Bottom-Up Estimate of YouTube's Daily Running Cost: Delivery, Storage, Compute and Staff

Estimate what it costs Google to keep YouTube running for one day. ```hint Split by what scales Break the cost into pieces that each grow with one driver you can estimate, such as hours watched, hours uploaded or headcount, and size the biggest driver first. ``` ### Clarifying Questions - Should the cost include payments to creators and to music and media rights holders, or only the cost of operating the service (infrastructure and people)? - Should infrastructure be priced at what Google likely pays on its own network and data centers, or at public cloud and CDN list prices? - Are all surfaces in scope (web, mobile apps, TV, Shorts, Music), or only the core video site and app? ### What a Strong Answer Covers - A clear equation that splits daily cost into delivery, storage, compute, people and, if in scope, creator and rights payouts - Labeled round assumptions for watch time, bitrate, upload volume and unit costs, with the arithmetic shown - Which term dominates, and how sensitive the total is to the unit-cost assumptions - Storage treated as the whole accumulated library in several encoded versions, not just one day's uploads - A sanity check of the result against an independent anchor such as traffic volume or annual revenue ### Follow-up Questions - Which single assumption moves your total the most, and how would you narrow it down? - If more viewing shifts to TVs and the average streaming quality rises, how does the daily cost change? - How would the cost of an hour of short-form viewing compare with an hour of long-form viewing?