“Now my balance sheet as an originator drops from a week’s worth of production to one loan’s worth of production.”

Mike Manning, Head of Institutional Finance, Ava Labs

That is the most useful way I have heard someone explain the promise of onchain private credit.

Watch on Youtube

Private credit has grown to roughly $2 trillion. Yet many facilities still run on spreadsheets, monthly reports, emailed PDFs, and weekly funding cycles. An originator may make a loan today and wait days for the money that funds it.

I sat down with Juan Montero, co-founder and CEO of Fence, Anant Matai, Investments & Product at Grove, and Mike Manning, Head of Institutional Finance at Ava Labs. We talked about what changes when the loan and the money move on the same rails, why most tokenization still stops at the wrapper, and what code should never be trusted to decide.

This is a playbook for building working private-credit rails, not a podcast about putting old assets inside new tokens.

Download the Full Report

This ad could belong to you. 🚀 Build credibility. Drive pipeline and revenue. We position you among 100,000+ digital asset decision-makers.

Let's talk

Why this matters now

The first wave of tokenization changed the ownership wrapper. A fund share or bond became a token, but the underlying loans, servicing, covenants, and cash movements often stayed in the old system.

The second wave is trying to change the operating model. A loan is created digitally. Its legal documents point to an authoritative record. Eligibility and covenant rules can be checked continuously. Repayment can arrive in digital cash and map back to that exact loan.

That distinction separates a faster transfer rail from a faster credit business. Our 51 Insights report estimates that working rails can compress funding from 5 to 15 days to 1 to 3 days, distributions from T+5 to T+30 to near real time, and refinancing from 4 to 8 weeks to 1 to 2 weeks. The loan does not become safer because it is onchain. The capital around it can move with less waiting and less clerical work.

About the guests

• Juan Montero is co-founder and CEO of Fence. The company builds operating infrastructure for asset-backed finance, from loan onboarding and borrowing-base calculations to covenants, cash flows, and reporting. Fence says it administers about $1.5 billion across live facilities.

• Anant Matai works across investments and product at Grove, an institutional credit protocol backed by Sky. Grove allocates onchain capital to tokenized credit and provides financing against eligible digital assets.

“The lenders are here ... what we’re really looking for is that end to end originators to originate assets that are on chain and borrow against them on chain.”

• Mike Manning is Head of Institutional Finance at Ava Labs, which develops the Avalanche network. He previously led blockchain and digital-currency work at Amazon and held roles at Provenance and Symbiont.

“This is a settlement and operational technology. It’s not a judgment technology.”

The discussion was part of Same Loans, Better Rails: The Institutional Rebuild of Private Credit, presented by 51 Insights and Avalanche.

Chapters

00:00 The $2 Trillion Private Credit Problem
00:56 Meet the Panel
02:38 Why Private Credit Still Runs on Spreadsheets
06:38 Blockchain Without the Hype
09:44 Bringing Institutional Credit Onchain
12:41 Tokenization vs Onchain Lifecycle Management
15:04 The Money Rail Meets the Asset Rail
17:28 How Banks Save 80% in Operational Costs
19:06 Can Blockchain Prevent Double Pledging?
21:59 Is Regulation Holding Blockchain Back?
25:01 Why Banks Will Adopt This
26:50 Which Blockchain Should Institutions Use?
29:00 What Really Needs to Go Onchain?
31:56 The Future of Onchain Credit
34:13 Final Thoughts

Important links

• Event page and guest details

• 51 Insights report: The tokenized private credit opportunity

• Fence case study: BBVA and Payflow

• Grove Allocator

• Grove’s $250 million Avalanche deployment target

• Avalanche for institutions

1. The wrapper is not the product

Thesis: A tokenized claim is useful. A digitally native loan lifecycle changes the economics.

Mike drew a clean line between putting an ownership wrapper onchain and moving issuance, servicing, covenants, repayment, and settlement onto the same system.

“Asset-backed finance and securitization is composable finance without a composable stack.”

A wrapper gives investors a new way to hold an asset. It does not automatically change how the loan is originated, checked, funded, serviced, or enforced.

• The underlying loan needs a durable digital identity.

• The credit agreement needs machine-readable eligibility, covenant, and waterfall rules.

• Cash movements need to reconcile to the same loan record.

• The legal documents need to say which record controls ownership.

