Aptos vs Sui: L1s on the Move

Cody

October 18, 2022

Featured

1 Comment

[Disclaimer: GCR participated in Mysten Labs’ (Sui) Series B on 9/12/2022]

I. Introduction

The ‘L1 Wars’ were a highly publicized narrative from the 2020-2021 cycle, with major alternative Layer-1 blockchains capturing a tremendous amount of value and developer talent alongside Ethereum. Solana, Cosmos, Avalanche, BSC, and nearly a dozen other L1s held a market cap > $300B near the market’s peak. While this narrative has cooled and the rotatooors have hunkered down for the bear market, conversations around web-sized scalability, security and continued innovation around technical design have again been brought to the forefront. These conversations have only been supercharged by the gigantic funding rounds of two new L1 unicorns – Aptos and Sui – whose teams and architecture both derive from Meta’s now defunct Diem and Novi projects.

> Diem & Novi

Novi Financial, a wholly-owned subsidiary of Meta, as well as a member of the Diem Association, was the primary project being built by the digital asset team at Meta for Diem. It was a crypto payments wallet built from the same technology behind its predecessor, Calibra which was also announced by Facebook in June 2019. As the Diem Association began winding down in the face of continued regulatory scrutiny and lack of mainstream excitement to adopt a product contradictory to the overarching ethos of web3 and decentralization, the Novi pilot followed a similar fate.

Though the Meta-backed projects were sunset into their graves at 1 Hacker Way, the teams behind the technology have big ambitions to bring it back in radically differentiated ways. Most notably, where Diem aimed to be a permissioned network, Aptos and Sui are decentralized and permissionless, allowing anyone to join them as a validator.

Diem, initially known as Libra, was a permissioned blockchain-based stablecoin payment system proposed by Facebook (Meta) to be independently and cryptographically entrusted to the Diem Association– a membership organization of companies from payments, technology, & telecommunication to venture capital, and nonprofits – including Visa, Mastercard, eBay, Shopify, Anchorage, Coinbase, Ribbit Capital, a16z, Union Square Ventures, Mercy Corp, Uber, Lyft and many others.

> Moving Out

Move is an open-source programming language based on Rust that was developed by the Diem Association team to create customizable transaction logic and smart contracts. It is a novel language along with the Move Virtual Machine (MoveVM) that came out of the project. The mechanism was designed to maximize safety, without adding the cost of compilation to transactions, minimizing gas fees as compared to other PoS chains like Ethereum. It is touted as a safe and highly efficient language that can help devs avoid mistakes that could lead to exploits.

According to the whitepaper, the executable format of Move is “bytecode that is higher-level than assembly yet lower-level than a source language. The bytecode is checked on-chain for resource, type, and memory safety by a bytecode verifier and then executed directly by a bytecode interpreter.” The major factor that separates Move from other programming languages and allows Move to enjoy a high level of both security and expressivity is its use of resources, which is drawn from the mathematical idea of linear logic. To dig in deeper on the technical nuances of this language, Pontem Network– an Aptos product studio building a wallet and asset swap on Aptos– outlines it in much more detail here.

While both protocols leverage the Move programming language, Aptos and Sui employ slightly different models. Where Aptos utilizes the original MOVE language created by the Diem team, Sui is leveraging their own alternative version known as “Sui Move”.

II. Funding Overview

Aptos and Sui have seen a meteoric rise in mindshare on the backs of a few eye popping fundraisings rounds. Aptos Labs’ $200M strategic investment led by a16z in March 2022 followed by a $150M Series A in July led by FTX Ventures came sandwiched between Mysten Labs’ $36M Series A in December 2021 and more recent $300M Series B in September also led by FTX Ventures. Binance Labs and Dragonfly joined the party, each with strategic investments into Aptos in late September after Mysten announced their latest raise.

Building a new L1 is a massive undertaking with no perfect playbook. It is capital intensive, talent intensive, and takes years to build, scale, market, and move towards progressive decentralization. These recent funding rounds feel like a capital arms race to see who can build the biggest bear market war chest that produces meaningful innovation in the next expansion cycle. That said, there has been some loud criticism from the community and investors that these valuations are too high given the lack of a live ecosystem, governance framework, or users to back their 2x unicorn status in the depths of a deep bear market.

