Find answers to the problem of using auto-drive space

There are two issues that I have not been aware of in terms of specific technical parameters.
Please ask someone who knows to answer:

1: If Autonomy has a total of 188PB of space. What is the theoretical maximum storage space for auto drive?
If all 188PB of space is stored. What if there are new data that need to be stored later?

2: What is the initial size of the free space currently open to a single user? After the initial free space is used up. It is a monthly renewal fee. After renewing, you can permanently use the purchased space.

Theoretical Maximum Storage for Auto Drive

Current Network Capacity

Based on the chain specification in the Autonomys Subspace repository, here are the key parameters:

  • Total Network Pledged Space: 158 PiB (162,201.6 TB)
  • Replication Factor: 25x (defined in chain spec)
  • Theoretical Maximum Usable Storage: ~6.34 PiB (6,488 TB)

Source Code Reference

Found in subspace_repo/crates/subspace-runtime-primitives/src/lib.rs:35-37:

/// Minimum desired number of replicas of the blockchain to be stored by the network,
/// impacts storage fees.
pub const MIN_REPLICATION_FACTOR: u16 = 25;

Calculation

Effective Storage = Total Pledged Space ÷ Replication Factor
                  = 158 PiB ÷ 25
                  = 6.34 PiB (6,488 TB)

This means that out of 158 PiB total network capacity, approximately 6.34 PiB is theoretically available for actual user data storage, with the remainder used for 25x redundancy to ensure data reliability.

What Happens When Storage Fills Up?

The network uses a dynamic pricing mechanism to handle capacity constraints. The storage fee formula (from pallet-transaction-fees/src/lib.rs):

pub fn calculate_transaction_byte_fee() -> BalanceOf<T> {
    let credit_supply = T::CreditSupply::get();

    match (T::TotalSpacePledged::get() / u128::from(T::MinReplicationFactor::get()))
        .checked_sub(T::BlockchainHistorySize::get())
    {
        Some(free_space) if free_space > 0 => {
            credit_supply / BalanceOf::<T>::saturated_from(free_space)
        }
        _ => credit_supply,
    }
}

Key Behavior

As the network fills up:

  1. Available storage decreases → denominator in the formula gets smaller
  2. Storage costs increase proportionally → higher fees per byte
  3. Economic incentive created → farmers are incentivized to add more storage capacity when demand is high
  4. Network naturally expands → the system is designed to scale as needed

The formula ensures that:

storage_fee_per_byte = total_credit_supply / ((total_space_pledged / min_replication_factor) - history_size)

As storage approaches capacity, fees rise automatically, creating market pressure for expansion.

Free Storage Allocation & Pricing

The current free storage allocations on ai3.storage are sponsored by the Subspace Foundation as a grant that renews monthly. This initial free tier allows users to experience the service, but is not guaranteed to be a permanent fixture.

The team is actively working on a direct purchase mechanism that will enable users to buy storage on ai3.storage. This represents the only sustainable path forward, ensuring:

  • Users pay for their own storage needs
  • The Foundation’s grant program isn’t drained by sybil attacks (users creating multiple wallet accounts to abuse free allocations)
  • Long-term viability of the storage service

In short: Current free tier = temporary Foundation grant. Future model = user-paid storage purchases.

Thank you very much for your careful reply. I have a clear understanding of the capacity part now.
But I still have a question: if there is a third-party application (app) that is based on the storage space of ordinary free users. For example: blog. So for it, being free is a key step in attracting users. That is to say, the initial user accumulation. Very important. So how can we strike a balance between free and paid services? It means having a large number of users and being able to charge reasonable fees.
Currently, there are very few use cases for auto drive. How to promote and attract initial users?
And auto drive (DSN) is a clear highlight and difference between automomys and other non storage projects.

Or rather, auto drive is not suitable for the scenarios used by ordinary portal websites in the past. Only applicable to AI storage and computing?

It may make sense to build something like this using the Auto SDK rather than relying on Auto Drive. The “free” tier of storage is only available on Auto Drive. Using the Auto SDK would allow the publisher to set reasonable usage limits and monetize the service once it exceeds a free tier. The foundation might even consider providing a grant to help fund such a project.

There is no truly free option. Someone must cover the costs, and those fees compensate farmers for their hardware investments.

Thank you. Of course I am aware of this plan.
My viewpoint is:
After a large number of free users flooded into auto drive. Perhaps it is the peak period of space utilization. Only then will there be the emergence of paid users who will use AI3 as a payment method to purchase Auto Drive. Then users become familiar with both the advantages of auto drive and the use of AI3. As a third-party app, it also started by mining novice users. Of course, my logic may not align with your ideas. Never mind. We can discuss.

It sounds like you have ideas for utilizing the DSN, which is great! Ultimately, it comes down to the mechanics of funding some storage and possibly the development of the application itself. Both are strong use cases that I think would warrant serious consideration by the foundation.

I encourage you to submit a proposal and see where it leads.

Let me give you an analogy: a third-party app uses auto drive to do video blocking. So if it’s the app itself. You can apply for a lot of storage space at the foundation. But if these blogs are not stored in a way that is based on basic users, then the charging method becomes a difficult point: users pay for their own private space. If you make this app pay. It is impossible to complete the calculation. So the conclusion is that we still need to expand the size of our private free space (currently only 1MB)