Akash Network Whitepaper

Friday, February 19, 2021
Buy AKT coin
Save for later
Add to list
Whitepaper.io is a project of OpenBook

AKT: Akash Network Token & Mining Economics Greg Osuri, Adam Bozanich∗ Akash Network, Akash Network (Dated: January 31, 2020) Akash is a marketplace for cloud compute resources which is designed to reduce waste, thereby cutting costs for consumers and increasing revenue for providers. This paper covers the economics of the Akash Network and introduces the Akash Token (AKT). We describe an economic incentive structure designed to drive adoption and ensure the economic security of the Akash ecosystem. We propose an inflationary mechanism to achieve economic goals. We provide calculations for mining rewards and inflation rates. We also present mechanisms for allowing a multitude of fee tokens. ACKNOWLEDGMENTS computing, the three leading cloud service providers — Amazon Web Services (AWS), We thank Sunny Aggarwal (Research Sci- Google Cloud, and Microsoft Azure — dom- entist, Tendermint), Gautier Marin (Tender- inate the cloud computing market with 71% mint), Morgan Thomas (Co-Founder, Kassir), market share[1] and this figure is expected and Brandon Goldman (Frm. Lead Architect, to increase. These providers are complicated, Blockfolio) for providing valuable comments inflexible, restrictive, and come at a high re- that significantly improved the manuscript. curring cost with vendor lock-in agreements[6]. Increased cloud usage has made cloud cost optimization the top priority of cloud service I. INTRODUCTION users for three consecutive years[7]. Outside of the large incumbent providers, or- Cloud infrastructure is a $32.4 billion indus- ganizations do not have many options for cloud try[1] and is predicted to reach $210 billion by computing. Akash aims to create efficiencies in 2022[2]. the cloud hosting market by repurposing com- By 2021, 94% of all internet applications pute resources that go to waste in the current and compute instances are expected to be pro- market. cessed by Cloud Service Providers (CSP) with only 6% processed by traditional data cen- By leveraging a blockchain, Akash intro- ters[3]. The primary driver for this growth duces decentralization and transparency into is poor utilization rates of IT resources provi- an industry currently controlled by monopolies. sioned in traditional data centers as no more The result is that cloud computing becomes a than 6% of their maximum computing out- commodity, fueled by competitive free market, put is delivered on average over the course of available and accessible anywhere in the world, the year [4], and up to 30% of servers are co- at a fraction of current costs. matose[5] – using electricity but delivering no Akash is the world’s first and only Su- useful information services. percloud for serverless computing, enabling With 8.4 million data centers globally, an anyone with a computer to become a cloud estimated 96% of server capacity underuti- provider by offering their unused compute cy- lized, and accelerated global demand for cloud cles in a safe and frictionless marketplace. In this paper, we present an economic sys- tem that uses Akash Network’s native currency, ∗ [email protected], [email protected] AKT, to achieve economic sovereignty in our

2 decentralized computing ecosystem. We also work by validating and relaying trans- propose an inflation design for mitigating in- actions, proposing, verifying and final- herent adoption challenges that face an early izing blocks. There will be a limited market economy — lack of sufficient demand set of validators, initially 64, required to from the tenants (consumers of computing), maintain a high standard of automated which negatively impacts demand due to lack signing infrastructure. Validators charge of supply. We also present a mechanism for a delegators a commission fee in AKT. stable medium of exchange by solving for to- ken volatility, a major challenge for adoption Delegator: Delegators are holders of the of decentralized ecosystems. AKT and use some or all of their to- Note: This whitepaper represents a contin- kens to secure the Akash chain. In re- uous work in progress. We will endeavor to turn, delegators earn a proportion of the keep this document current with the latest de- transaction fee as well as block rewards. velopment progress. As a result of the ongoing Provider: Providers offer computing cycles and iterative nature