What to do with this: When someone pitches a tokenized credit product, ask where the underlying loan, covenants, servicing events, and repayments live. If the answer is still email, PDF, and spreadsheet, the token changed the wrapper, not the operating model.

Related read: The tokenized private credit opportunity

🚀 Build credibility. Drive pipeline. Win in digital assets. We produce institutional-grade research that helps digital asset companies own a category, then distribute it to 100,000+ decision-makers. Let’s talk.

2. Capital velocity is the business case

Thesis: The most valuable output is not a token. It is less time between creating an eligible loan and receiving the capital that funds it.

In a weekly borrowing-base process, an originator may carry days of new loans on its own balance sheet. If each loan enters the facility as soon as it passes the agreed rules, that inventory can fall sharply.

“Now my balance sheet as an originator drops from a week’s worth of production to one loan’s worth of production.”

“It turns a capital intensive model into a capital light model.”

Juan used a Fence facility with BBVA as the proof point. Fence’s published case study says the system handles 100,000 transactions a month, reduced interest expense by about 30%, and increased the advance rate by more than 10%. Fence separately reports up to 80% lower operating overhead and up to 40% lower cost of capital across clients. These are company-reported outcomes and should be read that way.

What to do with this: Measure idle funding days, reconciliation hours, advance rates, and the cost of carrying unfunded loans. Those numbers tell you whether new rails changed the business.

Related read: Inside JPMorgan’s $3T tokenization machine

3. One ledger does not stop every double pledge

Thesis: A ledger can prevent two claims against one digital asset only when that asset is the authoritative legal record from origination.

The panel discussed the alleged double and triple pledging in the collapses of First Brands and Tricolor. A shared record can make duplicate claims visible, but only inside the system that everyone recognizes.

“You can prevent double pledging of an asset once that it is on chain, but then the question is, well, how do I know the asset that’s on chain is the original one?”

A digital twin can still be pledged on one chain, another chain, and a traditional facility. A signature can prove who submitted a record. It cannot prove that the offchain asset exists or that no competing claim sits elsewhere.

• Origination must create the first authoritative asset record.

• Legal documents must define control and priority.

• All relevant lenders need access to the same ownership state.

• Servicing data and cash need to reconcile to the asset continuously.

What to do with this: Ask which record has legal priority, whether the same receivable can still exist offchain, and what happens when a servicer or borrower disputes the data.

Related read: DTCC’s $20T October debut

4. Lenders are ready. Originators are the bottleneck

Thesis: Capital is already willing to move onchain. The missing piece is a deeper supply of loans that are created, funded, and serviced onchain from the start.

“The lenders are here ... what we’re really looking for is that end to end originators to originate assets that are on chain and borrow against them on chain.”

Grove shows the allocator side of the market. As of August 2026, its site shows $2.80 billion in TVL and 16 active allocations. Its first Sky mandate put more than $1 billion into Janus Henderson’s JAAA strategy. On Avalanche, Grove announced a $250 million deployment target, which is different from saying the full amount has already been deployed.

The harder step is moving beneath the fund token. Originators need loan data, documents, controls, servicing events, and payment rails that lenders can rely on without rebuilding the deal by hand.

What to do with this: Build the allocator roadmap around originators who can create a legally native loan record and keep it current. Separate capital allocated to tokenized wrappers from capital funding digitally native loans.

Related read: Same loans, better rails

5. Code should run operations, not judge credit

Thesis: Smart contracts can execute the agreed rules. They should not decide whether a borrower deserves capital.

“Code should do ninety nine percent, we should flag that one percent.”

“This is a settlement and operational technology. It’s not a judgment technology.”

There is plenty to automate: eligibility checks, concentration limits, payment status, waterfalls, borrowing bases, and defined covenants. Future-receivables facilities are especially clean examples. A missed payment can remove a receivable from the borrowing base, while a digital-cash repayment maps to that loan and releases capital for the next one.

Underwriting, audited accounts, change-of-control analysis, bespoke waivers, and enforcement remain human work. The system also needs a controlled way to handle exceptions. A rigid smart contract that cannot process a sensible waiver creates a new operating problem.

What to do with this: Draw a line through the workflow. Put deterministic checks and cash movements below it. Keep credit judgment, legal interpretation, and exception approval above it, with a clear audit trail between the two.

Related read: Where automation ends in tokenized private credit

6. Serious counterparties beat endless chain choice

Thesis: Institutions choose a market, not a feature list.

