I had three empty drive bays and a spare Ryzen box sitting in the corner of my room for almost two months before I actually looked into what it takes to run a Filecoin storage node. Everyone online makes it sound like plugging in a NAS and watching FIL roll in. It isn’t that. Arweave gets the same treatment, sold as “store data forever, get paid forever.” Also not quite true. Both networks pay real node operators real money. But the hardware, the collateral and the setup are nothing like a weekend project. Here’s what actually goes into it.
Filecoin vs Arweave, two completely different storage bets
Filecoin rents out storage like a landlord rents an apartment. You commit hardware, seal data into sectors, prove you’re still storing it every single day, and you get paid FIL for the deal plus a shot at block rewards. Stop proving and you lose your commitment. Arweave works the opposite way. You pay once, upfront, and the network is designed to store that file for roughly two hundred years using a fund called the storage endowment. Miners get paid from that fund forever, not just for the length of a contract.
That difference changes everything about the hardware you buy. Filecoin rewards raw sealing throughput and continuous uptime. Arweave rewards how much of the historical “weave” you can hold and prove access to, fast, using a proof called Proof of Access. Pick the wrong mental model and you’ll buy the wrong machine.
What you need before you touch any hardware
Before you spend a rupee or a dollar on drives, be honest about three things.
- Capital. Filecoin needs collateral in FIL, not just electricity money. Arweave needs less upfront cash but still needs serious storage.
- Linux comfort. Both networks assume you can compile from source, read logs and debug a systemd service at 2am.
- Uptime discipline. Filecoin slashes you for going offline. Arweave just pays you less if your node can’t keep up. Neither forgives a router that resets itself.
If any of these three feel shaky, start with a testnet or a small calibration setup first. Losing pledged FIL to a beginner mistake stings a lot more than losing time.
How to run a Filecoin storage node
This is the part most guides skim past, so let’s actually get into it.
Filecoin hardware requirements
Filecoin’s official Lotus documentation lays out the baseline for sealing 32GiB sectors, which is what mainnet uses. Here’s what that means in real terms.
- CPU: 8 core processor minimum. AMD chips from the Zen generation onward or Intel chips from Ice Lake onward matter because they support SHA extensions, which speed up sealing dramatically.
- RAM: 256 GiB plus swap space. Not a typo. Sealing is memory hungry.
- GPU: An Nvidia card with at least 11GB of VRAM. This handles the heaviest part of Proof of Replication.
- Storage: A 2TB NVMe drive just for the sealing pipeline, separate from your actual long term storage disks.
- Network: Stable, high bandwidth, low latency. Retrieval deals depend on it.
Filecoin’s own documentation is blunt about this: don’t try to run the whole storage provider stack on one machine. Split sealing workers from your main lotus miner process if you’re doing this beyond hobby scale.
The collateral problem nobody warns you about
Hardware is only half the cost. Filecoin uses upfront token collateral, similar to proof of stake systems, sized to the amount of storage power you commit. The network calculates your initial pledge collateral as a share of FIL’s circulating supply relative to your sector’s quality adjusted power against the network baseline. In plain terms, the more storage you commit and the smaller the network baseline, the more FIL you lock up per sector.
There’s a cushion though. Seventy five percent of every block reward you earn vests over roughly 180 days instead of paying out instantly, and that vesting reward counts toward reducing your initial pledge. Miss a WindowPoSt proof deadline, go offline too long, or break a deal, and you get slashed. That’s real FIL burned, not just a warning.
The easier path into Filecoin: PDP warm storage
Here’s something most 2024 era guides completely miss. Filecoin now supports Proof of Data Possession, or PDP, through what the network calls warm storage. It’s built for exactly the problem above. You don’t need the full sealing rig, the GPU or the long term collateral lock to get started. It’s genuinely the fastest way for a smaller operator to become a provider on Filecoin Onchain Cloud without dropping five figures on hardware first.
Step by step Filecoin setup
- Build or rent a machine that meets the hardware requirements above.
- Install Lotus and its dependencies on a clean Linux server, Ubuntu is the safest bet.
- Create and fund your storage provider wallets, owner, worker and control addresses.
- Decide between full sealing (sector commitment plus GPU work) or PDP warm storage if you’re starting small.
- Pre commit a sector, then prove commit it once sealing finishes.
- Set up your window PoSt schedule and monitor it obsessively for the first month.
- Start accepting deals through the market or join a deal making pool if you’re not marketing storage yourself.
How to run an Arweave permanent storage node
Arweave calls its node operators miners, but the mechanics feel closer to running a very specific kind of storage server than Bitcoin style mining.
Arweave node hardware requirements
Arweave’s own hardware guide breaks mining into two phases: syncing and packing data, then mining it. Each phase stresses different parts of your rig.
- CPU: Needs hardware SHA2 acceleration to keep up with the network’s Verifiable Delay Function, which produces a new mining seed roughly every second. Extra cores won’t speed up VDF because it runs sequentially on a single thread.