The recent fundraises tell another interesting story. Exchange funds are piling into both projects, with FTX, Coinbase, and Binance each contributing to the recent rounds for both protocols. Their parent exchanges will be listing these tokens and gaining a tremendous amount of fees from traders rotating into yet another L1 narrative, while also capturing token allocations at an early discount. In addition to the exchange funds, it’s notable to see the caliber of funds backing these protocols– with 15 top funds from our recent VC rankings filling the cap tables. A number of Solana investors also find themselves funding these protocols that are similarly looking to maximize network capacity, including: Multicoin, Blocktower, Sino, ParaFi, a16z and Jump. While the exorbitant valuations are still being digested with skepticism, early backers are hoping for outsized returns via future token warrants.

III. Founder-Market Fit

The founding teams from both Aptos and Sui are deeply rooted in the Diem and Novi projects. There is no denying that both projects have the technical savvy to execute on the visions that have already amassed a combined valuation exceeding $4B. Both teams undoubtedly have a strong founder-market fit. The Mysten Labs team includes two co-authors of the MOVE programming language whitepaper as well as more R&D firepower from the Novi/Diem teams. Mo Shaikh as CEO of Aptos Labs brings business development and partnerships chops– an important aspect for any new protocol looking to gain traction in the ever growing L1 market. This focus was a clear differentiator for Solana and Polygon last cycle vs more technically minded protocols like NEAR and Algorand. Aptos also brought in several Solana alumni who have first hand experience scaling an alt-L1 including Solana’s ex-Head of Marketing to drive the Aptos ecosystem.

IV. Architecture & Consensus

Aptos and Sui are both Proof of Stake (PoS)-based blockchains and both protocols utilize a consensus mechanism called BFT, or Byzantine Fault Tolerance– operating on the idea that one-third of validators can go offline or have malicious intent and the network can still operate normally. From here the designs begin to diverge.

> HotStuff

Aptos relies on a modified BFT consensus HotStuff. In HotStuff a leader that changes with every round of voting suggests a new block and validators vote on it. Since all validators communicate with a single leader, the total number of messages sent out is much lower than if the validators communicated with each other. Once a block is agreed to be valid it reaches finality in what Aptos claims will be < 1 second. This is the key advantage marketed by Aptos over competitor Solana where finality can take ~2-6 seconds.

Similar to most L1s including Etheruem, Aptos leverages an account-based model where transactions are sequentially packaged into blocks and formed into blockchains. However, by designing Block-STM, an in-memory parallel execution engine, Aptos claims to be able to execute over 160k non-trivial Move transactions per second (130k demonstrated via a speed test) “by leveraging the preset order of transactions and combining Software Transactional Memory techniques with a novel collaborative schedule” [1]

> Narwhal & Tusk

Sui leverages a Narwhal & Tusk consensus algorithm for parallelization at the execution layer. Narwhal is the mempool module– it makes sure that transaction data is available. It can also be used on its own (without Tusk) with other consensus engines, such as HotStuff or Cosmos’ Ignite [2]. Tusk is the consensus module that orders the complex transactions submitted for consensus [3].

Given this DAG-based (directed acyclic graph), or object-centric data model, some would likely consider Sui a distributed ledger rather than a blockchain. With this design, transactions are not packaged sequentially into a blockchain, instead its many elements are linked together as a networked graph rather than a chain. As you can imagine, the DAG model can enhance scalability by splitting objects and exploiting their built-in properties. This design is also asynchronous which means it can withstand DoS (denial of service) attacks, and helps Sui market as a security-centric protocol.

With this design, Sui takes a dynamic new approach by eliminating consensus for many transactions altogether– simple token transfers by their owner to a different address, without any other transactions depending on them, are confirmed virtually instantly. The sender broadcasts the transaction, collects validators’ votes (handshakes), and receives a so-called certificate of validity. For more complex transactions involving ‘shared objects’ such as interactions with a smart contract that can be modified by more than one owner, Sui utilizes the more traditional BFT consensus outlined above. This approach could make Sui a desirable L1 for specific use cases where a dApp generates a large number of simple transactions, and needs them to be confirmed with low latency and is less concerned with decentralization– such as games or airdrops [4].