Mike argued that optionality can become fragmentation. A bank does not care that a product can move across 100 networks if the counterparties, cash rails, legal arrangements, and liquidity it needs are scattered across all of them.

“What opportunity through optionality does is it ... institutionalizes fragmentation.”

Juan and Anant added the practical tests: resilience, uptime, security, observable infrastructure, builders who can support the product, and reliable on- and off-ramps. Distribution comes from depth in a room that serious participants already want to enter.

What to do with this: Choose infrastructure by the counterparties it can convene, the legal and cash workflows it can support, and the reliability it has proved. Maximum portability is a poor substitute for one market that works.

Related read: Banks went onchain

The bottom line

The skeptic has a strong case. A bad loan does not become a good loan because its data is onchain. Enforcement still happens in courts. Bespoke credit still needs judgment. And one important question from the discussion went unanswered after the connection dropped: how does an allocator offer fast withdrawals when the underlying loans may take weeks to sell? Better records do not remove the liquidity mismatch.

But that is not the main claim these systems need to prove. They need to show that a good loan can be funded, monitored, paid, and transferred with less idle cash and less clerical work. Fence’s live facilities suggest that operating gain can be large. Grove shows that institutional capital is already willing to use the rails. Avalanche is betting that the market will gather around a small number of networks with enough depth to support it.

Underwriting decides whether the loan deserves capital. The rails decide how long that capital sits still.

That’s all for now, folks.

– Marc & Team

Share

What’s happening: On August 18, N3XT, a narrow bank in Cheyenne, opened its N3XT Digital Dollar to wallets it does not bank: public Ethereum, around the clock, non-account holders. Every other US bank deposit token is still a claim…

Read more

In today’s issue:

• Goldman pays up to $2.25B for an income-ETF shop,

• Fidelity turns staking into a dividend,

• Washington’s agencies lap Congress,

• and stablecoins ride the digital pound’s test rails.

One theme runs through all of it: nobody bought exposure this week. Everybody bought income.

Let’s get into it.

PS: This week, we’re testing a new format. Simpler, lighter, sharper. Tell us how you like it at the bottom of the newsletter.

Subscribe now

Goldman Buys the Yield, Not the Coin

Coupon clippers.

Goldman Sachs agreed to acquire NEOS Investments for up to $2.25B in cash and equity, its second ETF deal this year.

What’s happening: NEOS, founded in 2022, runs $30B across 19 options-based income ETFs. The one everyone’s watching is BTCI, its bitcoin high-income fund: $1B+ in assets, under 4% of the deal, and an advertised annualized distribution rate around 27%.

• Bitcoin trades near $64K, about half its October high.

• Ether is down about 37% since January. The 10-year pays 4.65%.

• Derivative-income ETFs are already a $180B category, growing 70%+ a year since 2021, per Morningstar data cited by Goldman.

When the price stops paying, investors want the asset to.

Why it matters: Goldman filed its own bitcoin premium income ETF in April. It never launched. Four months later, it agreed to pay up to $2.25B for the firm that got there first.

In ETFs, track record and assets compound. Buying two years of head start beats building. BlackRock’s rival bitcoin income fund launched in June and holds $59M. BTCI holds $1B+. It’s rare to watch BlackRock lose a category, and Goldman is paying to keep it that way.

Between the lines: BTCI’s payout comes from selling options against bitcoin’s price, and part of each distribution can be your own capital coming back. The advertised rate is ~27%; the one-year total return was about minus 42%.

When someone quotes a crypto income ETF’s yield, ask for total return.

Punchline: Wall Street figured out how to charge active fees on a passive asset. Beta compressed to a few basis points; a 27% distribution supports a 0.99% expense ratio. The product being sold is changing from the price to the paycheck.

Fidelity makes ether pay a dividend

Staking claims.

Fidelity filed to let its $898M ether ETF stake up to 100% of holdings and pay rewards out as quarterly cash distributions.

The details:

• The fund keeps 85% of gross staking rewards; 15% goes to the sponsor, custodians, and node operators Blockdaemon, Figment, and Galaxy.

• Staking starts only once the SEC declares the filing effective.

• Grayscale has staked its ether ETP since October; 21Shares pays distributions; BlackRock’s ETHA still doesn’t stake.