of the Akash development (usually unused) on the Akash network process, the resulting code and implementa- and earn a fee for their contributions. tion will likely differ from what this paper Providers are required to maintain a represents. stake in AKT as collateral, proportional We invite the interested reader to to the hourly income earned; hence, ev- peruse the Akash GitHub repo at ery provider is a delegator and/or a val- https://github.com/ovrclk as we continue to idator. open-source various components of the system over time. Tenant: Tenants lease computing cycles of- fered by providers for a market-driven A. Definitions price set using a reverse auction process (described in section below). Akash Token (AKT): AKT is the native token of the Akash Network. The core II. NETWORK OVERVIEW utility of AKT acts as a staking mecha- nism to secure the network and normal- The Akash Network is a secure, transparent, ize compute prices for the marketplace and decentralized cloud computing market- auction. The amount of AKTs staked place that connects those who need computing towards a validator defines the frequency resources (clients) with those that have com- by which the validator may propose a puting capacity to lease (providers). Akash new block and its weight in votes to com- acts as a supercloud platform providing a uni- mit a block. In return for bonding (stak- fied layer above all providers on the market- ing) to a validator, an AKT holder be- place so as to present clients with a single comes eligible for block rewards (paid in cloud platform, regardless of which particular AKT) as well as a proportion of transac- provider they may be using. tion fees and service fees (paid in any of Tenants use Akash because of its cost advan- the whitelisted tokens). tage, usability, and flexibility to move between Validator: Validators secure the Akash net- cloud providers, and the performance benefits

3 of global deployments. Providers use Akash be- stakers as describe in sec. IV A. cause it allows them to earn profits from either Containerized Applications dedicated or temporarily-unused capacity. A unit of computing (CPU, Memory, Disk) is leased as a container on Akash. A con- App C App D App A App B App E tainer [8] is a standard unit of software that packages up code and all its dependencies, so the application runs quickly and reliably from Container Runtime (Docker) one computing environment to another. A container image is a lightweight, standalone, Host Operating System (Linux) executable package of software that includes everything needed to run an application: code, Physical Server (Bare Metal / Cloud VM) runtime, system tools, system libraries, and settings. Figure 1: A simple illustration of Any one with a physical machine (ie, com- containerized applications in relation to the puter, server) can slice the machine’s resources physical servers into containers using a process called virtual- ization. Docker is a company that provides widely adopted container virtualization tech- nology, and it is common to refer to contain- A. Proof of Stake Based Consensus ers as “docker images.” The relation between a physical computer and a container is illus- Akash employs a blockchain secured by a trated in fig. 1). Proof-of-Stake consensus model as a Sybil resis- All marketplace transactions are on the tance mechanism for determining participation Akash blockchain. To lease a container, the in its consensus protocol and implements the tenant (developer) requests a deployment by Tendermint [9] algorithm for Byzantine fault- specifying the type(s) of unit(s), and the quan- tolerant consensus. Tendermint was designed tity of each type of unit. To specify a type of to address the speed, scalability, and environ- unit, the tenant specifies attributes to match, mental concerns with Proof of Work with the such as region (e.g. US) or privacy features below set of properties: (e.g. Intel SGX). The tenant also specifies the maximum price they are willing to pay for each a) Validators take turns producing blocks type of unit. in a weighted round-robin fashion, mean- ing the algorithm has the ability to seam- An order is created in the order book (upon lessly change the leader on a per-block acceptance by a validator). basis. The provider(s) that match all the require- ments of the order then place a bid by com- b) Strict accountability for Byzantine faults peting on price. The provider that bids the allows for punishing misbehaving valida- lowest amount on the order wins (and match tors and providing economic security for requirements), upon which a lease is created the network. between the tenant and the provider for the order. For every successful lease, a portion Anyone who owns an Akash token can bond of the lease amount (Take Fee) is paid to the (or delegate) their coins and become a valida-

