts/yesterday-i-went-to-mars-7106663/episodes/if-a-500k-engineer-isn-t-burning-250k-in-tokens-a-year-an-alarm-should-go-off
Duration seconds: 211

## Resource

This episode looks at a line attributed to Jensen Huang that stopped the reader mid-scroll: if an engineer earning $500K a year isn't spending $250K on AI tokens annually, an alarm should go off. It works through what that framing actually implies — token spend as a proxy for effort, measured against salary — and sets it alongside a run of related stories: Salesforce freezing engineer hiring while committing billions to Anthropic tokens, Meta ranking 85,000 employees by token usage, Uber burning through its entire annual AI budget by April. There's also a quieter data point underneath all of it: tech layoffs in 2026 are running at over a thousand people a day, around half attributed to AI, with engineers now being cut at the resume stage who would have sailed through to interviews a year ago. A brief look at one visible pitfall, too — Microsoft engineers reportedly consuming $8,000 in tokens per day each, until leadership stepped in — and what happens when a measurable number becomes a performance metric: the same structure as meaningless overtime at a company that rewards long hours. A quiet reflection on what it feels like when token costs and labor costs start being discussed in the same breath, and why the more important question may not be how much to use AI, but where to hand things off and where to hold the work yourself.

## Actions

- request_transcript: `POST https://stenobird.com/v1/public/podcasts/yesterday-i-went-to-mars-7106663/episodes/if-a-500k-engineer-isn-t-burning-250k-in-tokens-a-year-an-alarm-should-go-off/transcription-requests` — Idempotently request low-priority transcript generation for this episode.
- read_markdown: `GET https://stenobird.com/podcast/yesterday-i-went-to-mars-7106663/if-a-500k-engineer-isn-t-burning-250k-in-tokens-a-year-an-alarm-should-go-off.md` — Read the agent-friendly Markdown representation of this episode resource.

A page view does not enqueue transcription. Agents should invoke `request_transcript` explicitly when they need this episode processed.

## Transcript

Full transcripts are not published on public pages unless there is a clear rights basis.