Been seeing a lot of discourse on X about agents triggering a bank run/lowering bank NIM by sweeping non-interest bearing deposits into yield. I think this is possible but going to take a lot longer than people think:
1. Retail banking is one of the highest switching cost products and deeply concentrated. The top 5 banks hold roughly half of US consumer deposits and banks with over $10B in consumer deposits hold close to 85%. These banks are incentivized to not be accessible to agents and are already not developer friendly. We have seen this with the big banks pushing back on open banking by charging for data access and reducing what is available. Muse can read your balances via Plaid but cannot move the money.
2. Well over half of consumer deposits are held by households 55+. This cohort is less likely to be an early adopter of an AI finance tool not being endorsed or directly offered by their bank.
3. The top 10% of households by wealth hold approximately 60% of deposits (top 1% alone is nearly 25%). Most of this money already has an advisor or brokerage attached reducing the incremental value proposition of a standalone consumer agent. The bottom 50% hold 5.5% of deposits, so the mass market agent story is less existential to bank funding.
Net net, I think in the fullness of time agents do become the primary interface for personal finance and cheap deposits shrink over time. But most of that comes from AI democratizing access to services the wealthiest people already get, and it will be relatively slow because these tools are new and people will need to establish trust before having them manage their money.
1. Retail banking is one of the highest switching cost products and deeply concentrated. The top 5 banks hold roughly half of US consumer deposits and banks with over $10B in consumer deposits hold close to 85%. These banks are incentivized to not be accessible to agents and are already not developer friendly. We have seen this with the big banks pushing back on open banking by charging for data access and reducing what is available. Muse can read your balances via Plaid but cannot move the money.
2. Well over half of consumer deposits are held by households 55+. This cohort is less likely to be an early adopter of an AI finance tool not being endorsed or directly offered by their bank.
3. The top 10% of households by wealth hold approximately 60% of deposits (top 1% alone is nearly 25%). Most of this money already has an advisor or brokerage attached reducing the incremental value proposition of a standalone consumer agent. The bottom 50% hold 5.5% of deposits, so the mass market agent story is less existential to bank funding.
Net net, I think in the fullness of time agents do become the primary interface for personal finance and cheap deposits shrink over time. But most of that comes from AI democratizing access to services the wealthiest people already get, and it will be relatively slow because these tools are new and people will need to establish trust before having them manage their money.
13
10
84
7,035
Noah Levine retweeted
Introducing @OpenReserveBank , a blockchain native de-novo digital bank, finally out of stealth. This week the Office of the Comptroller of the Currency (OCC) granted conditional approval of our full-service U.S. national bank charter application.
Preliminary conditional e novo national bank is a privilege our team does not take lightly. It is a significant milestone and the beginning of the most demanding phase of the work.
We're building toward real-time settlement across fiat and blockchain rails, direct Federal Reserve connectivity, tokenized deposits, and GENIUS Act-designed payment stablecoin infrastructure.
Thank you to the entire OpenReserve team, our board, investors, and to the OCC who ran a rigorous review. More to come!
See what we are building: openreserve.ai
Preliminary conditional e novo national bank is a privilege our team does not take lightly. It is a significant milestone and the beginning of the most demanding phase of the work.
We're building toward real-time settlement across fiat and blockchain rails, direct Federal Reserve connectivity, tokenized deposits, and GENIUS Act-designed payment stablecoin infrastructure.
Thank you to the entire OpenReserve team, our board, investors, and to the OCC who ran a rigorous review. More to come!
See what we are building: openreserve.ai
35
24
230
45,552
Noah Levine retweeted
Money-transfer startup Felix Pago raised $200 million to expand beyond remittances and offer services such as lending and savings to Latin American immigrants in the US bloomberg.com/news/articles/β¦
9
3
29
41,568
Noah Levine retweeted
Thank you @alive_ @nlevine19 and the whole @a16zcrypto team for partnering with us. Looking forward to building the future of financial services together
1
1
20
637
Awesome tool to see which teams are building stablecoins projects globally.
One of the hardest questions in stablecoins shouldn't be:
"Who else is building and where?"
