Running a Solana node: cost and disk
Solana nodes are server-class: many fast cores, hundreds of GB of RAM and several NVMe drives. Disk use is capped by ledger limits, so the cost is hardware and bandwidth, not bloat.
A 4 TB disk cannot be timed yet
Validator (Agave)
Official requirements
- CPU
- 12 cores / 24 threads, 2.8 GHz+, AVX2 and SHA extensions
- RAM
- 512 GB
- Disk
- Separate NVMe drives: accounts ≥ 1 TB, ledger ≥ 1 TB, snapshots ≥ 500 GB (OS ≥ 500 GB optional)
- Bandwidth
- ≥ 2 Gbit/s symmetric (10 Gbit/s recommended)
- Source
- Anza docs: validator requirements (undated, checked 2026-11-12)
Data and growth
- Data now
- — No data
- Growth
- — No data
No published or measurable growth figure
Bloat clock (disk 90% full at today's growth)
What it costs
Server: none of the plans in our price list has enough RAM, cores or NVMe for this node (2.70 TB needed). Expect custom bare metal.
Home build: not realistic for this node (needs more cores, RAM or disk than a mini PC holds).
RPC node (Agave)
Official requirements
- CPU
- 16 cores / 32 threads or more
- RAM
- 256 GB
- Disk
- Accounts and ledger on separate NVMe drives; more ledger disk for more history
- Bandwidth
- not published
- Source
- Anza docs: RPC requirements (undated, checked 2026-11-12)
Data and growth
- Data now
- — No data
- Growth
- — No data
No published or measurable growth figure
Bloat clock (disk 90% full at today's growth)
What it costs
Server: none of the plans in our price list has enough RAM, cores or NVMe for this node (2.70 TB needed). Expect custom bare metal.
Home build: not realistic for this node (the docs ask for server-class hardware).
Also on the Network Desk for Solana
How this works and where the data comes from
Data size. Where possible we read the uncompressed size of the latest node snapshot that PublicNode publishes for that chain and client (refreshed every few hours to weekly by them; we check every 6 hours). For Bitcoin we use the total block-data series from blockchain.com plus the documented chainstate size. Otherwise we use the figure in the official docs, with its date.
Growth. Best available, in this order: the change in snapshot size over at least 10 days (we build this history ourselves, so it appears after two weeks); Bitcoin's 60-day block-data growth; growth stated in official docs; raw block data from 20 sampled blocks × blocks per day (a lower bound, because state, receipts and indexes come on top). Pruned nodes and capped ledgers are shown as flat.
Bloat clock. Months until the disk is 90% full = (disk × 0.9 − data now) ÷ growth per month. It assumes today's growth rate; gas-limit increases and network upgrades can speed it up.
Cost. Cheapest configuration in our dated price list whose cores and RAM meet the official requirement and whose disk holds max(1.3 × today's data, 90% of the recommended disk). Bare metal fits any node; cloud/VPS plans with local disk fit non-validator nodes unless the docs ask for server-class hardware; a VPS plus network block storage only where the docs do not require local NVMe. EUR and USD plans are compared separately. Home build: the smallest listed mini PC with enough cores and RAM, plus one extra NVMe data drive if needed, at store list prices on the listed date, plus electricity at the stated assumption. Bandwidth is not priced.
Gas-limit scenario (Ethereum). Growth × (new gas limit ÷ today's gas limit). It assumes blocks stay as full as today and that disk growth scales with block gas; repricings such as EIP-8037 change the mix, so read it as a rough upper guide.
Sources: PublicNode snapshots; blockchain.com charts; litecoinspace.org; PublicNode JSON-RPC (block sampling); Official node documentation of each chain (linked per row); Provider price pages (linked in the price table)
Information only, not financial advice. Data comes from public sources and can be delayed or wrong; check the primary source before acting.