4 tor, making the validator set open and permis- sionless. The limited resource of Akash tokens acts as a Sybil prevention mechanism. Vn (t) = dlog2 (2t) · Vi,0 e (1) Voting power is determined by a validator’s So, in 10 years, there will be Vn (10) = 277 bonded stake (not reputation or real-world validators as illustrated in fig. 2 identity). No single actor can create multiple nodes in order to increase their voting power as the voting power is proportional to their bonded stake. Validators are required to post a “security deposit” which can be seized and burned by the protocol in a process known as “slashing”. These security deposits are locked in a bonded account and only released after an “unbonding period” in the event the staker wishes to unbond. Slashing allows for punish- Figure 2: Number of validators over the years ing bad actors that are caught causing any attributable Byzantine faults that harm the well-functioning the system. The slashing condition and the respective III. AKT: THE AKASH NETWORK TOKEN attributable Byzantine faults and punishments are beyond the scope of this paper. (For more information on these, please review Akash Net- The primary functions of AKT are in stak- work Technical White paper). ing (which provides security to the network), lease settlement, and in acting as a unit of mea- sure for pricing all currencies supported by the marketplace. Although AKT can be used for 1. Limits on Number of Validators settling transactions in the marketplace, the leases can be settled using a multitude of to- kens as described in later sections of this paper. Akash’s blockchain is based on Tendermint However, transaction fees and block rewards consensus which gets slower with more valida- are denominated in AKT. The income stak- tors due to the increased communication com- ers earn is proportional to the tokens staked plexity. Fortunately, we can support enough and length of staking commitment. That said, validators to make for a robust globally dis- AKT performs three main functions: Resolve, tributed blockchain with very fast transaction Reward, and Reserve. confirmation times, and, as bandwidth, stor- age, and parallel compute capacity increases, we will be able to support more validators in A. Resolve the future. On Genesis day, the number of validators Vi Akash relies on a blockchain in which a set of is set to Vi (0) = Vi,0 = 64 and the number of validators vote on proposals. Each proposal is validators at time t year will be: weighed by the proposer’s voting power, which

5 is the total tokens they staked and the tokens A. Take Fee bonded to them (stakers can delegate voting power to validators). For every successful lease, a portion of the lease amount (Take Fee) goes to a Take In- come Pool. The Take Income Pool is later dis- B. Reward tributed to stakers based on their stake weight (amount staked and time remaining to unlock, Users of AKT stake tokens to subsidize op- described in detail in the following sections). erating and capital expenditures. Stakers are The Take Rate depends on currency used for rewarded proportional to the number of tokens settlement. The proposed take rates at Gene- staked, the length of lockup time, and the over- sis when using AKT (TokenTakeRate) is 10% all tokens staked in the system. Lockup times and 20% when any other currency(TakeRate) can vary anywhere from one month to one is used. The TokenTakeRate and TakeRate year. Flexibility in lockup encourages stakers parameters is subject to community consensus who stake for shorter periods (bear markets), managed by the governance. in a self-adjusting inflationary system that is designed to optimize for lower price pressure B. Settlement with Exchange Rate during bear markets. Lockin The lease fees are denominated in AKT, but C. Reserve they can be settled using any whitelisted to- kens. There is an option to lock in an exchange Fees on Akash can be settled using a mul- rate between AKT and the settlement currency. titude of currencies along with AKT. How- This