How do compute costs compare to web 2.0 alternatives?

Running IC services is very cheap compared to other L1 solutions; how do compute costs compare to services like AWS?


The costs for various functions are below. You will have to know what you want to do and then do a good bit of math.

impl CyclesAccountManagerConfig {
pub fn application_subnet() -> Self {
    Self {
        // Setting the fee to create canisters to a significant amount to
        // discourage potential DoS attacks that involve creating a ton of
        // canisters on the network.
        canister_creation_fee: Cycles::new(1_000_000_000_000),
        compute_percent_allocated_per_second_fee: Cycles::new(100_000),

        // The following fields are set based on an experiment where we
        // measured how many resources a representative benchmark on a
        // representative subnet is using.
        update_message_execution_fee: Cycles::new(6_000_000),
        ten_update_instructions_execution_fee: Cycles::new(40),
        xnet_call_fee: Cycles::new(9_000_000),
        xnet_byte_transmission_fee: Cycles::new(70_000),
        ingress_message_reception_fee: Cycles::new(22_600_000),
        ingress_byte_reception_fee: Cycles::new(140_000),
        // 40 SDR per GiB per year => 40e12 Cycles per year
        gib_storage_per_second_fee: Cycles::new(1_270_000),

Thanks so much for spotting - this gives a good sense.