A big shout out to Pontem Network, whose blog helped me synthesize these highly technical sections.

> Scalability

Having dug into these novel protocol designs, it’s clear that scalability is the major value add to the ecosystem that these projects look to unlock. Along with major conversations taking place around what web-sized scalability looks like in blockchain, both Aptos and Sui aim to become leaders in this race by maximizing network capacity. The below chart highlights how early speed tests match up to the established L1 competition. Time will tell if the teams are able to execute on their early promises.

V. Traction

It’s still far too early to tell how the MOVE Wars will shape out for Aptos and Sui. However the large capital inflows will no doubt supercharge traction for both protocols, with early signs of ecosystems already beginning to emerge. There are already 181 projects being built on or supporting integration of the Move programming language according to Move Market Cap.

> Aptos Ecosystem

A number of Aptos specific browser wallets have launched in testnet since the late summer fundraising, including Pontem Wallet, Fewcha, Martian and a few others are available in Chrome and iOS. Pontem Network, the Aptos product studio behind the aforementioned wallet also launched LiquidSwap, the first AMM on Aptos. In addition, a few NFT marketplaces have been built on the network with BlueMove, Topaz and Souffl3 (which currently only supports the Martian Wallet) each launching on testnet beta. Aptos Naming Service was also built by the Aptos Foundation. Rounding out the early defi ecosystem, a number of additional dApps have captured attention, including Aptin Finance (a lending platform), Mover (EVM bridge to Aptos) Vial (algorithmic liquidity protocol), Mobius (non-custodial liquidity protocol) and AptosLaunch, the first decentralized launchpad on Aptos. LayerZero Labs also recently announced integration with Aptos on their Mainnet which went live on October 17. Here is a full list of projects building on Aptos put together by the community.

> Sui Ecosystem

The Sui ecosystem has seen rise to a number of independent browser extension wallets following the core team launching the Sui Wallet in July. These include Wave, Suiet, Hydro and a few others. MoveEx has emerged as the first DEX built on Sui along with BlueMove, the NFT marketplace for Aptos, also integrating the Sui network. A recent partnership with Axelar Network aims to help Sui developers bridge to the EVM-compatible applications. A nascent GameFi ecosystem has also begun to emerge, but far too early to even get a whiff of the potential winners.

> On Tokenomics [Updated on October 19]

Mysten Labs has announced the SUI tokenomics which you can explore in much more detail here. While they have provided general information around the token mechanics, they have yet to release a formal allocation breakdown by cohort.

Aptos on the other hand has taken a more opaque approach which has led to a tidal wave of early skepticism and disappointment among the community. Ahead of mainnet launching on 10/17 there was no public release of tokenomics or allocation breakdown to the public. It wasn’t until the day after mainnet launch and just a short time before APT was to be listed on top exchanges when tokenomics were formally announced, leading to some prominent players even calling this week’s launch of APT token launch as predatory to retail investors, and “among the worst designs I’ve ever seen, and I’ve seen a lot” according to Olaf from Polychain. This frustration is attributed to the fact that the 51% of token supply claimed by Aptos Foundation to be allocated to the ‘community’ in reality must be arbitrarily distributed by the Foundation via yet another opaque process– leading many to believe the Aptos Foundation and Aptos Labs actually hold >67% of the supply today– with no strong model to soundly distribute and decentralize the network. Also of concern is that according to the block explorer, apscan.io, 821 million of the 1 billion total supply is staked – implying that undistributed tokens could be potentially earning staking rewards. Aptos Labs co-founder and CEO, Mo Shaikh acknowledged the rocky launch on Twitter.

VI. Final Thoughts

The L1 game was saturated after last cycle. However, many other top L1s emerged deep from previous bear markets only to slingshot into the following expansion. Don’t discount the ability for this to happen again as new conversations around continued innovation and web-sized scalability take root.
We will see entire ecosystems and sectors emerge that are built on the Move programming language– perhaps even some alternative L1s to Aptos and Sui that could iterate on the innovation.
More projects may emerge and get funded that are built specifically on Narwhal & Tusk consensus mechanisms as more builders push the boundaries on innovating on scalability.
Another Rust-based programming language will drive more developer adoption to other rust-based blockchains such as Solana, further eating at Solidity market share in the industry.