protects providers and tenants from the ever, the market order book uses Akash Token price volatility of AKT expected to result from (AKT) as the reserve currency of the ecosystem. its low liquidity. AKT provides a novel settlement option to lock For example, suppose a lease is set to in an exchange rate between AKT and the set- 10 AKT s and locks an exchange rate of tlement currency. This way, providers and 1 AKT = 0.2 BT C. If the price of AKT tenants are protected from the price volatility doubles, i.e., 1 AKT = 0.4 BT C, the tenant of AKT expected to result from its low liquid- is required to pay 5 AKT . Conversely, if the ity. In this section, we also present a mecha- price of BTC doubles while keeping the price nism “Transaction Ordering using Consensus of AKT same, i.e., 1 AKT = 0.1 BT C, then Weighted Median” as described in sec. IV D to the tenant is required to pay 20 AKT . establish exchange rates without the need for an oracle. C. Fees Using a Multitude of Tokens IV. SETTLEMENT AND FEES In order to avoid issues of network abuse (e.g. DOS attacks), all transactions and leases This section describes various fees incurred on Akash are subject to fees. Every transaction by users of Akash Network. has a specific associated fee, GasLimit, for

6 processing the transaction, as long as it does “votes” for the value of each token on-chain as not exceed BlockGasLimit. a transaction. The GasLimit is the amount of gas which Lets say for example, there are five is deducted from the sender’s account balance validators {A, B, C, D, E} with powers to issue a transaction. {0.3, 0.3, 0.1, 0.1, 0.2} respectively. They Unlike most other blockchain platforms that submit the following votes for their personal require fees to be paid in the platform’s native views of each token: cryptocurrency, such as Ethereum [10], Bitcoin A : AKT = 1, BTC = 0.2 [11], and Neo [12], Akash accepts a multitude B : AKT = 2, BTC = 0.4 of tokens for fees. Each validator and provider C : AKT = 12, BTC = 2 on Akash can choose to accept any currency D : AKT = 4, BTC = 1 or a combination of currencies as fees. E : AKT = 1.5, BTC = 0.5 The resulting transaction fees, minus a net- These values are stored on-chain in a ordered work tax that goes into a reserve pool, are list along with their validator that placed the split among validators and delegators based vote. on their stake (amount and length). AKT : [1A , 1.5E , 2B , 4D , 12C ] BTC : [0.2A , 0.4B , 0.5E , 1D , 2C ] D. Transaction Ordering using The proposer takes a weighted mean (by Consensus Weighted Median stake) of the votes for each whitelisted token to determine a consensus relative value of each In order to prioritize transactions when mul- token, where w̄(xn ) = W eightedM ean(xn ) : tiple tokens are used, validators need a mech- AKT : w̄([1, 0.3], [1.5, 0.2], [2, 0.3], [4, 0.1], [12, 0.1]) anism to determine the relative value of the BTC : w̄([0.2, 0.3], [0.4, 0.2], [0.5, 0.2], [1, 0.1], [2, 0.2]) transaction fee. For example, let us assume which give us the relative value for each to- we had an oracle to inform us that the relative ken: AKT = 2.8 and BTC = 0.58 respectively. value of BTC is 200 AKT, and that of ETH is 0.4 AKT. Suppose we have two transactions of equal gas cost, and the transaction fees on V. TOKEN ECONOMICS AND them are 10 BTC and 6000 ETH, respectively. INCENTIVES The first transaction’s fee is equivalent to 2000 (10 x 200) AKT and the second transaction’s Providers earn income by selling computing fee is equivalent to 2400 (6000 x 0.4) AKT. The cycles to tenants who lease computing services second transaction will have a higher priority. for a fee. However, in the early days of the In order to get these relative values with- network, there is a high chance the providers out using an oracle, we can employ a Consen- will not be able to earn a meaningful income sus Weighted Median using Localized Validator due to a lack of sufficient demand from the Configuration [13] mechanism. tenants (consumers of computing), which in In this method, each validator maintains a turn hurts demand because of lack of supply. local view of the relative values of the tokens To solve this problem, we will incentivize in a config file which is periodically updated, the providers using inflation by means of and the relative value is achieved by using block rewards until a healthy threshold can be a weighted mean, meaning they submit their achieved.