• Meanwhile: a record 34% of all ETH is now staked, while ether fell a third from January. And Galaxy’s $125M on-chain yield fund went live with $100M of SharpLink’s ETH treasury, committed in May.

Why it matters: The ETF wrapper keeps absorbing reasons to hold crypto directly. First custody, then options, now the staking coupon, paid in cash like a dividend.

Between the lines: The headline fee stays 0.25%, but 15% of staking rewards is a second toll that never shows up in the expense ratio. At today’s rates, that’s roughly another 0.4% a year if the fund stakes everything.

Looking ahead: Expect every US ether ETF to file the same feature within months. “The fund pays you” is a better pitch than “the fund tracks ether,” especially with ether down 37% this year.

Compare staking ETFs on the reward split, not the expense ratio. The split is where the real fee hides.

Subscribe now

Washington’s Agencies Stopped Waiting

Two clocks.

Three moves in five days put the regulators ahead of Congress.

What’s happening:

• Majority Leader Thune filed cloture on the Clarity Act, setting a 60-vote test on September 15 to open floor debate.

• The SEC scheduled a vote for this morning on proposing “Regulation Crypto Assets,” its first crypto-specific rulemaking. Then it cancelled the meeting hours before the gavel. No reason posted, no new date.

• The OCC reported 40 new bank charter applications in 18 months, 13 from digital-asset firms, Kraken and Revolut among them. Comptroller Gould: firms dealing in digital assets “should have a path to becoming a national bank.”

Why it matters: Galaxy just cut the Clarity Act’s 2026 odds from 50% to 30%. So the agencies are building their own regime instead: a proposed offering rule at the SEC, a charter pipeline at the OCC.

For a company, a charter in 2026 beats a statute in 2028.

Between the lines: This morning’s pulled meeting is the tell. Agency timelines can vanish overnight, and what an agency gives, the next administration can take back. Rules reverse; statutes stick. The rational play is the one the industry is making: take the charter now, keep pushing the bill.

Looking ahead: September 15 needs seven Democrats. Watch that count.

Related: 187: Wall Street flipped the switch

Stablecoins ride the digital pound

Plumbing, not an app.

Polygon Labs published details of its test inside the Bank of England’s Digital Pound Lab, run with NOBO Finance and Dun & Bradstreet.

What’s happening: In the experiment, an exporter gets paid on stablecoin rails while the UK importer settles in simulated digital pounds on the Bank’s test system. One payment, both forms of money. No real funds moved; Phase 2 wrapped in July, and findings feed the Bank and Treasury’s digital pound decision later this year.

Why it matters: Central banks spent years framing CBDCs and stablecoins as rivals. The Bank of England’s lab just let a participant team run them in the same payment. That’s a quiet shift.

Between the lines: The Bank’s June draft rules point the same way: a temporary £40B issuance guardrail per systemic sterling stablecoin, and up to 70% of reserves in short-term gilts.

Punchline: The UK design is coming into focus. The central bank settles the core; regulated stablecoins do the reach. A digital pound as plumbing, not an app.

⚡ Quick Hits

• Nasdaq agreed to acquire LeveL Markets, the third-largest US ATS, anchoring a new always-on markets unit.

• MUFG will test real-time settlement of Japanese government bond repos on Canton, a ~$1.7T market.

• Wintermute registered as a US broker-dealer and plans ~$1B of AI and trading-infrastructure spend, per Bloomberg.

• Securitize reported Q2 volume of $5.3B, up 147%, in its first earnings report as a public company.

• The US Treasury sanctioned Iran-linked exchanges Shelbit and Aban Tether for moving IRGC funds.

• The Bank of Russia cleared bitcoin, ether, and USDT for exchange trading, with a proposed retail cap near $3,600/year per broker.

• Bybit secured a US court injunction freezing assets from its $1.5B hack after suing North Korea and Lazarus directly.

• Anchorpoint, the Standard Chartered-led venture, began rolling out HKDAP, Hong Kong’s first regulated local-dollar stablecoin.

• Coinbase made Abu Dhabi its global tokenization hub after winning an ADGM permission covering tokenized securities.

💰 Money Moves

• Erebor, the crypto-friendly bank backed by Palmer Luckey, is in talks to raise $1.5B at a $9.5B valuation, six months after launch, per the FT.

