Skip to content

The CCS Interview

Cartesi's Bruno Maia: a Linux VM that ran Doom on Ethereum infrastructure

Bruno Maia, ecosystem growth lead at Cartesi, explains how the project's application-specific rollups run a full Linux OS alongside Ethereum, and why he thinks that opens the door to richer on-chain games and AI. The conversation, recorded in August 2023, also covers modular blockchains, the Doom demo and Cartesi's plan to launch on Ethereum mainnet.

Interview by Ashton Addison Recorded August 2023 5 min read

Key takeaways

  • Cartesi is a modular execution layer that runs a Linux-based RISC-V virtual machine on top of Ethereum, enabling significantly more computational capacity than the EVM alone.
  • Linux powers over 90% of devices globally and provides developers with familiar programming tools, libraries, and languages compared to the limited EVM environment.
  • Cartesi uses optimistic rollups for dispute resolution, maintaining Ethereum's security guarantees while enabling deployment across multiple layer-one and layer-two blockchains.
  • The modular blockchain approach allows Ethereum to focus on decentralized consensus and data availability while delegating computation to specialized execution layers like Cartesi.
  • Application-specific rollups built on Cartesi can execute thousands more computation cycles than a single Ethereum block, enabling complex real-world applications on blockchain.

A full Linux machine beside Ethereum

Ashton Addison: For anyone who hasn't seen our first interview, what does Cartesi do?▶ 03:15

Bruno Maia: Cartesi is a modular execution layer, used in a specific rollup design called application-specific rollups. What gives it value on top of that design is a RISC-V machine that runs an entire Linux OS. It's the first time the blockchain ecosystem has had such a refined and sophisticated VM running a full Linux. For developers that is a game changer, both for the kinds of programs you can build and for how you approach programmability in blockchain.

Ashton: Linux has been central to computing since the early internet, yet we rarely hear about it in blockchain. Why is that, and why is Cartesi using it now?▶ 04:56

Bruno: Linux is a common layer for computing. Ethereum, by comparison, is a very primitive world computer with very limited computational capacity and data storage. That is a bottleneck to reaching the mainstream, because you can't solve real-life problems with so little capacity. But its greatest achievement is the decentralization of the truth. Whatever runs on Ethereum is validated by thousands of nodes, which gives you verifiability and tamper-proofing. We are not trying to replace Ethereum; on the contrary, we are augmenting what Ethereum achieved. With Linux you can choose your programming language and use the libraries and tooling that developers already know. Everything keeps the same security guarantees as Ethereum, through the dispute resolution of optimistic rollups.

We are not trying to replace Ethereum; on the contrary, we are augmenting what Ethereum achieved.

Why the future looks modular

Ashton: Modular blockchains are getting a lot of attention. Do you think they will become the status quo?▶ 08:22

Bruno: We are still in the early days of building tomorrow's blockchain infrastructure, so it is very much an experimental field. Ethereum, as the first, hit problems earlier than other chains. It realized that a world computer validating every single thing could not run enough computation, or cut transaction costs enough, for mass adoption. So the base layer concentrates on the decentralization of the truth through consensus, and on increasing data storage. The rest of the roadmap relies on upper layers. But many layers built on top so far made architecture choices that face similar limitations. They look unloaded now because it's early, but with time they will be busy too. If you want to scale computation enough for real-world applications beyond typical financial solutions, you need application-specific rollups on top of those layers. Each use case demands different characteristics, and modularity gives you that plasticity.

Ashton: Everything seems to tie back to the Ethereum Virtual Machine. Will everyone keep building on it?▶ 16:57

Bruno: It's a choice that depends on the requirements and the economic value of the application, and Cartesi is not the only way to implement smart contracts. I expect Ethereum's base layer to be used more for settlement and storage than for running your code directly, and the EVM will be used extensively for years. It is simple by design. When thousands of nodes validate the same computation, you can't have huge computation capacity, so you need the simplest virtual machine that still allows any kind of programming. Cartesi lets people do more sophisticated things, and brings in millions of developers who work in languages like Python and Java, with their libraries and tooling. It's already happening in our hackathons. Someone from Australia built ChatGPT inside Cartesi, implementing a large language model, which is pretty amazing.

Games, Doom and on-chain logic

Ashton: Could gaming be one of the near-term use cases for this kind of extra computation and storage?▶ 11:44

Bruno: Absolutely. Games show how limited today's infrastructure is. A blockchain game still has to be fun to play, with real gameplay dynamics, otherwise it makes no sense. So far we have seen centralized games add a small layer of blockchain, mostly for compensation or wearables. Even then, the wearables are minted and used in gameplay under centralized logic, which is awkward. With far more computation you can build much more logic on-chain, on layers that are fully verifiable, immutable and protected by the blockchain's consensus. One community member built a chess game where AI robots, written in Python, battle each other, and the player is actually programming his robot for the fight. You can also layer in different incentives, compose with other games and even build machine-to-machine economies.

Ashton: I saw a Cartesi and Doom partnership. What's going on there?▶ 14:48

Bruno: It's not a formal partnership with the software house that created Doom. Doom is an iconic game, the cradle of first-person shooters together with Wolfenstein. Someone in the blockchain community tweeted a challenge to Ethereum: if it's the world computer, why can't it run Doom? Which is a valid question. A developer from our community replied, challenge accepted, and showed Doom running on a Cartesi machine on top of Ethereum infrastructure, with an application-specific rollup. It's the first time the entire logic of Doom has been composed into this kind of virtual machine and executed. You take code developed over more than 30 years, compile it for our version of Linux, and all the libraries and languages are ready. It shows the complexity our virtual machine can handle, and it's an open invitation to experimental developers.

Not every application you build in the world needs a blockchain.

Barriers to adoption and the road to mainnet

Ashton: Why aren't more Web2 developers building decentralized applications beyond finance?▶ 20:33

Bruno: There isn't a single reason. The a16z state of crypto report for 2023 was very clear that a key roadblock to use cases beyond DeFi is the lack of computation capacity, what the industry calls block space. Massively increasing that capacity is a core part of our design, and we believe it is a good step toward new, real-world applications. There is also an adoption cycle for any new technology. And not every application you build in the world needs a blockchain. Some things are blockchainizable, and for others it's fine to run on a centralized service like AWS, with super high performance, global latency and real-time response. Blockchain will not completely replace the world of computing.

Ashton: What's on the roadmap toward the end of this year and into 2024?▶ 22:41

Bruno: We are in full adoption mode and launching on mainnet. For an application-specific rollup that is a novel concept. Cartesi will launch an application on Ethereum mainnet, to show that our technology is ready for other developers who want to deploy on mainnet. We will keep improving usability and form partnerships to extend how Cartesi's technology can be composed, to give more publicity to different use cases. We are in touch with several important players in the ecosystem. I believe the blockchain industry is much more collaborative than competitive, because different projects are good at building specific pieces that will be useful in the future. We welcome all developers who want to do more than Web3 with Cartesi, in any programming language they want.

This interview has been edited for length and clarity. Recorded August 2023.

Bruno Maia
Bruno MaiaBruno Maia is a co-founder and COO of Zeom, a London-headquartered fintech serving the Brazilian market. He is a Web3 business development and growth executive best known for his work at Cartesi. He is a Brazilian telecommunications engineer with a master's degree from the Pontifical… Full profile →
Share this interviewXLinkedIn
▶ Watching the interview