Key takeaways
- Pontem operates as a product studio with live products including Pontem Wallet and Liquid Swap AMM on Aptos, now developing Lumio Layer 2 for Ethereum.
- Move VM is a Rust-based programming language created by Meta in 2019 that offers better security features than Solidity, including native formal verification and parallel transaction execution.
- Lumio aims to provide a vendor-neutral way for traditional finance companies to deploy applications on Ethereum using Move or alternative VMs instead of being locked into the EVM.
- Solidity was designed as a proof-of-concept in 2014 and has structural vulnerabilities like reentrancy errors that caused the DAO hack, whereas Move was purpose-built for blockchain applications.
- Pontem plans to port its existing wallet, AMM, and lending protocols built on Move to Ethereum through Lumio while maintaining composability across different blockchain environments.
A product studio heading to Ethereum
Ashton Addison: What is Pontem, and what are you building?▶ 00:22
Alejo Pinto: The best way to think about Pontem is as a product studio, like a game studio with several individual products underneath. Two are live: Pontem Wallet and the Liquid Swap AMM, both category leaders on Aptos. Now that we have those two beachheads, we're filling out the vertical integration of our product stack, down to the execution layer. That's Lumio, which is on testnet. It will let us launch the Move VM, and in future other alternative VMs, in a credibly neutral way on Ethereum. As traditional finance companies like Fidelity and BlackRock come to Ethereum, they can choose how to deploy their applications, rather than being locked into the EVM. Solidity is a bit like JavaScript, and you wouldn't build mission-critical banking infrastructure on JavaScript.
Ashton: Is the virtual machine a protocol that connects different blockchains?▶ 03:07
Alejo: Think of it as a credibly neutral way to deploy applications: the AMM, the wallet, and also lending protocols built with Move. Think of public blockchains as public clouds. If you were vendor-locked into building your backend only with JavaScript, that would be a really bad experience. We're essentially upgrading Ethereum, the public cloud infrastructure, so that we can deploy applications built with a better virtual machine and a better language, and be competitive there. This can be applied to other blockchains too. Initially these environments will be isolated, but we're also researching ways to connect them.
We're essentially upgrading Ethereum, the public cloud infrastructure, so that we can deploy applications built with a better virtual machine and a better language.
The case for Move over Solidity
Ashton: Move is new to many people. Why haven't other blockchains adopted it?▶ 04:35
Alejo: Move has actually been around since 2019. It was conceived at Meta, then Facebook, and built specifically for the Libra project. You can think of it as the best of Solidity and the best of Rust. It's designed after Rust, with a resource model and static typing, and it supports concurrency natively, so you can do parallel transaction execution. It also has safety features built in. The Kyber swap hack was an overflow error. On Ethereum you'd rely on a third-party library for formal verification, whereas in Move it's native to the VM. Solidity is kind of a proof of concept. In 2014 even Vitalik didn't think his idea had legs, and they had to come up with an entire new language. It has worked pretty well, but re-entrancy errors led to the DAO hack, and Ethereum literally forked the chain to give people their money back. If they had used Move from the beginning, re-entrancy errors wouldn't have been an issue. Aptos and Sui chose Move because that's what they incubated at Meta, and there's no reason you couldn't port it back to Ethereum. The fact that no one has done it is a big market opportunity.
Ashton: What would Uniswap gain by moving a version 4 to Move?▶ 13:51
Alejo: For the end user the improvements are marginal. Move natively supports parallel transaction execution, so two pools, say WETH/USDC and WBTC/USDC, can run at the same time. The EVM is sequential. The caveat is that parallelized EVMs are coming. Arguably Move also performs better, and each transaction might cost less gas. You also gain safety: no re-entrancy errors, which is a major hack vector, and a native prover for formal verification. For the user it should feel relatively the same, just a little cheaper and faster. For developers it's a more solid app that's less likely to get hacked. Even marginal improvements in performance and safety can create a network effect. If two products are otherwise a commodity, the one with the edge is more likely to get ahead. The drawback is the Lindy effect of the EVM, but as more applications are built with Move, that should start to catch up.
Upgrades, security and Lumio's next steps
Ashton: How would existing DeFi apps on Ethereum move to Move, and what does that mean for security?▶ 09:22
Alejo: Users should be aware of upgradeability. When a protocol is upgradeable, people holding multisigs can upgrade it. The developers could turn malicious or the multisig could be hacked, and the protocol could be upgraded to add a withdraw function and drain the funds. Look at Serum: when FTX collapsed, no one knew who had the upgrade keys. Check the docs of your favorite dapps to see whether they're actually decentralized. If there are upgrade paths, make sure the multisigs are secured by trusted people, and remember those people could still get hacked or be malicious. Uniswap is immutable, so even if you held a gun to the developers' heads they couldn't take the money out. But that means launching V1, V2, V3 as entirely new contracts and migrating liquidity each time. There's no reason the next versions of these protocols couldn't launch in Move and Rust. Incumbents from the EVM like Sushi and PancakeSwap have already started to launch versions on Aptos. On Lumio you also get composability between the EVM and Move, and you could have it with EVM L2s like Base and OP Mainnet. So legacy contracts keep working while you build new versions in Move. We're trying to give people choice, and with choice comes wider adoption.
Ashton: Pontem has raised $6 million in financing for Lumio. What are the next steps?▶ 16:06
Alejo: Lumio is a proof of concept on testnet right now. The funding lets us scale the team and hire more people to take it to mainnet. We need a lot of resources to do that.
If it's going to be free, it probably has to live on the blockchain. That's the only machine humans can't shut down.
AI, DAOs and on-chain trading
Ashton: Does AI play a role in what Pontem is building?▶ 16:55
Alejo: We have an AI product called Pontem AI. It's an LLM we've embedded with crypto information, so you can ask it about Aptos and Ethereum, and we have on-chain data in there. That's a fairly superficial use of AI. Fundamentally AI is about data and compute aggregation, and right now that sits on servers like Microsoft's, which control the data and what gets published. I asked ChatGPT to show me an image of poop and it said that goes against its content policy. We're being censored from looking at poop, which is kind of scary. The fix is to decentralize AI: put data on something like IPFS and build networks of GPUs that provide compute to train and run models. If you extrapolate AI 100 years, you'll have a sentient being. Is that being going to be Microsoft's, or is it going to be free? If it's going to be free, it probably has to live on the blockchain. That's the only machine humans can't shut down. You can also imagine DAOs where the parameter inputs, like a lending protocol's LTV, are set by algorithms instead of humans. You should probably still have humans check that it doesn't go wrong, but theoretically you could do this today.
Ashton: What are you focused on in 2024, and what should the public expect?▶ 26:03
Alejo: We're focused on growing our current products, the AMM and the wallet, and by extension Lumio adoption as we launch on Ethereum. I don't want to reveal too much about Liquid Swap, but expect a really low-latency, high-performance DEX, which I don't think has been done concertedly yet. I think we'll relaunch within the next six months. On Aptos and other next-generation blockchains you can start to match the performance of centralized exchanges. People use FTX or Binance because they don't want to wait an hour for a transaction or pay $4 for a basic trade. On Aptos, Solana and Sui it starts to feel a lot like a traditional web2 app. So why are people still using Binance and OKX after FTX imploded, when there are on-chain alternatives where you custody your own keys? I think it's because the infrastructure isn't yet there for it to feel like a web2 app, but that's coming.
This interview has been edited for length and clarity. Recorded February 2024.