7 In this section, we describe the economics of 30 days. Additionally, they delegate (voting mining and Akash Network’s inflation model. power) to validator v by bonding their stake An ideal inflation model should have the fol- via BindValidator transaction. lowing properties: A staker is a delegator and/or a validator to whom delegators delegate. Every provider • Early providers can provide services at is a staker, but not every staker is a provider; exponentially lower costs than in the there can be stakers who are pure delegators market outside the network, to acceler- providing no other services, and there can be ate adoption. stakers who are pure validators providing no • The income a provider can earn is pro- other services. portional to the number of tokens they At any point, a staker can: a) Split their stake. stake (or any piece of their stake) into two • The block compensation for a staker is pieces. b) Increase their stake l by adding proportional to their staked amount, the more AKT. c) Increase the lock time T , where time to unlock and overall locked tokens. T > Tmin . • Stakers are incentivized to stake for Stakers choose to split their stake because longer periods. the compensation is dependent on lock time L • Short term stakers (such as some bear which will be addressed in later sections. market participants) are also incen- tivized, but they gain a smaller reward. • To maximize compensation, stakers are C. General Inflation Properties incentivized to re-stake their income. 1. Initial Inflation A. Motivation If we assume Akash will have the same num- ber of tokens locked as NuCypher [14] and Akash Network aims to secure early adop- DASH [15]: λ = 60%, then 1 − 40% of the tion by offering exponential cost savings as a supply of AKT will be in circulation. The value proposition for tenants, and the efficiency adjusted inflation rate for inflation, I will be: of a serverless infrastructure as an additional value proposition for tenants and providers. I These value propositions are extremely com- I∗ = , (2) 1−λ pelling, especially for data and compute inten- sive applications such as machine learning. Considering that ZCash [16] had I ∗ = 350% (turn around point during the overall bull mar- ket), which makes I = 140% APR, it is rea- B. Stake and Bind: Mining Protocol sonably safe to set the initial inflation to be I0 = 100% APR (meaning 1/365 per day). A provider commits to provide services for at least time T and intends to earn service income r every compensation period Tcomp = 2. Inflation Decay 1 day. Providers stake Akash tokens s and specify an unlock time t1 , where minimal lock- Assume that all miners have the maximum time t1 − t should not be less than Tmin = compensation rate. We define the inflation

8 decay factor (time to halve the inflation rate) 3. Staking Time and Token Creation to be T1/2 = 2 years in this case. Inflation depending on the time passed from the Genesis We will reward the full compensation (γ = 1) t, then looks like: to the stakers who are committed to stake at least T1 = 1 year (365 days). Those who stake   for Tmin = 1 month will get close to half the −T t t compensation (γ ≈ 0.54). In general, I(t) = I0 · 2 1/2 = I0 exp − ln 2 , (3) T1/2 In this case, the dependence of the token min(Ti , T1 )   γ= 0.5 + 0.5 , (6) supply on the time t is: T1 t   I0 T1/2 Z − t M (t) = M0 + I(t) dt = M0 + 1 − 2 T1/2 Ti,initial > Tmin , (7) 0 ln 2 (4) where the unlocking time Ti means the time If we let I0 be the relative inflation rate, then left to unlock the tokens: Ti = t1 − t. t1 is the I0 = i0 M0 . For 100% APR, i0 = 1 and I0 = time when the tokens will be unlocked, and M0 , which gives us the maximum number of t is the current time. The initial Ti cannot tokens which will ever be created (as illustrated be set smaller than Tmin = 1 month, but it in fig. 3): eventually becomes smaller than that as time passes and t gets closer to t1 .   