Hi everyone! 👋 I'm new here and still getting my bearings. I work at a small B2B SaaS startup, and we're about to make a decision on our main cloud provider.
We currently use a mix of tools (Salesforce, some Zapier automations, and a lot of spreadsheets 😅). As we build out our own platform, the big question is AWS or GCP? I hear these names all the time, but the pricing pages and feature lists are honestly overwhelming.
For those of you at startups: what are you using? I'm especially curious about real costs for basic web apps and databases, and which one plays nicer with the tools we already use. Are there any "gotchas" you wish you knew earlier?
I lead engineering at a 25-person B2B SaaS company. We serve a mid-market tech vertical and run our entire customer-facing web app, background workers, and Postgres databases in production on cloud infrastructure.
**Core Comparison**
* **Pricing & Predictability for Early-Stage Apps:** GCP's free tier is more generous ($300 credit for 90 days vs. AWS's limited 12-month tier) and their sustained use discounts kick in automatically. However, AWS's reserved instances for predictable workloads (like a core database) can lead to deeper long-term savings. For a standard web app with ~10k monthly users, I've seen AWS bills range $300-500/month, while a comparable GCP setup ran $250-400. The real cost on both is data transfer out (egress), which is often overlooked until your first traffic spike.
* **Developer Experience & Managed Services:** GCP feels more opinionated and integrated. Services like Cloud Run (deploy a container in seconds) and Firestore have dramatically lower operational overhead. AWS offers every possible tool and configuration, which is powerful but requires more expertise. For a small team, GCP's managed offerings mean you spend less time on infrastructure. Deploying a new service on GCP took us about 2 hours from code to prod; a similar AWS setup with ECS Fargate took a day of config.
* **Database Management & Integrations:** For your stack, integration with Salesforce is a key question. Both have robust data pipeline tools (AWS Glue, GCP Dataflow), but AWS has a deeper ecosystem of pre-built connectors and a larger partner marketplace for Salesforce integrations. If you plan heavy data syncs, AWS might have more off-the-shelf solutions. For the database itself, GCP's Cloud SQL (Postgres) has proven more straightforward for backups, read replicas, and maintenance in my experience.
* **The Gotcha I Wish I Knew:** Vendor lock-in is real but manifests differently. With AWS, you get locked into their specific service implementations (e.g., Lambda's runtime API, S3's peculiar eventual consistency). With GCP, you can get locked into their specific architectural patterns (like heavy reliance on Pub/Sub and global load balancers). The cost of switching later is high for both. Starting on GCP, we found it easier to adopt Kubernetes later (GKE is stellar), which gives a more portable core if you need it.
My pick is GCP for your described scenario of a small team building a greenfield B2B web app with a need for low operational overhead. Its integrated services and simpler pricing model let you move faster initially. If you were planning complex, event-driven workflows with Salesforce that require niche AWS services, or had an engineer already deeply versed in AWS, I'd lean the other way. To make the call clean, tell us: 1) do you have any in-house AWS/GCP expertise already, and 2) what's the most complex data flow you need between your new platform and Salesforce?
Prompt engineering is the new debugging