The theme of this past quarter is that crypto continues to march forward through the diffusion of onchain finance. Startups, fintechs, banks and asset managers are all meaningfully building on blockchains and finding success. We’re also starting to see onchain finance merge with AI to create financial products that understand us and can move money on our behalf. The combination of programmable assets and abundant intelligence promises to define the next era of financial services. 

At the same time that crypto is becoming an increasingly important piece of infrastructure for the financial system, future attack vectors are growing - most notably the threat of quantum computing and advances in AI’s ability to solve math. This will require adequate attention and investment from the industry over the coming years to build robust defenses.

Onchain finance continues to march forward

The convergence of onchain finance with traditional finance is a continual point of focus for us and we’d be remiss if we didn’t provide an update after what was demonstrated this past quarter. Crypto has never had a bigger impact than it does today and it’s doing so by bringing financial services onchain. There are many examples of how this is playing out but we’ll just touch on a couple.

Robinhood Chain launched at the start of the quarter and finished September with just over $1bn in deposits and $39m in revenue for the month. A lot of September’s traction came from memecoin trading, which some are quick to dismiss. However, as we’ve seen with many successful products in the past, speculation is often the Trojan horse for a wider range of financial products and consumer onboarding to follow.

Tokenized stocks, and the ability to borrow against them, are one of those. Both Robinhood and Coinbase launched tokenized stocks this quarter, along with the SEC issuing a five-year “Innovation Exemption” to set their guidelines around regulating this activity. By late September, tokenized stocks had reached around $146m in value on Robinhood Chain and around $22m on Base. Across all blockchains, tokenized stocks now count over $3bn in value, $10bn in monthly volume, and are held by more than 4m wallets.

Morpho launched lending markets on Base that let users borrow against these stock tokens. At launch in mid-September, there was around $55k in active borrows against $104k in collateral.

It’s still early, but this has always been our view: DeFi protocols explode in utility as more assets get tokenized and get put to work across lending, trading and everything else onchain.

Agentic finance: a tangible synthesis of crypto and AI

In our view, the future of finance is both onchain and intelligent. Bringing assets and markets onchain makes them programmable, while AI makes them intelligent. While the potential of combining these two technologies has largely remained theoretical, we're now starting to see them come together to create compelling products.

The clearest example so far is AI agents that can move money and interact with financial markets on behalf of users. Coinbase has been building the infrastructure for this with AgentKit, which gives AI agents their own wallets and the ability to transact onchain. Gina, one of our portfolio companies, allows users to program agents in natural language to trade assets across spot, perpetual and prediction markets.

We believe onchain is the natural economic home for agents. An agent can own its own wallet, move money at any time, and interact with open financial protocols without needing a traditional bank or brokerage account. This makes it possible for agents to do more than analyse markets and recommend trades. They can actually participate in them.

Quantum and AI risks becoming more acute

Security has been garnering more attention for two reasons: there is both more value onchain to secure, and advances in quantum and AI are increasing the risk of future attack vectors.

Of the two, the quantum risk is a little further out, but the industry is already taking it very seriously. Coinbase set up an independent advisory board on quantum in January and is already migrating its cryptographic systems to be post-quantum ready. The Ethereum Foundation formed a dedicated post-quantum team the same month and is targeting full quantum resistance by 2029.

The AI risk is more acute. In September, OpenAI announced that one of its models may have solved a mathematics problem that had stood for decades. The risk is that models get smart enough to crack elliptic curve cryptography, which underpins the security of almost every blockchain as well as TradFi systems.

Both quantum and AI will keep improving, and the industry needs to begin preparing now. Plenty of startups are working on these problems. We’ve already backed some and will look to back more.