- RAM: Scales with how many partitions you’re mining. Each 3.6TB mining partition adds real memory overhead, and repacking data into newer formats can spike RAM usage fast.
- Storage: Data is organized into 3.6TB partitions. A dedicated data_dir needs at least 256GB, ideally on NVMe or SSD, since it gets hit constantly by every node service.
- Network bandwidth: The single biggest bottleneck during the syncing and packing phase. Undersized bandwidth means your node falls behind the weave.
Node types you can actually choose from
- Solo miner: Runs everything yourself, keeps all rewards, carries all the hardware load.
- Coordinated miner: Splits partitions across multiple machines you control.
- Pool miner: Shares hashpower and rewards with others, lower barrier to entry.
- VDF server: Just computes the VDF sequence and feeds it to other miners. Interestingly, Apple’s M series chips have posted some of the fastest VDF benchmarks recorded, even though full mining isn’t supported on macOS yet.
- Validator: Doesn’t mine at all, just verifies the network.
How Arweave node rewards actually work
This is where Arweave earns its “permanent storage” reputation. When someone uploads data, they pay once. A portion of that payment goes straight to the miner as an upfront reward, historically around fifteen percent of the fee at current storage prices. The remaining eighty five percent drops into the storage endowment, a fund that earns yield over time and pays miners incrementally across roughly two hundred years, assuming storage costs keep falling the way they have for the past fifty years.
Consensus itself is called Proof of Access. To mine a block, you have to prove you’re storing a randomly selected historical chunk, the recall block. That single design choice is what actually keeps the whole network’s history online instead of letting old data quietly rot on nobody’s disk.
Filecoin vs Arweave hardware and rewards, side by side
| Factor | Filecoin | Arweave |
|---|---|---|
| Core hardware need | 8 core CPU, 256GB RAM, 11GB VRAM GPU, 2TB NVMe | SHA2 capable CPU, scales with partitions, 3.6TB per partition |
| Minimum practical entry | Full sealing rig or PDP warm storage (lighter) | Data_dir 256GB NVMe plus bandwidth for syncing |
| Upfront cost driver | FIL collateral tied to committed power | Mostly hardware and bandwidth, less collateral pressure |
| Reward source | Deal payments plus block rewards | Storage endowment payouts plus small upfront cut |
| Penalty for downtime | Slashing of locked collateral | Lower proving rate, no direct token slash |
| Storage duration model | Fixed length deals, renewable | Roughly 200 years, paid from endowment |
Mistakes that kill new node operators
- Buying consumer grade drives for sealing work. They die fast under sustained write load.
- Underestimating bandwidth costs. Both networks punish slow syncing, and residential internet plans often aren’t built for it.
- Ignoring the collateral math on Filecoin. People commit more sectors than their FIL can safely back and end up in fee debt.
- Running everything on one box. Filecoin’s own team says this outright. One machine, one point of failure.
- Skipping monitoring. A missed WindowPoSt deadline on Filecoin or a stalled sync on Arweave can quietly cost you weeks of rewards before you notice.
Is it still worth it in 2026
Honestly, it depends on what you’re optimizing for. FIL has traded in a fairly soft range through most of 2026, so pure block reward speculation isn’t the pitch anymore. What’s changed is Filecoin’s push toward Onchain Cloud and PDP warm storage, which lowers the barrier for smaller operators who don’t want to buy a GPU rig on day one. Arweave, meanwhile, is less about mining income and more about the AO compute layer building on top of its permanent storage base, which is quietly pulling in new demand for the underlying weave.
If you’re chasing fast passive income, neither network will feel exciting right now. If you actually care about decentralized storage infrastructure and you’re willing to run real hardware properly, both are still legitimate ways to get paid for it. Just go in with your eyes open about collateral, bandwidth and the fact that neither of these is a “set it and forget it” setup.
FAQ
Do I need a GPU to run a Filecoin storage node? Only for full sealing. PDP warm storage lets you provide Filecoin storage without the Nvidia GPU requirement, though it comes with its own tradeoffs around deal types.
How much FIL collateral do I actually need? It scales with your committed storage power relative to the network baseline and circulating supply, so there’s no fixed number. Smaller commitments need proportionally less collateral.
Can I run an Arweave node on a laptop? Technically you can run a VDF server or validator on modest hardware, even a Mac. Full mining with packing and proving needs real storage partitions and bandwidth, so a laptop won’t cut it for actual mining rewards.
What happens if my Filecoin node goes offline? You risk missing WindowPoSt proof deadlines, which leads to slashing of your locked collateral and a drop in your storage power on the network.
Does Arweave slash tokens like Filecoin does? No. Arweave doesn’t use the same collateral and slashing system. Falling behind on proving just reduces your chances of winning block rewards, it doesn’t burn your existing AR.
Is PDP storage as profitable as full Filecoin sealing? It’s generally lower reward per unit of storage since it skips the heavier proof requirements, but the lower hardware and collateral cost makes the return on investment more accessible for smaller operators.