Today we launched a completely rebuilt Stablescape to help answer that.
- 6,000+ stablecoin companies
- Interactive market maps
- Rich company profiles
- And more
You can now submit existing one, and play around with interactive charts.
See it for yourself here: stablescape.xyz
1
1
22
2,033
Noah Levine retweeted
Here's a hypothetical scenario of where stablecoin netting makes sense:
The volume of agentic micropayments the Cloudflare CEO is talking about below cannot be handled by any known blockchain architecture. Even a corpo DLT with 5 validators can't scale that high (and even if it did, the centralization would eliminate the point of using stables in the first place).
So instead what Cloudlfare could do is net the micropayments. It's highly likely an AI agent paying a micropayment to access the content of one Cloudflare website also needs to access many others in a short window. So Cloudflare can play the role of trusted central counterparty, performing multilateral netting on all the micropayments that go to a particular website and settling on-chain periodically.
The best chain to do this on would be a highly performant ETH L2, because rollups do their own type of netting, posting the blobs and proofs of a bunch of transactions happening on fast L2 blocks to the slower L1. Tiered netting is standard practice in TradFi.
If this model sounds familiar to you, it's the same trusted service card schemes perform between banks and merchants. So in theory, Visa and Mastercard could offer multilateral netting for agentic micropayments as well.
The problem they will have is that they are offering one set of users a similar service to what they offer another but at a fraction of the cost. That's a tough juggling act and incumbents seldom pull it off. Cloudflare doesn't have this problem.
The volume of agentic micropayments the Cloudflare CEO is talking about below cannot be handled by any known blockchain architecture. Even a corpo DLT with 5 validators can't scale that high (and even if it did, the centralization would eliminate the point of using stables in the first place).
So instead what Cloudlfare could do is net the micropayments. It's highly likely an AI agent paying a micropayment to access the content of one Cloudflare website also needs to access many others in a short window. So Cloudflare can play the role of trusted central counterparty, performing multilateral netting on all the micropayments that go to a particular website and settling on-chain periodically.
The best chain to do this on would be a highly performant ETH L2, because rollups do their own type of netting, posting the blobs and proofs of a bunch of transactions happening on fast L2 blocks to the slower L1. Tiered netting is standard practice in TradFi.
If this model sounds familiar to you, it's the same trusted service card schemes perform between banks and merchants. So in theory, Visa and Mastercard could offer multilateral netting for agentic micropayments as well.
The problem they will have is that they are offering one set of users a similar service to what they offer another but at a fraction of the cost. That's a tough juggling act and incumbents seldom pull it off. Cloudflare doesn't have this problem.
CT missed the most important earnings call of the week. It wasn't a crypto company. It was @Cloudflare
Everyone caught the wallet announcement. The real alpha was in the call itself.
Today Cloudflare monetizes the internet through subscriptions: security services, AI platform spend, pools of funds. A mix of SaaS and IaaS economics. The CEO was explicit that AI agents are about to break that model.
Cloudflare sits in front of roughly 20% of internet traffic. Here's what that traffic looks like from the call:
- AI agent requests up 1,700% YoY
- Agents crossed 50% of total network traffic this quarter. First time in history non-human traffic is the majority. Management admitted it happened faster than their own models
- Their projection: if trends hold, non-human traffic outnumbers human traffic 1000x within 5 years
The monetization shift is the key part. The ad-supported internet doesn't work when the visitor is an agent. Cloudflare's CEO answer: block malicious bots for free, charge good agents a tiny fee per request. Fractions of a penny. They want to be the ones defining that layer.
Now the throughput math here:
- Cloudflare handles ~500M requests per second
- They estimate 1 to 10% is monetizable via micro/nanotransactions
- That means 10M TPS on day one, scaling to 100M TPS
Visa peaks at ~20k TPS
The CEO's framing: "we're building this while others compete with Visa." Three to four orders of magnitude beyond card rails. No existing payment network can settle this. It has to be something new.
Two conclusions I keep coming back to:
- Being short L1 throughput is being short agentic workflows. If agent traffic gets monetized per request, the settlement layer needs to scale orders of magnitude beyond anything live today.
