Decentralize AI or only big tech wins; ETP vs. ETF; defend, play, trust
Editor’s note: This post originally appeared in our web3 newsletter — a guide to trending topics in crypto with insights and resources from engineers, researchers, and others on the a16z crypto team. Subscribe to see it in your inbox every other week.
✨Spotlight: AI x crypto
with Dan Boneh, Ali Yahya, Ben Fielding, Harry Grieve; Sonal Chokshi and Jason Rosenthal
This week we’re highlighting a popular — and more relevant than ever — theme: how crypto can benefit AI, and how AI can benefit crypto.
Building AI systems requires ever-larger deployments of computational power for training today’s largest and most powerful models. But privileged access to computing power and the economies of scale for large data centers gives big tech companies an advantage over startups in the race to capture value from AI. To level the playing field, startups and other companies need affordable access to large-scale computing power. So in this just-released (and final) talk in our Crypto Startup Accelerator CSX Spring 2024 playlist, the cofounders of Gensyn discuss the importance of building a permissionless, decentralized, machine learning protocol upon which AI models can train… and on virtually any compute device in the world. watch the talk here
This week, the web3 with a16z podcast also resurfaces the AI x crypto theme — threading the tension between centralization vs. decentralization. The discussion covers everything from deepfakes, bots, and the need for proof-of-humanity in a world of AI; to large language models like ChatGPT; user control, privacy and security; zero-knowledge machine learning (zkML); to media, art, and more. listen to the episode/ read the Q&A here
see also:
- Forget antitrust, regulate to let tech disrupt itself: on reining in intermediaries, and how interoperability is the best antidote to someone standing in the middle
- GPT@home: Why the future of training is decentralized: on the current state of frontier model training; history of distributed computing from SETI to Folding@Home; challenges, recent breakthroughs, and the road ahead
- On AI & personhood credentials — paper about the need for privacy-preserving ways to distinguish who is real online (with OpenAI, Microsoft, Harvard Society of Fellows, University of Oxford, a16z crypto, MIT, and several others)
❓Why Ethereum ETPs (vs. ETFs) are so significant, plus answers to other FAQs
A couple months ago, U.S. stock exchanges made available exchange-traded products (ETPs) based on Ether, the asset native to the Ethereum blockchain, for the first time… following the listing of Bitcoin ETPs earlier this year. While much has been said about these ETPs as investment products, what do they mean for crypto as an asset class? Why do they matter to the industry — for Ethereum specifically, and Solana and others potentially?
And while we’re at it, what even is an ETP? What’s the difference between that and the commonly (but mistakenly) referred to “ETF”? Plus, what are some advantages and disadvantages of BTC and ETH exchange-traded products? In this explainer, Scott Walker — chief compliance officer at a16z and former senior specialist, special counsel, and attorney-advisor for the SEC — provides answers to both basic and industry-specific frequently asked questions. read on to learn more
🛡️News you can use: A legal defense fund for creators using crypto
Artists and creators have long been early adopters of blockchain technology — including NFTs — to help them directly monetize and control their work, ensure authenticity, and earn royalties through smart contracts. And unlike most of the crypto industry, creators seem to have been spared from the ire of the Securities and Exchange Commission (SEC)… until now.
But well-meaning artists should not have to be intimidated by government agencies when experimenting with new technologies that may better serve their fans and livelihoods. So several organizations, including a16z crypto, partnered to help offer legal services and consultations to artists and creators facing potential government action — or who simply want to ensure they’re pursuing their NFT projects in a legal and compliant manner. learn more here
🌌Application to play (test): On games as reality
This recent episode of the web3 with a16z podcast digs deep into game design, philosophy, and gaming mechanics — with the maker of one of the world’s iconic massive multiplayer online role playing games: EVE Online. Hilmar Pétursson, CEO of CCP Games (makers of EVE Online), shares CCP’s unique worldview as well as the studio’s next act: a blockchain-based overhaul of the sci-fi game that’s currently inviting people to apply as playtesters before launching widely to the world.
In discussion with a16z crypto CTO Eddy Lazzarin, the two also go into the technology and economic design of the sci-fi simulation. Their conversation also touches on how niche cults can break into mainstream culture; how slow databases can make for fun gameplay; and more. listen (or watch) here
🔐Trend: Trusted execution environments (TEEs)
Simply put, TEEs provide an isolated environment where applications can be executed — allowing operators to run a job inside of that trusted execution environment regardless of the rest of the computing system. They are thus a powerful mechanism for designing secure systems, with many applications in cloud computing and beyond. And though they have been around for a while, TEEs have been advancing recently thanks to ongoing work by the community of people working on blockchains. Below are a couple resources on the topic from our site and around the web:
On TEEs for blockchain applications by Ari Juels (a16z crypto research seminar presentation, 2023)
Trusted Execution Environments (TEEs) are a powerful security tool, and their functionality in principle subsumes a gamut of cryptographic primitives. Cornell University’s Juels discusses the use of TEEs in blockchain infrastructure — focusing on block-building (one of the first specific applications where hardware-based TEEs took hold). He discusses leveraging TEEs to enforce “fair” ordering of user transactions. He also suggests an approach to addressing security concerns in TEEs, and privacy-preserving systems more generally. watch here
Confidential computing proofs: An alternative to cryptographic zero-knowledge by Mark Russinovich, Cédric Fournet, Greg Zaverucha, Josh Benaloh, Brandon Murdoch, Manuel Costa (ACM Queue, September 2024)
This article attempts to relate zero-knowledge and confidential-computing proof concepts, which were developed by different communities under different trust assumptions. A variety of TEEs are now available to help achieve soundness and zero knowledge — but with important differences. While TEEs are not suitable for all proof applications, the article explores these tradeoffs and characteristics for proof systems. access pdf here
— a16z crypto editorial team
You’re receiving this newsletter because you signed up for it on our websites, at an event, or elsewhere (you can opt out any time using the ‘unsubscribe’ link below). This newsletter is provided for informational purposes only, and should NOT be relied upon as legal, business, investment, or tax advice. Furthermore, the content is not directed at nor intended for use by any investors or prospective investors in any a16z funds. Please see a16z.com/disclosures for additional important details, including link to list of investments.