Disclosure: Multicoin has established, maintains and enforces written policies and procedures reasonably designed to identify and effectively manage conflicts of interest related to its investment activities. Multicoin Capital abides by a “No Trade Policy” for the assets listed in this report for 3 days (“No Trade Period”) following its public release. At the time of publication, Multicoin Capital holds Arweave ($AR) tokens.
Editor's Note: This essay was co-authored by Tushar Jain, co-founder and managing partner at Multicoin Capital, and Spencer Applebaum, associate at Multicoin Capital. Multicoin Capital is a thesis-driven investment firm that invests in cryptocurrencies, tokens and blockchain companies.
Over the past decade, a number of businesses have come up with ingenious ways to put empty properties and other “idle” assets to better use. Airbnb shocked the hospitality industry by making money off of otherwise empty homes and bedrooms. Uber upended the taxi industry by using otherwise idle cars as taxis. Building on this theme, decentralized storage networks aim to disrupt the cloud storage industry by increasing the use of otherwise unused computer storage.
Cloud storage is a market worthy of challenge. Over the last decade, the new cloud paradigm has displaced the legacy on-premise servers, birthing new tech behemoths like Amazon Web Services (AWS), Alibaba Cloud, Microsoft Azure, Google Cloud Platform (GCP) and many others. Today, market research firms estimate that the global market for cloud storage will reach $137.3 billion by 2025 while growing 22.3% per year. This market is enormous.
Recognizing this opportunity, many teams began building open source solutions in late 2016 and early 2017 including Filecoin, Storj, Sia and SAFE. In 2018, Arweave launched a mainnet based on similar ideas around decentralized storage, but with a different purpose: permanent storage. Unlike the other decentralized storage networks—which aim to compete with legacy cloud providers on cost and/or performance—Arweave uses permissionless crypto-economic incentives to create a new kind of service that wasn’t possible before.
Filecoin mining is booming in China, and FIL is now trading at around $28, implying a fully diluted network valuation of $56 billion [at time of writing, Oct. 21]. The mainnet just launched, the market is excited and hype is at an all time high. Meanwhile, Arweave recently raised capital from prominent investors including Andreessen Horowitz, Union Square Ventures and Coinbase Ventures. In April 2020, Arweave 2.0 launched, beating Filecoin to market, and it has been growing consistently since.
As investors evaluate the forthcoming Storage Wars, big questions loom. Which network will developers choose, if either? Can decentralized storage networks compete on cost against Web2 giants like Amazon and Alibaba? How should investors compare the two approaches to decentralized storage?
In this essay, we explore these questions and provide a framework to evaluate the decentralized storage market. This essay is not investment advice, but rather a framework for evaluating one of the most exciting markets in Web3. With that backdrop, the first step is understanding the differences and trade-offs between Filecoin and Arweave.
Contract-Based vs. Permanent Storage
Both Filecoin and Arweave enable decentralized, trust-minimized, censorship-resistant data storage. Both are built using blockchain technology. And both networks can be used to store data for long periods of time, either for archival purposes or for real time applications like website hosting. At first glance, they are quite similar. So rather than start by evaluating Filecoin’s and Arweave’s respective blockchains, a better place to begin is how they intend to offer storage to end users.
Filecoin's economic model mirrors that of centralized cloud providers: contract-based storage. Contract-based storage can be more simply thought of as a pay-as-you-go model. Users pay a network of nodes that store X bytes of data for Y period of time with Z retrievability guarantees. Storj, Sia and SAFE use the same model.
Arweave, on the other hand, introduces an entirely new economic model to the market, one that was never possible before the advent of permissionless crypto networks: permanent storage. With permanent storage, users pay a one-time, up-front fee to store the data forever. Permanent storage creates an entirely new market (we’ll get to this later). The Arweave protocol accomplishes this by leveraging crypto-economic game theory and creating an endowment to compensate miners for ensuring data availability, reliability and permanence.
The Filecoin Proposition
Filecoin and other contract-based decentralized storage protocols (Sia, Storj) primarily compete on cost. They claim to be able to offer lower costs than centralized providers because they utilize otherwise idle hard drive space. These networks also offer a higher degree of censorship resistance than traditional cloud storage providers.