Shorter stake periods (for lower rewards) i0 T1/2 Mmax = M (∞) = M0 1 + ≈ 3.89 M0 , result in a lower daily token emission. Con- ln 2 (5) sidering that miners will most likely stake for short periods during a bear market, we can where M0 is initial number of tokens. expect token emission to decline during a bear market, which will help to boost the price. Therefore we can expect this mechanism to support price stability. The emission half decay time T1/2 ∗ = T1/2 /γ , where γ is the mean staking param- ∗ ∗ eter, is also prolonged when γ < 1. T1/2 pro- longs to 4 years instead of 2 if all stakers have γ ∗ = γ = 0.5. The total supply over time (eq. 4) at γ ∗ = 6 1 will then look like: Figure 3: Token supply and tokens locked over years with an initial inflation of 100% APR that is halving every 2 years i0 γ ∗ T1/2 ∗  − T ∗t   M (t) = M0 1 + 1−2 1/2 . ln 2 (8)

9 D. Delegate Pool Distribution E. Mining strategies and expected compensation The exponential is a solution of a differential equation where inflation is proportional to the In this section, we look at three possibili- amount of not yet mined tokens: ties: a staker liquidating all the compensation while extending the lock time (Liquidate min- ing compensation), a staker adding all the com- ln 2 I(t) = (Mmax − M (t)) (9) pensation to their current stake, and a miner T1/2 waiting for their stake to unlock after time T . Each of these possibilities could have differ- ent distributions of γ. Let’s consider γ = 1 dM = I(t) dt. (10) and γ = 0.5 as the two extreme values of γ. Let’s take the amount of tokens locked to be where M (t) is the current token supply with λ = 60%, as in DASH. M (0) = M0 and dt can be equal to the mining period (1 day). Each validator can trivially calculate its dM using few operations using 1. Liquidate Mining Compensation the token supply M from the last period. The amount of mined tokens for the validator pool In this scenario, all stakers in the pool are p in the time t can be calculated according to liquidating all their earnings every Tcomp pe- the formula: riod. The total amount of tokens staked in the network can be expressed as S = λM . Assume all the delegators have equal amounts of stake sv ln 2 δmv,t = δM (t), (11) bound to the pool. The amount of stake stays S T1/2 constant in this case, and equal to mi = s, making mv = sv and γ = γv where, γv is the X mean staking parameter of the pool. Then, δMt = δmv,t , (12) the pool mining rate (i.e. the cumulative pool v reward) is: where sv is the number of tokens bound to the validator’s delegate pool v and S is drv Sv ln 2 the total number of tokens locked. Instead of = γv (Mmax − M (t)) . (14) dt λM (t) T1/2 calculating all the sum over v, each validator can add their portion δmv,t . When we substitute M (t) from eq. 8 and The distribution factor for a delegate bound integrate over time, we find total pool compen- to pool v is: sation: 1 M (t)   γ s γ κ= + , (13) rv (t) = Sv ln , (15) 2 γv Sv ∗ γ λ M0 γv is the aggregate stake compensation fac- If ∆rv (t) = rv (t) − C where C is validator’s tor for the pool and Sv is the sum of all tokens commission, that brings individual staker’s bound to the pool. compensation to be:

10 1  γ ∗γ λ   γ s M (t)  r(t) = κ · ∆rv (t) = + · ∆rv (t) s(t) = s(0) . (18) 2 γv Sv M0 (16) If γ = 1 (staking for 1 year) and λ = 60% Assuming the validator commission is 1%, (60% of all AKT are staked). With C = 0.1 · if γ = 1 (staking for 1 year+) and λ = 60% r(t), staker compensation in AKT starts from (60% of all nodes in the network are staking), 0.45% per day, or 101.6% during the first year delegate compensation in AKT starts from of staking. 0.45% per day, or s(1) − s(0) = 176.5% during We should note that if other miners stake the first year of staking. for less than a year (γ ∗ < 1), the inflation rate decays slower, and the compensation over a given period will be higher. 