This essay is meant to explore the decentralized cloud computing space. Akash is one of the leaders in the space and Akash is used as a sample to build an investment thesis. This is not a recommendation pitch.
About Akash
- Akash network is a decentralized cloud computing marketplace that connects those who need computing resources (clients) with those that have the excess computing capacity to lease (providers). Akash also runs the platform to provide the infrastructure required to host and run workloads
- AKT (Akash Token) is the native token used for staking, rewarding, and lease settlements. Settlements can be made in other currencies too
- Akash enables the offering of computing resources at a fractional price compared to that offered by centralized competitors such as AWS, Google, and Azure. The network generates income in the form of transaction fees and take-rate (as % of lease amount)
Below is the description of the team:
Akash can be classified as shown below:
To understand how Akash is built and how does its technology stack look like, refer to the diagram below:
Stakeholders
There are different stakeholders running the network. These are:
Validators secure Akash Network by validating transactions and verifying blocks. The network will commence with 64 validators which will grow over time and reach 277 validators by the tenth year (and continue increasing in line with this). Validators are incentivized by charging delegators a commission fee in Akash Token.
Delegators are holders of Akash Token that stake their tokens for the purpose of securing the network. Delegators receive a proportion of the transaction fee as well as block rewards in exchange for staking.
Providers are those that offer their under-utilized capacity on Akash Network and earn income for their contributions. Providers lease these computer units (CPUs) as container units on Akash. A container unit is a standard unit of software that packages up code and all its dependencies. This model ensures that applications run quickly and reliably from one environment to the other. Providers are required to maintain a stake in Akash Token as collateral which is proportionate to their hourly income earned.
Tenants are those individuals that lease capacity offered by providers. They can discriminate between providers by requesting specific regions, privacy features, and the maximum price they are willing to pay for the lease. The provider is chosen by way of an auction process.
These stakeholders interact in the following way as shown below:
Why decentralized cloud is an attractive proposition??
- The decentralized cloud computing market is an attractive opportunity in the web 3.0 regime where technology-enabled applications will like to be truly decentralized by breaking away from centralized cloud providers
- Moreover, cloud services will remain an essential commodity to power applications in the future. Hence, demand will not be a problem?
- The current dominance of centralized cloud providers poses challenges and threats as mentioned below?
- As blockchain-based applications’ adoption is increasing, various building blocks of the stack are also getting decentralized to enable true decentralization as shown below?
How the decentralized cloud market might look like at scale??
Akash’s positioning
Akash operates in both the storage and computing segment. Competitors can be categorized into 3 categories – (1) Only storage, (2) Only computing, and (3) Both storage and computing. The storage category has seen relatively more scaled players followed by only compute and both storage and computing.
Competitors comparison
Akash leads on many relevant parameters. Regarding the active providers' parameter, it will be useful to understand how many providers are providing the lion-share of computing resources for both Akash and Golem. However, I could not get that data.
Akash value proposition
Another big value proposition is the low computing cost provided by Akash compared to other centralized services
Business model
- Akash solves the critical pain point of cloud service-seeking customers by providing cost advantage and easy to deploy the solution
- Akash makes money by charging take rates from leases and incentivizes network participants to maintain network integrity
- Akash’s business model is de-risked by adequate incentivization of network participants, liquidity provisions, and measures to control volatility??
Token-economics
- At Genesis, 100M token was supplied. The 2.5 years-release schedule of the tokens is displayed below.
- Overall, there will be 388,539,00 supply of AKT tokens. The split of tokens in the steady state will be as shown in the table
- The steady-state supply distribution is attractive with 2/3rd of the supply dedicated towards network participants
Operating metrics & major partnerships
Operating metrics have fallen from their peak as activities across dApps have slowed down in sync with price correction in major currencies.
Akash Network will be able to run smart contracts on Solana’s smart contract platform. By interoperating with Solana, users will be able to run dApps and settle using the Solana token on Akash
Dapps from Polygon can run on Akash. More than 70% of Polygon’s 700 dApps are related to decentralized finance (DeFi) and provide investing, lending, and various other financial services. These include DeFi bluechip dapps such as Sushiswap, Curve, Aave, Balancer, and Kyber.?Polygon developers are encouraged to apply for up to $100,000 from the Akash Developer Grant Program
Supporting inter-blockchain compatibility will enable Akash Network to operate with multiple blockchains more easily. ThorChain has already promoted the movement from centralized cloud platforms to Akash Network; and
Oracle provider Chainlink will be utilized to ensure that Akash Network’s users have access to the most secure and accurate price data available which will support decision-making processes on the cost of lease fees.
Revenue/cash flow potential for the network (in 10 years)
What do we need to believe in?
