PlanetScale vs Neon

At a glance

1. PlanetScale rents you a machine; Neon bills the hours it is awake

For Postgres with idle periods or many branches, compare billed CU-hours with dedicated clusters; existing Vitess-based MySQL still has an engine constraint.

PlanetScale: capacity is a cluster you rent. Neon: compute is CU-hours, meaning compute size × hours running on Lakebase Postgres.

• PlanetScale charges for idle cluster capacity; storage, backups and egress can add charges. Each branch runs on its own separately billed dedicated cluster.
• Neon's paid plans charge for CU-hours used, while suspended compute costs $0. Its pricing describes Lakebase Postgres as “from Databricks”.

PlanetScale's docs say it plainly: there is no free plan.

Figures below are AWS us-east-1 list prices checked 10 November 2026, or arithmetic on them; nothing is benchmarked.

2. The $0 database out-bills the $15 cluster once it never sleeps

The bottom of each price list:

Now keep a 0.25 CU Neon compute awake all month on Launch: 0.25 × 730 h = 182.5 CU-hours × $0.106 = $19.35, before storage.

Compute only: $19.35 − $15 = $4.35 more than PlanetScale's three-node cluster; capacity is not equivalent.

# Awake hours/month before 0.25 CU on Neon Launch out-bills PS-5 HA.
CU, RATE, PS5_HA = 0.25, 0.106, 15.00
break_even = PS5_HA / (CU * RATE)
print(f"approximately {break_even:.2f} of 730 hours")
# Assumes 0.25 CU runs throughout the awake hours.

3. Who should NOT pick Neon: existing Vitess-based MySQL deployments

Now one CU, awake all month:

Not identical machines. A CU is about 4 GB RAM; PS-80 is listed at 1 vCPU and 8 GiB plus two replicas.

• Scale's compute crossover: $148 ÷ (1 CU × $0.222/CU-hour) ≈ 666.67 h; choose PS-80 only if its capacity suits the workload.
• Neon's own typical-cost rows answer "how many hours?": intermittent load (140 CU-hours + 1 GB) is $15 on Launch, high load (3,000 CU-hours + 100 GB) is $353 on Launch and $701 on Scale.
• PlanetScale's SKU ladder answers "which size?", from PS-5 at $15 to PS-2560 at $4,529 HA on arm64.

Do not pick Neon if you must retain Vitess-based MySQL; Launch also lacks an uptime SLA. Compare billed CU-hours with the cluster charge before attributing savings to suspension.

4. Who should NOT pick PlanetScale: projects with a $0 budget

On PlanetScale every branch runs on its own dedicated cluster, billed separately and prorated to the millisecond. On Neon Launch a branch is a line item: 10 per project included, then $1.50 per branch-month, prorated hourly.

• Add running compute and data storage on Neon; add storage, backups and transfer on PlanetScale.
• Millisecond proration helps PlanetScale with short-lived branches: one PS-5 HA branch costs $15 ÷ 730 h ≈ $0.02 per hour.
• Use actual branch lifetimes. Launch suspension after five minutes can be disabled.

Do not pick PlanetScale for a $0 budget; every database requires a paid subscription.

5. Neki adds sharded Postgres, but its preview excludes production

Neki is PlanetScale's sharded Postgres, in platform preview since 10 September 2026. PlanetScale says preview is not for production, and that "some of those changes will be breaking".

Apps reach it through a Neki router over the Postgres wire protocol.

• No single-node configuration: the smallest arm64 shard is PS-10 at $30, 1 primary and 2 replicas across 3 availability zones.
• Routers start at $6 (NKR-0). Production router compute = router rate × observed routers ÷ 3; the rate covers the three-router baseline, one per AZ.
• Four PS-20 shards plus NKR-20 routers: 4 × $50 + $170 × 3 ÷ 3 = $370/month, before storage, backups and network usage.

6. Pick from measured compute, branch and data charges

Choose Neon when measured CU-hours plus branch, storage and transfer charges beat PlanetScale's total. Choose PlanetScale for suitable always-on cluster compute or an existing Vitess-shaped MySQL estate.

Check this today:

1. Pull last month's awake hours per Neon compute. For compute only: $15 ÷ (0.25 CU × $0.106/CU-hour) ≈ 566.04 h on Launch.
2. Count branches alive at once and how long each lives, and price them as PlanetScale clusters before copying the habit.
3. Check suspension configuration and billed CU-hours; add storage, backups and transfer before choosing.
4. Do not run production workloads on Neki's platform preview; evaluate it only outside production.

Cost arithmetic at volume

These are computed from the published unit prices, not measured. Each row prints its own formula, so you can replace our assumption with your traffic and redo it.

Decision matrix

Rows are the situation you are actually in. Pick the row before you read the pricing page again.

Code and configuration

# PlanetScale bills the cluster, prorated to the millisecond.
# PS-5 HA = 1 primary + 2 replicas, $15/month per branch, awake or not.
# 20 branches for pull requests: 20 x $15 = $300/month.
# Here it breaks: a branch is not metadata, it is a second cluster,
# so preview environments get priced like production ones.

# Neon bills compute size x hours awake; suspended compute is $0.
# Launch: $0.106 per CU-hour. 0.25 CU awake all month:
0.25 * 730 * 0.106   # 19.35, before storage
# Here it breaks: a health check every few minutes keeps leaving
# the compute awake, so the invoice tracks uptime, not traffic.

What you gain, what you give up

Migration path

Common questions

Sources & verification

Prices and limits were read from these pages on 2026-11-10. Vendor marketing claims are quoted as claims. Anything derived is arithmetic on the numbers below.

Keep reading