• NVIDIA partnered with Apollo, BlackRock, Blackstone, Brookfield, Goldman, and KKR on platforms targeting $500B+ of third-party capital for AI compute. MoUs for now, not committed capital. The pitch: treat the GPU like a building that pays rent.

• Morgan Stanley launched an initiative to facilitate ~$1.5T of US innovation-infrastructure financing over ten years.

• Riot Platforms signed a 20-year, $9.1B lease for 191MW of AI capacity with a frontier AI lab, which Bloomberg reports is Anthropic.

Same instinct across chips and crypto this week: everything becomes an income product.

51 View: The last time tech vendors financed their own customers at scale, it was Lucent and Nortel in the dot-com years. It ended badly.

📚 Boardroom Reads

• Stablecoins in Emerging Markets (IMF). South Africa’s dollar-stablecoin trading grew from under ZAR 4B to ZAR 80B in three years.

• FCA departs from the traditional stablecoin model (OMFIF). UK issuer capital set at 1% of issuance, half of MiCA’s 2%.

• Solana Q2 2026 (Galaxy Research). RWA holdings crossed $3B; public equities now the largest category.

• Violent Crypto Wrench Attacks: H1 2026 (Chainalysis). $30M stolen in physical attacks in H1, pacing past 2025’s record.

• The AI investment race (BIS). Models AI over-investment at 1.5x the efficient level, against $700B of 2026 capex.

📅 On the Calendar

• Date TBD: SEC’s cancelled Regulation Crypto Assets vote; watch for a new Sunshine Act notice

  This post has bonus content for paid subscribers. Upgrade to get full access.

That’s all for now, folks.

– Marc & Team

Ps: Save your spot for our next webinar. Space is limited.

RSVP

I’m sitting down with the people dvising the banks and building the stablecoin rails those banks will plug into:

• ​Christian Schmid, Managing Director & Senior Partner at BCG, global lead for Corporate & Investment Banking.

• ​Alexander Paddington, Managing Director & Partner at BCG, Global Leader of Transaction Banking

• ​Sergio Mello, Global Head of Stablecoin Solutions at Anchorage Digital

35 minutes of moderated panel. 10 minutes of live Q&A.

For CEOs, board members, and heads of strategy at banks, FMIs, asset managers, and custodians.

📅 September 16 at 10:00 AM ET.

🚨 Space is limited. RSVP to secure your spot.

RSVP

What’s happening: On August 4, Cloudflare launched Cloudflare Wallets and cloudflare.pay, giving AI agents a verified identity and a stablecoin wallet their owner controls. Investors liked the look of it; shares jumped 6.1% to $300.11, a 52-week high.

Subscribe now

Topline

The pitch isn’t that agents can pay. It’s that they can be traced, capped, and cut off. Here’s how it works:

• Account Wallets hold the money. A human funds them, sets the rules, and can pull the plug.

• Virtual Wallets go to individual agents, each with a spending allowance, an approved-recipient list, a max transaction size, and automatic flags for weird spending.

• cloudflare.pay handles (think: research.acme.cloudflare.pay) give each agent a public identity, backed by cryptographic keys via Web Bot Auth.

• Settlement runs in USDC across ten chains, with Coinbase’s x402 facilitator as the default. Handles are free and live now; funding and wallet issuance land “in the coming months.”

Handle reservation is free and live today. Funding, onramps, and Virtual Wallet issuance arrive “in the coming months.”

Settlement runs in USDC across ten chains, including Base, Ethereum, Solana, and Polygon, with Coinbase’s public x402 facilitator as the default. The x402 protocol has settled 160.6 million transactions worth $41.2 million, roughly 26 cents each ($41.2M ÷ 160.6M = $0.257).

CEO Matthew Prince put the pitch in one line: “When an agent shows up at your door, you need to know who sent it. Cloudflare can give agents a face — a link to the human or organization that owns them.”

This ad could belong to you. 🚀 Build credibility. Drive pipeline and revenue. We position you among 100,000+ digital asset decision-makers.

Let's talk

Why it matters

• Cloudflare rebuilt the corporate expense program with agents as the employees. An employee gets a card, a limit, a vendor list, and a manager who can cancel the card. A Virtual Wallet is the same object: a budget, an allow list, a hard cap, an off switch. Enterprises do not adopt autonomy, they adopt auditability. Nobody signs off on software that spends freely. Everyone knows how to sign off on a card program, because procurement has run them for fifty years. The design also bounds the new attack surface: a prompt-injected agent with a $100 weekly allowance and a five-vendor allow list is an expense report problem, not a treasury event.