3. Take mining compensation and spindown When the node spins down, the staker does not extend the time for end of staking t1 , and the compensation is constantly decreas- ing as the time left to unlock becomes smaller and smaller, effectively decreasing γ gradu- ally towards 0.5. This is the default behavior. To avoid this, the staker should set t1 large enough, or increase t1 periodically. Figure 4: Daily compensation over time assuming 60% tokens locked for lock times of 4. FAQ 1 year and 1 month How many tokens will ever be in exis- tence? We will start with 100 million tokens, and the maximum amount of tokens ever cre- 2. Re-stake mining compensation ated will be 389 million, as illustrated in fig. 3 Instead of liquidating mining compensation, What is the inflation rate? The infla- it could be re-staked into the pool in order tion rate will depend on how many short-term to increase the delegator’s stake. In this case, miners and long-term miners are working in the actual stake s is constantly increasing with the system. Depending on this, the initial in- time: flation will be between 50% APR (if all miners are very short term) and 100% APR (if all min- ds s ln 2 ers commit for a long term). The inflation will =γ (Mmax − M (t)) . (17) decay exponentially every day, halving some dt λM (t) T1/2 time between 2 years (if all the miners are long If we substitute S(t) from eq. 8 and solve term) and 4 years (if all the miners are short this differential equation against s, we get: term). fig. 5

11 21-day “unbonding” period – considered lock up – and there is no incentive to commit for more extended periods. Whereas, stakers in Akash can choose to commit for one month to a year, for which they will receive ~54% and 100% compensation respectively. B. NEO Figure 5: Annual inflation over the years when tokens are locked with long and short According to NEO’s white paper [12]: commitments NEO network has two tokens. NEO representing the right to man- VI. RELATED WORK age NEO blockchain and GAS rep- resenting the right to use the NEO The majority of proof of stake networks such Blockchain. as Ethereum 2.0 [17], Tezos [18], and Cardano [19] all use a single token model. However, At the surface, NEO’s primary utility is a there seem to be networks that are experiment- staking token and GAS is the fee token. How- ing with more novel models. In this section, we ever, after closer observation, NEO’s model is will review some of these systems and explore very different from Akash’s model. the differences with Akash’s token model. Firstly, NEO is used as a mechanism to determine how many votes each NEO account gets without a requirement to stake tokens. A. Cosmos Hub Each account can vote for as many validator candidates as they wish and each validator Akash and Cosmos Hub use Tendermint [9] candidate they vote for receives the number of Consensus Algorithm and share a core set of votes equivalent to the amount of NEO in the values with interoperability and user experi- voter’s account. ence. Similar to Cosmos’s Atom [13], AKT’s With regards to the fee, NEO’s chain only primary utility is to provide economic security supports a single fee token, unlike Akash’s to the network. Akash’s model variously im- multi-token model. Furthermore, unlike proves Cosmos’s model. First, AKT provides a Akash, NEO does not provide volatility pro- mechanism to normalize compute prices for the tection for the GAS tokens. marketplace auction. Secondly, Akash intro- duces a mechanism to lock in an exchange rate to a reserve currency of choice to mitigate mar- C. EOS ket drive volatility risk of AKT when leasing computing for more extended periods. Finally, EOS’s delegated proof of stake consensus [20] Akash’s block reward distribution is propor- has similarities with Akash’s model but is ex- tional to the time and amount of a stake, unlike tensively different. In EOS, each token holder Cosmos’s model where the distribution is ho- can stake their tokens in order to vote for block mogeneous for a fixed time. Cosmos imposes a producer and in return, they are rewarded in