- The fee math for L1s flips. Base fees have collapsed across ETH, SOL, everywhere. MEV is getting internalized by apps. Hard to build a base fee revenue case at human scale. But at 10M TPS and $0.001 per transaction, you're looking at ~$315B a year in base fees alone. At 100M TPS the number gets silly.
Stablecoins and crypto are the end-game here for Agentic finance @jerallaire @circle
5
11
26
4,923
Great time chatting with @sytaylor and @bamazizimesh about the future of agentic commerce!
π¨ Ep. 10 of Agentic Commerce:Β Why Stablecoins Wonβt Kill Cards
@sytaylor & @bamazizimesh are joined by:
πΒ @nlevine19, Partner, @a16zcrypto
To discuss:
π Agentic commerce is still very early in development
π New class of merchants and buyers emerging
βοΈ Stablecoin and bloceeds
π³ Stablecoins won't kill cards but fill gaps
β Card networks offer trust and brand value
π° Value accrues above payment protocols not in them
ποΈ Wallet aggregation and headless merchants are key
πΈ Compute orchestration and real time payment opportunities
π Jevons paradox and streaming payments in finance
***
Timestamps:
00:00 Introduction
2:20 Agentic commerce is still very early in development
4:14 New class of merchants and buyers emerging
6:17 Stablecoin and blockchain for AI payment needs
7:37 Stablecoins won't kill cards but fill gaps
11:50 Card networks offer trust and brand value
13:23 Value accrues above payment protocols not in them
16:45 Wallet aggregation and headless merchants are key
21:51 Compute orchestration and real time payment opportunities
29:20 Jevons paradox and streaming payments in finance
***
π ππ¦π’π³π€π© 'ππ°π¬π¦π―πͺπ»π¦π₯ ππ°π₯π€π’π΄π΅' ππ― π π°πΆππΆπ£π¦. ππ±π±ππ¦, ππ±π°π΅πͺπ§πΊ π°π³ π’π―πΊ ππ°π₯π€π’π΄π΅ πππ’πΊπ¦π³! π
2
3
14
1,443
Noah Levine retweeted
DAOs were a failure in the last era of crypto.
But it may have just been the wrong time for them. Two big things are different now: (1) regulatory clarity and (2) AI agents.
We spent the last 10 years rediscovering the hard way that direct democracy is a bad idea. Turns out, nobody wakes up in the morning eager to adjudicate risk parameters or protocol updates.
It didn't help that it was illegal to experiment with different DAO structures. The vicious hostility of Biden and Gensler toward the space killed all innovation.
But it's a new day.
It is once again legal to experiment. You could choose to play with forms of representative democracy, or multi-cameral systems, or hybrids of permissioned and permissionless participation.
Not to mention, AI agents can now step into many kinds of roles that humans will never want to perform. They may help DAOs live up to their name by making them truly Autonomous.
The design space for new DAOs is infinite, for they are just software. What new experiments are people running?
But it may have just been the wrong time for them. Two big things are different now: (1) regulatory clarity and (2) AI agents.
We spent the last 10 years rediscovering the hard way that direct democracy is a bad idea. Turns out, nobody wakes up in the morning eager to adjudicate risk parameters or protocol updates.
It didn't help that it was illegal to experiment with different DAO structures. The vicious hostility of Biden and Gensler toward the space killed all innovation.
But it's a new day.
It is once again legal to experiment. You could choose to play with forms of representative democracy, or multi-cameral systems, or hybrids of permissioned and permissionless participation.
Not to mention, AI agents can now step into many kinds of roles that humans will never want to perform. They may help DAOs live up to their name by making them truly Autonomous.
The design space for new DAOs is infinite, for they are just software. What new experiments are people running?
100
34
287
73,649
Noah Levine retweeted
Three barriers have kept institutions from adopting blockchain infrastructure: performance, regulatory clarity, and privacy.
Performance and regulatory frameworks have advanced, and Canton was built to address privacy from the start.
@alive_ and @nlevine19 on why that matters.
Performance and regulatory frameworks have advanced, and Canton was built to address privacy from the start.
@alive_ and @nlevine19 on why that matters.
14
16
97
9,073