• Roughly 20% of all websites already route through Cloudflare, and it now monetizes both directions of agent traffic. Since July 2025, its pay-per-crawl program has charged AI crawlers for access to content. That is the seller side. Wallets fund the agents doing the paying. That is the buyer side. Cloudflare already terminates the traffic; now it wants to clear both legs of transactions it proxies. There is an irony worth noticing: the company that built its reputation blocking bots is becoming their passport office. Web Bot Auth turns “this bot is legitimate” into “this bot is legitimate, and here is its budget.” A small technical step, a large commercial one.

• Cloudflare announced its NET Dollar stablecoin in September 2025. Eleven months later, Wallets ship with USDC and Coinbase’s facilitator. NET Dollar has not launched. The read-through: distribution beats issuance. Even a platform with Cloudflare’s reach concluded that plugging into existing USDC liquidity beats bootstrapping a new coin. Every corporate treasury weighing “should we issue our own” should study this sequence. The wallet, identity, and policy layers are where differentiation lives. The settlement asset is becoming a default you adopt. Circle wins another default it did not have to fight for.

• The average x402 transaction is 26 cents. The standard card fee on that same transaction is 31 cents (2.9% × $0.26 + $0.30 fixed = $0.31, per Stripe’s published standard US card pricing). Card rails are excluded from agent payments by arithmetic, not ideology. Honesty requires the caveat: those 160.6 million transactions include test and infrastructure traffic, and $41.2 million in settled value is a rounding error next to the $7.5 trillion stablecoins moved in March alone. The volume proves the mechanism works. It does not yet prove the market exists.

Between the lines

Read more

“We did over $1.2 trillion of onchain trading volume across spot and perps.”

One app on Solana processed $1.2 trillion in trading volume. Not an exchange with servers and order books. An app that runs entirely on a blockchain. Its president spent years at KKR, one of the largest private equity firms in the world, before he switched sides.

His name is Xiao-Xiao J. Zhu, President of Jupiter, the biggest DeFi platform on Solana.

Watch on Youtube

At KKR he led the firm’s digital assets and blockchain strategy. Now he runs a company with 20+ onchain products, $3 billion in TVL, and a plan to put US stocks, a stablecoin, a neobank, and AI agents on the same rails. His core argument is simple: the value in crypto has moved from blockchains to applications. He calls it the fat app thesis. And he thinks onchain finance is still 10x to 100x smaller than its real market.

This isn’t a recap. It’s the playbook: the six best ideas from the conversation, the exact quotes, and what to do with each one.

About Xiao-Xiao J. Zhu: Xiao-Xiao J. Zhu is President of Jupiter, Solana’s largest onchain finance platform. Before Jupiter, he was Digital Operating Partner at KKR, where he led technology value creation across the portfolio and ran the firm’s global digital assets and blockchain strategy, backing crypto funds and companies including Anchorage Digital. He has seen both sides: how value gets built in traditional private equity, and how it gets built onchain.

“The biggest companies in crypto are basically centralized exchanges or market makers who are extremely intransparent and are running on centralized databases.”

Why this matters: Jupiter is what the next generation of financial institutions might look like. It started as a DEX aggregator three or four years ago. Today it is the number one trading venue and the number one TVL protocol on Solana, it launched a stablecoin backed by BlackRock’s BUIDL fund, and in May it put regulated US equities onchain with Jump Trading and Securitize. Robinhood, Coinbase and OKX already route through its APIs. We recorded this live at Proof of Talk in Paris. Here it is in six ideas.

🎯 Jump to the best parts

[00:00] Cold open: $1.2 trillion and the road map
[00:30] Live from Proof of Talk in Paris
[01:05] From KKR to the biggest DeFi app on Solana
[02:26] The inflection point: blockchains finally got fast
[03:12] The fat app thesis
[03:43] What Jupiter is: 20+ products, $1.2T in volume
[04:44] Jupiter Lend, JLP, and their own stablecoin
[05:52] Why “onchain finance,” not DeFi
[07:56] The 100x gap: millions of users vs. Binance’s 300M
[09:00] The two unlocks: RWAs and agentic finance
[10:31] Agents don’t do KYC
[11:39] The Jupiter agent kit is live
[12:37] How institutions plug in today
[14:29] Bitwise and up to $1B into Jupiter Lend
[15:11] Jupiter Global: the onchain neobank
[15:40] Tokenized US equities with Jump and Securitize
[17:07] What tokenized stocks actually unlock
[17:52] The road map: super app, neobank, JupNet
[18:38] Wrap

