Written in September 2024, while Pete was running the Techstars Web3 Accelerator.
Co-written with Hugh McGirr and first published by Techstars when applications opened for the Web3 2025 program. This is a condensed version. The full piece is on the Techstars site.
Networks and ownership
Hugh and I started from networks: groups of people, devices, data and assets that become more valuable as they grow, and that mean more to the people in them when those people own a piece. Our earlier theses argued that giving contributors and users real ownership is one way to grow a market. Blockchains make that workable at scale, because code and incentives can do the coordinating that intermediaries used to charge for.
We were clear that this is early. Most people still spend their time on centralized platforms whose algorithms are tuned for retention. We wanted founders building something that grows like a city, open to anyone who turns up, rather than another walled garden.
What we look for
New technology changes behavior, and changed behavior opens room for business models that incumbents either miss or can't pursue. Those openings tend to show up at the edges of big markets, where they are easy to wave away.
Copying a web2 service onto a blockchain rarely works. A web3 product usually has to be something like ten times better on whatever matters to its users, whether that is the experience, the cost of capital, access to knowledge or how resources get used. Composability is what makes that reachable for a small team. Tokens, wallets and open protocols can be combined across applications, so founders build on work that already exists instead of starting from zero.
And we back founders who can describe a credible path to open, decentralized services and pull together the resources to build it.
Three layers of adoption
We grouped the opportunity into three layers.
At the developer layer, the long-running trade-off between security and usability was the problem to solve. We were looking for composable security primitives that ship alongside composable UX, plus tooling that helps developers scale: account abstraction, MPC, cross-chain interoperability, privacy and ZK, app-specific rollups, and better on- and off-ramps.
At the protocol layer, the question was which incentives can coordinate users at large scale and keep them coordinated. Tokens are programmable assets and can bootstrap a network with less risk. We also pointed to dollar stablecoins, already used outside the US for savings, remittances and trade because they settle quickly and hold their value. Our worry was that too many incentive designs still looked like a casino.
At the application layer, digital currencies were becoming part of global financial infrastructure, with real use in payments, custody, tokenized real-world assets, prediction markets, gaming and IP. We expected those uses to spread into bigger markets over time, social included.
AI as the accelerant
We argued that AI and blockchains need each other. AI needs blockchains for attestation, so anyone can check where data and content came from. Blockchains need AI to coordinate and make sense of data at global scale.
Reading it now
Several threads here carry straight into the Norio View, stablecoins and AI agents in particular. The difference is focus: Norio narrows in on finance onchain and the bridge that gets financial firms onto it.