Given Filecoin’s clout, it’s reasonable to expect that it will capture some share of the Web3 storage market. However, in practice, it’s unlikely that it will be able to sustainably undercut Amazon’s pricing. Filecoin will effectively subsidize costs for storage buyers in the early days with FIL issuance, but (1) printing tokens to subsidize costs cannot last forever without adversely affecting token price, (2) Amazon, Alibaba, Tencent, Microsoft and Google have a lot more capital that they can use to subsidize prices if it comes down to a price war, (3) Amazon cross-subsidizes their S3 business lines such as storage, compute, database, etc.
In every one of Amazon’s business units, they vertically integrate and radically slash costs. For example, Amazon has steadily increased its footprint in the shipping value stack, and today Amazon ships upwards of 50% of its own packages in the U.S. (whereas just a few years ago Amazon relied on third parties for 100% of shipments). They take a similar approach in cloud services, offering everything that developers need at massive scale and low cost. Storage customers benefit from these economies of scale (as per the chart below), but take on platform risk.
How does Amazon accomplish this? They can negotiate bulk purchase agreements that allow them to access cheaper storage and power. Per a study called Cost to Support Compute Capacity conducted by Ponemon Institute LLC, the average annual cost/kW of power ranges from $5,467 for data centers over 50,000 square feet to $26,495 for data centers between 500-5,000 square feet. Amazon also purchases hardware in large bulk, which gives them access to cheaper parts that they then pass on to cloud customers.
Today, it’s challenging for an individual with idle storage space to participate on the supply side of the Filecoin network (the recommended hardware is 128-256 GB of RAM and a 24 Core CPU). As such, the network at launch is dominated by miners who have invested large amounts of capital in powerful mining machines. We know from Space Race, Filecoin’s testnet, that at least 145 of 230 PiB of storage was provided by the top 10 miners. It is unlikely that these medium-sized mining facilities will have lower data center costs than Amazon, which runs some of the largest data centers in the world.
Based on hardware configuration, Filecoin miners cannot outcompete traditional cloud storage providers on cost. Thus, the primary “feature” of Filecoin is not cost, but rather ideology and a greater degree of censorship resistance.
The Arweave Proposition
The Arweave Protocol provides permanent storage as a service. It does so not by creating contracts between users and storage providers, but by creating crypto-economic incentives for miners to replicate as much data as possible. Permanent data storage is an entirely new service that Amazon, Google and others cannot offer.
To store a file on Arweave, a developer creates a transaction that pays some amount of AR tokens as a network fee (currently $5/GB) to store data forever. In comparison, Amazon S3 charges $0.276/GB per year for their low tier pricing, implying that Arweave is 18x as expensive as Amazon. Arweave does not claim to compete with Amazon on cost. Arweave users are explicitly paying a premium for something that Amazon cannot offer—permanent storage.
We have identified two market segments that are adopting Arweave now. Players in these two sectors are less price sensitive and need permanent storage:
Blockchains for data availability - Blockchains are meant to store the history of a transaction network forever. Arweave enables Layer 1 and Layer 2 teams to store a copy of their ledger permanently. This is key for auditability and redundancy. Several prominent teams such as Solana and SKALE are finalizing their integrations with Arweave for this purpose now, and we expect more chains to announce similar initiatives in the coming months.
Internet Archiving - The Internet Archive recently announced that they will leverage Arweave to fulfill their mission; they are the non-profit organization that hosts the famous Wayback Machine project, which hosts old websites even after their original creators take them down. While it’s not widely known, “link rot” is a massive problem. For example, over 49% of links cited in U.S. Supreme Court decisions are broken, according to research in 2013. The internet is exhausting data at a compounding rate. As more data is created, more links break.
While we do not know how large the market for permanent storage is, we do know that popular blockchains will need to store massive amounts of data once they achieve web scale. We also know that humanity creates a massive amount of data, and there are organizations that actively seek out ways to store humanity’s history in fault tolerant ways.
There are several other markets that we can reasonably expect to value permanence of data storage, and be willing to pay a premium for it:
Journalists who want to make sure their reporting is available forever to shine light on the truth;
Political dissidents who want to ensure that governments can’t censor their thoughts;
Lawyers working on personal estates or trusts;
NGOs or foundations who want to store their records forever;
People who want to store personal memories for distant future generations.