Important Links

• Jupiter: https://jup.ag

• Securitize / Jump / Jupiter tokenized equities announcement: PR Newswire

• Jupiter Lend x Bitwise (Ethena market): PR Newswire

• LinkedIn: https://www.linkedin.com/in/xiao-xiao-j-zhu-12078730

Watch or listen now: YouTube • Apple Podcasts

🔒 The full breakdown is for subscribers

Our biggest takeaways from this conversation

1. The value moved from blockchains to apps.

For years the money in crypto was made at the protocol layer. You bought the chain, not the things built on it. Zhu says that flipped, and it flipped because blockchains finally got fast enough to build real products on.

“Value was initially in crypto created at the protocol level, at the blockchain level, to now really an era of the fat app thesis.”

The irony he points out: everyone came to crypto for decentralization, but the biggest crypto companies are centralized exchanges and market makers running on ordinary databases. Not because they were lazy. Five to seven years ago, chains simply couldn’t handle the volume.

• The inflection came in the last two to three years, when Solana and newer L1s and L2s solved most of the scalability problems.

• The result is a new generation of apps like Jupiter and Hyperliquid: permissionless, self-custodial, and built fully onchain, at global scale.

• The question he thinks matters now: what applications, brands and user experiences can you build at scale on top of blockchains? Not which chain wins.

What to do with this: if your digital asset exposure is all protocol-level, you own the last cycle’s thesis. Look at where usage and fees actually accrue now: the application layer.

Related reads:
→ 184: Kraken is buying DeFi

This ad could belong to you. 🚀 Build credibility. Drive pipeline and revenue. We position you among 100,000+ digital asset decision-makers.

Let's talk

2. Jupiter is Robinhood, rebuilt fully onchain.

Read more

On the podcast this week, BCG’s global banking lead Christian Schmid gave me the line that ended up explaining the whole week:

“Waiting isn’t a strategy.”

Nobody waited this week. Mastercard closed its $1.5 billion BVNK acquisition on Monday, with up to $300 million more tied to an earnout. Western Union shipped a stablecoin card on Tuesday. Visa switched on stablecoin payouts for Visa Direct clients on Wednesday, on a network reaching 18 billion endpoints. And Cloudflare gave AI agents their own stablecoin wallets.

Last week we showed how the banks took the settlement layer with tokenized bank money.

This week the stablecoin side answered at the front door: the card, the app, the phone, the agent.

Circle’s earnings, out Wednesday, explain the rush. USDC moved $14.8 trillion on-chain last quarter, up 151% in a year. But the float, the pile of dollars Circle earns interest on, averaged just 25% growth, and the quarter-end pile was smaller than three months earlier. The stock fell about 4% anyway. The money is becoming a rail, and rails pay whoever owns the customer, not whoever holds still.

What to expect now: more incumbents buying the conversion layer, and stablecoin economics splitting in two: float income for issuers, transaction income for whoever owns the front door.

Below we unpack: the week’s deals, where the Clarity Act endgame stands, and why Circle’s chart matters more than its earnings.

Signals at a glance:

• Circle moved $14.8 trillion and the stock fell anyway

• Mastercard and Visa moved on the stablecoin conversion layer

• Western Union shipped a stablecoin card

• Wells Fargo tokenized the deposit

• Tether halved its safety buffer while making $1.5 billion

And 10+ more signals below.

Subscribe now

The 51 Signal

Whoever owns the front door owns the business.

Read with PRO.

📍 Clarity Act: State of Play

The last pre-recess window. Majority Leader Thune said Sunday a vote would happen before the August recess. As of Wednesday, no cloture motion had been filed. Still open: illicit-finance provisions and the ethics language on senior officials profiting from crypto. Senator Lummis told Fox Business: “I don’t think we’ll be leaving on Friday. I think we’ll go into the weekend.” If it slips, the next plausible window is after the elections. The industry is pushing hard; Chris Dixon’s case for the bill argues the GENIUS Act showed that clear rules can draw investment.