12 resource units such as CPU, RAM, and NET //www.marketsandmarkets.com/PressRelea that can be spent for transactions on the net- ses/cloud-infrastructure.asp work. However, like in NEO, the staking token [3] “Cisco Global Cloud Index: Forecast EOS is not staked by the block producers, and and Methodology, 2016–2021 White Paper” it is not slashable in the case of misbehavior. [Online]. Available: https://www.cisco.co In EOS, staking means, stakers are putting m/c/en/us/solutions/collateral/service- tokens in a lockup period and not necessarily provider/global-cloud-index-gci/white-paper- contributing to the functionality of the net- c11-738085.html work. Stakers earn rewards in CPU, RAM, and [4] J. Kaplan, N. Kindler, and F. William, NET that are used to purchase computational “Revolutionizing Data Center Efficiency McK- resources on the network. These resources are insey and Company.” [Online]. Available: not transferrable. CPU and NET are only https://www.sallan.org/pdf -docs/McKin spendable by the receiver, whereas RAM can sey_Data_Center_Efficiency.pdf be traded with other users in a Bancor-style [5] “Uptime Institute Comatose Server Sav- marketplace [21]. ings Calculator.” [Online]. Available: https: EOS burns these resources upon spending, //uptimeinstitute.com/resources/asset/coma instead of giving them to block producers. The tose-server-savings-calculator validator compensation model is unclear, con- [6] “Prime Leverage: How Amazon Wields sidering transaction fees is not the primary Power in the Technology World” [Online]. mechanism. EOS is seemingly a single to- Available: https://www.nytimes.com/20 ken network, despite having nuances and addi- 19/12/15/technology/amazon-aws-cloud- tional steps. competition.html [7] “RightScale 2019 State of the Cloud Re- port.” [Online]. Available: https://www.fle VII. CONCLUSION xera.com/about-us/press-center/rightscale- 2019-state-of-the-cloud-report-from-flexera- This paper explains the network and min- identifies-cloud-adoption-trends.html ing economics of Akash Network and presents [8] “What is a Container?” [Online]. Avail- various incentives and utilities of different to- able: https://www.docker.com/resources/wha kens in the staking and fees mechanisms. The t-container Akash Token (AKT) acts as staking token and [9] E. Buchman, J. Kwon, and Z. Milsosevic, reserve currency for the network while using a “The latest gossip on BFT consensus” [Online]. multitude of tokens for settlement. Available: https://arxiv.org/abs/1807.04938 [10] G. Wood, “Ethereum: A Secure De- centralised Generalised Transaction Ledger.” [1] “Worldwide Market Share Analysis: IaaS [Online]. Available: https://gavwood.com/pa and IUS” [Online]. Available: https://www.ga per.pdf rtner.com/en/newsroom/press-releases/2019- [11] N. Satoshi, “Bitcoin: A Peer-to-Peer 07-29-gartner-says-worldwide-iaas-public-cl Electronic Cash System.” [Online]. Available: oud-services-market-grew-31point3-percent- https://bitcoin.org/bitcoin.pdf in-2018 [12] “NEO Whitepaper.” [Online]. Available: [2] “Cloud Infrastructure Market - Global http://docs.neo.org/docs/en-us/basic/white Forecast to 2022” [Online]. Available: https: paper.html

13 [13] S. Aggarwal, “Cosmos Multi-Token amending crypto-ledger.” [Online]. Available: Proof of Stake Token Model” [Online]. Avail- https://tezos.com/static/white_paper- able: https://github.com/cosmos/cosmos/b 2dc8c02267a8fb86bd67a108199441bf.pdf lob/master/Cosmos_Token_Model.pdf [19] A. Kiayias, A. Russell, B. David, and [14] M. Egorov, M. Wilkinson, and, R. Oliynykov, “Ouroboros: A Provably Secure “NuCypher: Mining & Staking Economics” Proof of Stake Blockchain Protocol.” [Online]. [Online]. Available: https://www.nucyph Available: https://iohk.io/research/paper er.com/static/whitepapers/mining-paper.pdf s/#ouroboros-a-provably-secure-proof-of - [15] E. Duffield and D. Diaz, “Dash: A stake-blockchain-protocol Payments-Focussed Cryptocurrency.” [Online]. [20] D. Larimer, “EOS: Technical Whitepa- Available: https://github.com/dashpay/dash/ per.” [Online]. Available: https://github.com wiki/Whitepaper /EOSIO/Documentation/blob/master/Tec [16] “ZCash Emmission Rate.” [Online]. hnicalWhitePaper.md Available: https://z.cash/technology/ [21] “EOS RAM 101: Non-Technical Guide- [17] “Ethereum 2.0 White Paper.” [Online]. book for Beginners.” [Online]. Available: Available: https://github.com/ethereum/wi https://medium.com/coinmonks/eos-ram- ki/wiki/White-Paper 101-non-technical-guidebook-for-beginners- [18] L. M. Goodman, “Tezos: a self- 6f971322042e