I’m a solo operator with no paying customers yet. I wanted more predictable baseline costs while iterating on a side project, so I migrated from App Engine to Cloud Run.
I fed my setup, budget, and constraints as context into Gemini CLI, asked it to search official documentation and best practices, and followed that guidance.
I removed –min-instances=1, expecting autoscaling to reduce idle spend. The commit message claimed “60% cost savings.” The actual outcome was roughly an 1,800% increase and a surprise decline message when I went to buy coffee on my credit card.
From Nov 2 to Dec 14, Cloud Run accrued ~$4,676. There was no traffic spike, no abuse, no application bug. The services were mostly idle.
What compounded: CPU and memory were over-provisioned (4 CPU, 16Gi) from earlier experimentation. Each deploy created a new revision, and I deployed frequently while iterating. Those revisions must have stayed warm longer than expected. Autoscaling plus revision sprawl meant more active resources than intuition suggested, even without traffic.
The billing alert failure: I had alerts enabled at $50. I received one early notification, then silence as spend climbed another $4,600. There were no clear signals that my cost profile had materially changed.
I contacted Google Cloud Billing Support looking for clear guidance or partial relief. After review, they declined any adjustment and closed the case. As a solo dev without an account team, there was no escalation path beyond accepting the charges.
Where it stands now: After right-sizing resources and cleaning up revision sprawl, daily costs dropped from ~$200 to under $5 on my billing dashboard. I’m cautiously optimistic but not certain it’s fully resolved.
For those running Cloud Run longer term: How do you actually cap downside as a solo dev? Do you set hard budget caps and accept downtime? Are there deployment patterns that avoid revision sprawl? Is App Engine still preferable purely for cost predictability? What guardrails work that don’t depend on constant manual billing checks?
I wasn’t chasing scale. I was trying to be careful with money while building alone. I’d appreciate hearing what others would do differently.
the answer is staring you right in the face:
> I fed my setup, budget, and constraints as context into Gemini CLI
> The commit message claimed “60% cost savings.”
don't outsource your critical thinking to a chatbot.
or, if you feel you simply must have the chatbot do this work for you, supervise it more closely. instead you ignored it for 6 weeks:
> From Nov 2 to Dec 14, Cloud Run accrued ~$4,676.
I’ve seen runaway cloud costs at my employer a few times, with a few different services and it took a fair amount of time to figure out the monitoring/alerting. They may change their service agreements or pricing structure in hard to decipher ToS, etc. If the cloud provider won’t refund or credit a company that has a representative, they aren’t going to pay any attention to a solo dev or small team.
I would build locally on your laptop and start with a $5/month VM until you get a paying customer and know what size your system needs.
Also don't outsource your thinking to LLMs, it's a useful tool, but once you do, it's brainrot for programmers.
First of all, thank you so much for obviously writing part of this via a Large Language Model. Second of all, what kind of argument is "The commit message claimed '60% cost savings'" - do you have any idea what you were actually doing? And lastly, addressing your question:
> Do you set hard budget caps and accept downtime?
If you have no clue what you're doing, yes! Especially for early prototyping, why not? IaaS offerings will also just create downtime for you as well if you need more resources than you've provisioned. It's normal. Either you set up a system where you can rely on dynamic scaling or you don't and set hard limits.
You asked your cloud provider to provision resources, and you were billed for them. If you can't handle working with a cloud provider, you might want to look into less scalable but in turn more cost stable infrastructure solutions.
A little more context: I’ve been on GCP for 4 years, App Engine for the majority of it. Expensive but stable. I’ve used Gemini in the past to reduce costs successfully, so this wasn’t my first attempt at optimizing.
I take ownership of the outcome, but the config behavior still doesn’t match my mental model and Google support hasn’t been able to clarify how to properly scope this either, which is why I turned here.