- Decentralized cloud service providers can provide service features and quality such as up-time similar to that provided by the centralized providers. This will help in the adoption
- Decentralized cloud services are currently used by smaller-scale projects. We need more proof points for it to cross that valley and be able to serve enterprise clients in order to become a meaningful market
- If there are 4-5 dominant players compared to 2-3 players, then there will be pricing pressure and the network will have to settle for a lesser take rate, which will reduce revenue/token number at scale
- Cashflow per token is expected to be $44 and $4 in bull and bear cases, respectively
Valuation multiples for similar businesses
- Exchange businesses and other technology platform businesses are used as comparable because of their 2-sided market business models. Both blockchain and non-blockchain-based firms in the above categories are used
- Protocol revenue is the demand-side revenue or the revenue kept by the network. It is similar to FCFF
- 19x seems a reasonable multiple to assume for valuing Akash because at scale Akash will become a large cap 2-sided market business and the growth will be slower
Valuation and return analysis
Assuming that the decentralized cloud adoption happens, the return potential is huge with 647x in the bull case and 26x in the bear case
Roadmap
- Unbounded Network Bandwidth: Workloads running on Akash have a fixed cap with bandwidth usage. This prevents data-intensive applications from using Akash. Removing this limit by creating a bandwidth market enables more use-cases for Akash
- UI Component Library: The UI component library is an open-source repo of code components. Because engineers aren’t always designers, their Akash initiatives have inconsistent branding and UI experiences, leading to confusion and mistrust from users
- SSL Support: Akash currently does not provide a mechanism to attach x.509 certificates to workloads and defers to an external router (Cloudflare) to provide SSL capability. This introduces additional friction as one is required to find external support.
- Provider Onboarding Framework: Better capacity plan and revenue visibility for providers
- Developer Hub: Communicate system architecture, important concepts, and value propositions
- Lease Renegotiation: Renegotiating the contract during the lease life cycle provides stability with computing prices
- Log retention: For better debugging facility. Currently, logs get lost when a container restarts as they are currently streamed from Kubernetes. The logs will be shipped to a 3rd party log storage system such as Loki
- Akash Javascript SDK: Javascript SDK simplifies extending Akash in Javascript applications by providing libraries to improve access to the Akash API
- Deployment GUI: deployment web interface will enable the adoption of Akash to beginner and intermediate users, thus growing the network
- GPU Marketplace: Unlock new uses cases such as ML by adding the ability to source GPUs
- Provider Management Dashboard: Managing a provider on Akash is currently only available with CLI which is cumbersome and limited
- Overlay Network: Overlay network connects workloads deployed on Akash in a private network with private IP space, removing the onus on the user to secure the network
- Sovereign Docker Registry: Akash currently supports public docker containers hosted on external registries. Public containers can leak information, but private registries have better security features
- Managed Services Market: Market for Managed Backend Services (such as Databases) to reduce the operational burden for Tenants and to increase adoption of the network?
Future diligence points and Risks
- How will the lease renegotiation proposed in the road map work? Is it a balanced interest or favors either of the parties, i.e., providers or tenants?
- What are the current and future initiatives to onboard providers?
- What is the bottleneck today – is it getting more providers or tenants? Conversely, which side of the market is more influential in driving adoption?
- If switching cost will be less in decentralized cloud providers, how will be the stickiness built in the business?
- Since Akash does not pool resources, tenants have 1-to-1 relationship with providers. If a tenant requires a large computing resource, it has to sign many leases unless there’s a big provider. Many leases increase the management hassle. What is the workaround for this problem?
- Dive deep into performance benchmarking against the centralized counterparts, and assess the gaps, if any. We can compare variables such as - the ability to cancel and spin up deployments in seconds, pay per use model, geographical distribution
- How to protect/de-risk a provider going offline? Below are current solutions but these either come for an additional cost or are not much reliable
- Have replica deployments across more than one provider and have a load balance or simple DNS round robin between the deployments which would mean one going offline would not be impactful.
- Choose providers that have been audited by Akash, knowing these providers are far less likely to go offline.
- Negative network effect: If adoption in any one side of the market is not picking up, then the negative working effect will lead to withdrawal on the other side of the market.
- Who are the top 10 providers on Akash Network, and how many providers contribute to the majority of the computing resources on Akash?
- 7 validators hold 35% of the voting power and to get a quorum only 40% of the total staked tokens are required. Hence, governance is not that strong yet. How can governance be made stronger? (Source)
- Since Akash Network is built using Cosmos SDK, Akash Network faces L1 protocol risk. Strength and adoption of the Cosmos ecosystem will affect Akash’s performance?
