Rosa Del Mar

Daily Brief

Issue 103 2026-04-13

Pricing And Experimentation Friction

Issue 103 Edition 2026-04-13 8 min read
General
Sources: 1 • Confidence: Medium • Updated: 2026-04-13 04:02

Key takeaways

  • Convex offers a free tier that includes up to 40 deployments and up to six team members.
  • Supabase is positioned as Postgres with direct querying and optional SDKs/RLS for client-to-database synchronization via its compute plane, while Convex is a higher-level application-data abstraction where developers do not directly touch SQL.
  • Environment variable management is described as a remaining Convex workflow gap, with a desire for code-defined env var declarations that auto-sync like other deployment artifacts.
  • Convex pricing includes free monthly quotas such as one million function calls and then charges per additional million calls plus metered compute (GB-hours) and storage overages.
  • Convex supports non-deterministic work via actions that can call external services and then invoke mutations to persist results, while queries and mutations are intended to remain pure database operations.

Sections

Pricing And Experimentation Friction

  • Convex offers a free tier that includes up to 40 deployments and up to six team members.
  • Supabase's free plan limits users to two active projects, requiring pausing or deleting an existing project to test additional ones.
  • Supabase billing is organization-based such that upgrading one project to Pro can cause all projects in that organization to become paid instances with a minimum monthly charge per project.
  • Convex pricing includes free monthly quotas such as one million function calls and then charges per additional million calls plus metered compute (GB-hours) and storage overages.

Abstraction Layer And Security Surface

  • Supabase is positioned as Postgres with direct querying and optional SDKs/RLS for client-to-database synchronization via its compute plane, while Convex is a higher-level application-data abstraction where developers do not directly touch SQL.
  • Convex requires developers to expose data through explicit queries and mutations written as code, whereas Supabase often relies on database schema design plus RLS policies to control what clients can access directly.
  • In Convex, schema and backend logic live in a dedicated project folder so database state and access patterns are described and versioned in code rather than configured primarily in a dashboard.
  • Convex provides strong TypeScript type inference so queries against tables or fields not in the declared schema produce immediate type errors in the editor/LSP.

Workflow Automation And Operability

  • Environment variable management is described as a remaining Convex workflow gap, with a desire for code-defined env var declarations that auto-sync like other deployment artifacts.
  • Convex logs in the cloud surface backend errors to aid debugging of deployed functions.
  • Convex Cloud Code is reported to lack working default LSP support, contributing to type issues that were later fixed by running TypeScript checks and addressing errors.
  • Convex automatically redeploys code changes to the dev instance and can block deployment when schema changes conflict with existing data until the conflicting data is resolved.

Compute Runtime And Scaling Model

  • Convex pricing includes free monthly quotas such as one million function calls and then charges per additional million calls plus metered compute (GB-hours) and storage overages.
  • Convex uses an isolate-based compute layer co-located with the database so requests run near data, enabling faster operations and more usage-proportional scaling than traditional always-on database instances.
  • Convex actions can optionally run on Node by declaring a Node runtime directive at the top of the file.

Side Effects Background Work And Integrations

  • Convex supports non-deterministic work via actions that can call external services and then invoke mutations to persist results, while queries and mutations are intended to remain pure database operations.
  • Convex offers installable backend components (e.g., Stripe integration, agent tooling, crons, and R2/Resend integrations) that can be added without leaving the codebase.
  • Convex Work Pools provide concurrency-limited job execution with queuing and features like retries and completion handling.

Watchlist

  • Convex file bandwidth pricing is described as relatively high, and the team is building a file-storage-to-CDN component to reduce those costs.
  • Environment variable management is described as a remaining Convex workflow gap, with a desire for code-defined env var declarations that auto-sync like other deployment artifacts.

Unknowns

  • What are the current, authoritative Supabase free-tier limits and organization/project billing rules, including any minimum monthly charges per project?
  • What are the current Convex free-tier quotas and overage rates (function calls, GB-hours, storage, file bandwidth), and how often have they changed?
  • How does Convex file bandwidth pricing compare in practice for media-heavy apps, and does the promised CDN-backed file storage component ship with materially different rates?
  • Under what workloads and constraints does the isolate-co-located compute model deliver the claimed performance and scaling benefits, and what are the limits (cold starts, CPU ceilings, runtime restrictions)?
  • How robust and general is Convex's real-time invalidation model for complex queries, high fan-out updates, and multi-tenant permissioning?

Investor overlay

Read-throughs

  • Usage metering plus a generous free tier may reduce experimentation friction and increase deployment sprawl, which could improve top of funnel and later expansion if overages feel predictable.
  • A higher level app data abstraction with code defined access surfaces may shift security and correctness burden from database policy configuration to code review, potentially appealing to teams prioritizing reviewability and type safety.
  • Co located isolate compute plus an actions escape hatch suggests a platform trying to bundle database, realtime invalidation, and background work into one surface, potentially reducing external infra if performance and limits hold.

What would confirm

  • Published, stable free tier quotas and overage rates with low frequency of changes, plus clear billing topology that avoids org wide minimums or hidden per project charges.
  • Environment variable workflow becomes code defined and auto synced, and tooling such as editor or LSP support improves enough that redeploy and debugging workflows stay reproducible at scale.
  • File storage to CDN component ships and materially reduces effective file bandwidth costs for media heavy apps, and realtime invalidation remains robust under high fan out and multi tenant permissioning.

What would kill

  • Free tier and overage rates change frequently or remain unclear, leading to unpredictable marginal costs that discourage multi deployment experimentation.
  • Isolate co located compute shows practical constraints such as cold start issues, CPU ceilings, or runtime restrictions that force many workloads onto actions or external services, undermining the claimed model benefits.
  • File bandwidth remains relatively high and the CDN backed component fails to reduce costs or complexity, making media heavy workloads uneconomic or operationally cumbersome.

Sources

  1. youtube.com