Week three of the revival, and the industry handed us a theme. This was the week subsidies, charters and whole chains got stress-tested. The Robinhood Chain retention data we’ve been waiting two weeks for is in, an L2 that once held $2 billion announced its own funeral, and the bank charters I covered two weeks ago are already in court. Signal in five minutes, sources linked, rabbit holes yours.
1. Robinhood Chain survived its first week without free gas (mostly)
The 90-day gas subsidy for Robinhood Wallet users expired on September 29, and every transaction on the chain now pays its own way in ETH. The early verdict is split. Total value locked held above $1.03 billion as of October 1, still over the billion-dollar mark first crossed on September 22, while daily DEX volume roughly halved from $1.88 billion to $947 million as memecoin trading cooled and chain fees fell about 31% week over week. Deposited capital, in other words, held up far better than trading activity.
My take. I flagged this as the number to watch two weeks running, so here is the honest read. The halving in volume looks worse than it is, since much of the memecoin churn ran through third-party bots and launchpads that never used the subsidy anyway, and some of what left was the rug-pull assembly line I covered last week finally packing up. The number that actually matters is the one that held. A billion dollars of capital stayed deposited when the free ride ended, which suggests the chain is quietly rotating from a meme casino toward the lending and tokenized-equity venue Robinhood designed it to be. October’s full-month data and Robinhood’s Q3 earnings in late October are the confirmation points. For those of us building here, a smaller, less mercenary chain with sticky deposits is a better home than a subsidized volume record.
2. Aave turned tokenized stocks into working collateral
Aave V4 on Base launched an Equities Hub accepting seven Coinbase tokenized stocks (Apple, Amazon, Alphabet, Meta, Microsoft, Nvidia and Tesla) as collateral for USDC loans, with Chainlink supplying the price feeds. The tokens are Coinbase-issued certificates backed 1 to 1 by shares in segregated custody, collateral factors run from 65% to 79%, initial caps sit at $32 million of USDC supply and $21 million of borrows, and access is limited to eligible non-US users. The stocks are collateral only at launch, no borrowing the equities themselves and no equity-against-equity positions.
My take. This actually went live on September 25 and slipped past me last week, which I regret, because it is the clearest answer yet to the activation problem I keep writing about. Two weeks ago Binance Research showed 88% of tokenized value sitting inert. This is what the other 12% looks like, a tokenized Apple share earning its keep as loan collateral at two in the morning on a Sunday. Note the design conservatism, though. Separate hub so USDC lenders opt in explicitly, modest caps, no recursive equity leverage, and reserves pause during corporate actions like splits. That restraint is the tell that everyone involved remembers what happens when novel collateral meets aggressive parameters. The question I posed in May about designing liquidation logic for collateral with off-chain pricing is now a production system, and its risk choices are worth studying before you ship your own.
3. Blast is shutting down
On October 2, Blast announced it will wind down its Ethereum layer 2, saying the costs of maintaining the chain exceed the revenue it generates and that there is no credible path to economic sustainability. Users have until October 26 to withdraw through the normal interface, after which assets remain recoverable only by interacting directly with bridge contracts on Ethereum mainnet. At its 2024 peak Blast held over $2 billion in deposits. By the end, TVL had collapsed to roughly $32 million.
My take. Hold this story next to item one, because they are the same story with opposite endings. Both chains bought activity with incentives, Blast with points and native yield, Robinhood with free gas. When Blast’s incentives faded, the capital left, and with roughly $30 million in TVL a chain cannot cover its own data and infrastructure costs. ZetaChain folded its L1 last month, and Blast makes two. The L2 shakeout is no longer hypothetical, and the dividing line is legible. Chains attached to a real distribution engine and a real asset class are consolidating the activity, and chains whose only product was yield-on-deposits are discovering that mercenary capital has no loyalty program. If you are deploying contracts or holding assets on a long-tail L2, the Blast playbook (temporary withdrawal pause, deadline, then manual bridge interaction) is your preview of the exit experience. Choose your chain like it might someday email you a withdrawal deadline.
4. Community banks are suing to kill the agent-bank charters
The Independent Community Bankers of America filed suit against the OCC, challenging the framework behind its recent wave of crypto trust bank charters. ICBA argues the agency exceeded its statutory authority by permitting national trust banks that neither accept deposits nor operate within the fiduciary limits it contends federal law requires, calling the charters a “side door” into the banking system.
My take. Two weeks ago I covered Catena and Agora getting preliminary OCC approval and called it the institutional answer to agent attribution. The incumbents noticed too, and their response was litigation. Read the complaint for what it is, an incumbency defense dressed in statutory interpretation, because community banks understand exactly what it means if businesses whose AI agents move money can get custody and fiduciary services from a purpose-built charter instead of from them. But do not dismiss the legal risk. The charters are preliminary and conditional, and a court loss for the OCC would freeze the cleanest regulatory path agentic finance currently has. Builders planning around these trust banks should treat the charters as probable rather than settled, and keep the x402-style payment-layer alternative in the design space. The race I described two weeks ago now has a third participant, the courts.
5. The Ethereum Foundation shipped private AI inference
The Ethereum Foundation and the Open Anonymity Project launched zkAPI, a zero-knowledge protocol for private AI. The flow is simple to describe. Deposit ETH into a vault, sign a zk proof, and receive a fixed amount of AI inference without linking your identity or data to the request. The Foundation framed it as moving past intermediaries that require users to tie their data and identities together, and an initial chat product built on the protocol is already live.
My take. Most AI-meets-crypto announcements are a token looking for a reason. This one is infrastructure with an obvious customer, which is any agent or application that needs to buy intelligence without building a surveillance trail. Today every agent’s inference calls route through an API key that identifies its operator, links its queries together and creates a log someone else controls. zkAPI proposes payment and access without identity, which is philosophically the exact opposite of the Catena model in item four, where the answer to agent trust is a chartered fiduciary that knows everything. I do not think these are competitors so much as the two poles the agent economy will oscillate between, accountable agents for finance and anonymous agents for inference. Vitalik’s roadmap post last week said Ethereum is becoming a cryptographic world computer rather than just a blockchain. This is what the first applications of that look like.
Bonus tidbits
The SEC is not slowing down. A proposed crypto custody framework landed October 1, with comments due October 20, giving advisers and regulated funds a path to hold crypto through qualifying custodians. Chair Atkins also vowed further rule changes to keep crypto markets onshore. The post-CLARITY agency sprint continues.
The UK opened its doors. The FCA began accepting applications under the UK’s new crypto regime on September 30, with the window running through February 2027 ahead of an October 2027 start. If US litigation risk worries you, the UK just published its onramp.
Bitget recovery check. Of the $387.5 million stolen two weeks ago, roughly $1.1 million had been frozen as of October 2. Withdrawals resumed and the protection fund held, but as a recovery rate, 0.3% is the honest number to internalize about exchange hacks.
The macro whipsaw. A soft September jobs report briefly pushed Bitcoin to $87,219 before a derivatives liquidation flush dragged it back. The week closed near flat with total market cap around $2.99 trillion. Rate-cut hopes now do the work rate hikes used to.
Russia is paying state employees in CBDC. The finance ministry confirmed some salaries now flow through the digital ruble, where every transaction sits on the government’s ledger and payments can be programmed or frozen. The dystopian reference implementation, shipping while the West debates.
That’s the week. Five things to think about, five more for your peripheral vision. If this earned a spot in your Monday, consider subscribing. Next week’s edition lands (around) the same time, same place, with the same promise.