What serverless costs when nobody is using it, product by product, 18 October 2026

From the products' own pricing pages on 18 October 2026, excluding storage, at 730 hours and US East rates: Lambda event functions, Cloud Run, Cosmos DB serverless, Neon and Redshift Serverless cost zero idle; Aurora Serverless v2 zero since November 2024 and OpenSearch's next-generation collections zero since May 2026. MSK Serverless bills $0.75 per cluster-hour, about $550 a month idle; OpenSearch classic collections have a floor of two capacity units, about $350 a month; Neptune Serverless a floor of one capacity unit, about $80 a month; ElastiCache Serverless a minimum of storage metered hourly, from $6 a month. The definitions were clear: the CNCF's whitepaper says 'there is no charge when code is not running', and the Berkeley paper warned in 2019 that few so-called elastic services could scale to zero.

Decided by. Nobody yet: a candidate for Dinis to read, comment on and decide (keep, rewrite, merge, cut, article first, later, not doing it).

Note. The monthly figures are Dinis's arithmetic and move with region and configuration; the article says so. Some floors have fallen over time: Aurora Serverless v2 had a floor of half a capacity unit, priced in 2022 at 'just shy of $45' a month, until scale to zero arrived in November 2024; OpenSearch Serverless started at four units, halved in June 2024, reached zero in May 2026. For the hands-on chapter the table is a dated figure to renew each year; the test it supports (ship-now-ship-serverless) does not date.

Where it comes from

Ephemeral town planner services What it means now (the table); What serverless meant (newsroom)

What it answers in the 2018 book

Serverless answered outline, Concepts

Where it goes

Chapter Serverless, for real, part Build it without a server (drafting); topic Building blocks: cloud, serverless, free models, git and CI.

Other ideas proposed for the same chapter: Building blocks: cloud as Lego, free models with your own key, git as a standard, The file system is the database: load, extract, transform, save, and keep every intermediate as a file, No server, by design: retire the back end too, and keep only commodity servers, Ship now, and ship your startup serverless: no servers of yours, no database, client side, with privacy by design and control of your agents, The name shows the glue: a first version is often a villager's work dressed as a town planner's, Move the cloud, not the code: a twin the application cannot tell from the real service.

Node idea:what-serverless-costs-when-idle in graph.json; see it in the graph. Source: admin/content/graph/ideas.json.