Lastly, and most importantly, we expect Arweave will enable the creation of new kinds of unstoppable applications that rely on permanent, immutable storage. The Arweave team has been fostering the development of this nascent ecosystem for years. At the Fall 2019 Multicoin Summit, Arweave founder Sam Williams gave a presentation outlining how the Permaweb will reshape the web as we know it. Today, Arweave is now in a similar stage to where Ethereum was in early 2016. There are dozens of developers building new kinds of Arweave-based apps including: ArDrive, Limestone, Evermore, Nest.land, Non-Zone, ArGo, Outpost, OpenBits, Verto, WeaveID and more.
Relative Network Valuations and Token Economics
Filecoin’s native token, FIL, has two functions:
Miners must stake FIL as collateral in exchange for the ability to host files. They must stake 0.1901 FIL per 32 GiB of data that can be converted into storage mining power.
FIL is used as a medium of exchange.
Payment tokens should be valued using the MV=PQ equation. For a simple model, we can use Filecoin’s own estimate of a $75 billion market size by 2021, and the USD M1 velocity of 3.9 (which is generous considering FIL is subject to the velocity problem). We use a velocity of 3.9 because there is a velocity sink in which storage nodes must stake FIL tokens to earn FIL-denominated rewards. Using a velocity of 3.9, were Filecoin to capture the entire market for cloud storage over Amazon and Alibaba, the FIL market cap would rationally be worth $19 billion ($75 billion/3.9) — 66% lower than where it trades today ($28 per FIL or a $56 billion fully diluted network valuation, at time of writing).
Given this analysis, it’s easy to see how Filecoin’s implied fully diluted network value is far ahead of its fundamentals.
On the other hand, Arweave’s native token, AR, is about $2.69, implying a fully diluted valuation of $178 million. It’s important to note that AR tokens are subject to a powerful token velocity sink (more so than FIL’s circular payment structure in which storage nodes escrow FIL tokens to receive FIL tokens). When a user pays to store data on the Arweave network, they don’t just pay the miner the storage fee, which the miner can then sell for USD. Instead, more than 83% of the fee goes into an endowment pool. The endowment pool is slowly paid out to miners over time for storing data. For each file, this endowment pool slowly approaches 0 over time, but never actually reaches 0. This is possible because the cost of storage falls over time. Over the last 50 years, the cost of storage has decreased at about 41% per year. Therefore, as demand for permanent storage grows, users buy AR and then lock it in the endowment, creating continued buying pressure. We believe AR’s economic model captures value much more effectively than FIL.
Or said another way, Arweave—as a fully-launched, fully-functioning network with novel token economics—is valued at .31% of Filecoin today. The chart below demonstrates the discrepancy. Filecoin is the red line. Arweave is the blue line.
The Storage Wars
It’s impossible to say if either of these networks and their valuations will grow or shrink, if Filecoin will achieve its vision of disrupting AWS and Alibaba or if Arweave will still be around in 20 years to serve files that users pay to store today.
What is known is that Filecoin has a thriving community and a lot of early investors who participated in its $257 million ICO in 2017. It’s also known that Filecoin has been under development for the past three years and that Juan Benet is an absolutely amazing engineer. The Filecoin team has built both the Interplanetary File System (IPFS) and CoinList. IPFS, which is designed to work with Filecoin, demonstrates the team’s competence in creating large-scale digital infrastructure, while CoinList showcases their ability to create well-functioning user-facing applications. Both of these characteristics will be paramount for success.
On the other hand, Arweave is interesting because it serves an entirely new market, one that is uniquely enabled by blockchain technology, and one that caters to both the existing storage market and unlocks new ones. The network is live and is growing. Decentralized application interfaces from projects like SushiSwap, Uniswap V2 and yearn.finance recently adopted Arweave, paving the way for thousands of applications to follow. This will all ultimately work to accelerate Arweave’s growth.
In summary, after examining (1) the technical and economic designs of both Filecoin and Arweave, (2) the present-day dynamics in the cloud storage markets, (3) the fact that Arweave unlocks a new market and differentiates on features beyond price and (4) that AR is valued at about .31% of FIL, we believe there is a compelling relative value play here.