Top Boardroom Reads & Data

• Stablecoins and (Non)Crypto Shocks: A 2026 Update (NY Fed Liberty Street Economics, July 31, 2026). How shocks from outside crypto reshape the reserves inside it.

• Dollarisation and monetary control: what lessons for the rise of stablecoins? (BIS Working Paper 1370, July 21, 2026).

• It’s time to pass the CLARITY Act (Chris Dixon, a16z crypto, July 22, 2026). The industry’s most complete brief for the bill in the Senate’s last pre-recess window.

• 3 charts on the tokenized stocks boom (a16z crypto, July 20, 2026). Tokenized stocks hit ~$1.7B market cap, 5x in a year, with monthly transfer volume up 170x.

Subscribe now

This ad could belong to you. 🚀 Build credibility. Drive pipeline and revenue. We position you among 100,000+ digital asset decision-makers.

Let's talk

Circle’s quarter-end float fell

What happened: Circle reported Q2 results on Wednesday: total revenue and reserve income of $701 million, up 7% year over year but slightly under Street estimates, with net income of $48 million. USDC ended the quarter at $73.3 billion in circulation, down about 5% from the prior quarter, while average circulation rose 25% year over year. On-chain transaction volume hit $14.8 trillion, up 151%. Circle nearly doubled its 2026 “other revenue” guidance to $310-330 million and confirmed its Arc blockchain goes to mainnet on September 16, with validators including BlackRock, DTCC, Visa, Mastercard, ICE, Standard Chartered, MoneyGram and Sumitomo. The stock closed down about 4% at $63.25.

“We have built the platform for the internet financial system – for traditional and digital finance, real-world assets, and the institutions that move the world’s capital.”
Jeremy Allaire, CEO of Circle, in the results release

51 View: The bull case says USDC is becoming the internet’s settlement rail and Arc gives Circle a way to charge for the flow. The bear case says Circle is a money-market fund whose assets just stalled, dressed up as a network. The September 16 mainnet launch is Circle reaching for network economics on top of its float business. Look at who signed up to validate it: the same card networks and custodians that spent this week wiring into distribution. They would all rather own a seat on the rail than rent one.

Go deeper with PRO:

The card networks moved on the conversion layer

What happened: Mastercard completed its acquisition of BVNK on Monday. It paid $1.5 billion, plus up to $300 million in an earnout. BVNK moves roughly $30 billion a year in stablecoin payments across 130 markets, with 25+ licenses. Its clients include Worldpay, Deel, Rapyd and Flywire. Two days later, Visa and zerohash switched on stablecoin prefunding and payouts for eligible Visa Direct clients. That rail reaches more than 18 billion cards, accounts and wallets in 195+ countries.

“...stablecoins are increasingly addressing real-world needs in areas like cross-border B2B payments, remittances, payouts, settlement and treasury flows.”
Jorn Lambert, Chief Product Officer at Mastercard, in the closing announcement

51 View: A card network cares less about whose dollar moves than that it moves through them. Mastercard is the first big listed network to buy stablecoin infrastructure outright rather than partner into it. The price tells you the urgency. The conversion layer now carries billion-dollar price tags because it’s one of the few layers most users still can’t route around. Somewhere, crypto dollars have to become spendable ones.

Western Union turned the remittance into a stablecoin

What happened: Western Union launched Stablecard on Tuesday with card-infrastructure firm Rain. It’s live in 37 markets, with a target of 60 by year end. The app receives Western Union transfers directly into USDPT, the company’s dollar stablecoin issued by Anchorage Digital Bank on Solana. A linked Visa card spends it at 175 million merchant locations, with Apple Pay and Google Pay support. No credit check, no minimum balance. Fireblocks provides the wallet infrastructure.

51 View: Look at what Western Union kept and what it gave away. It kept the customer, the brand and the app. It handed the money to a federally chartered crypto bank, the settlement to Solana, and the spending to Visa. Everyone said stablecoins would kill this company. Instead, the 175-year-old incumbent used them to ship the product stablecoins were always supposed to be: a dollar account for people without a bank. A remittance that used to end as a cash pickup can now end as a dollar balance with a Visa card attached. Watch the receive side. Western Union moves roughly $100 billion a year. If even a small share starts parking in USDPT, Western Union’s corridors start doubling as stablecoin distribution.

Wells Fargo tokenized the deposit

What happened: Wells Fargo announced tokenized deposits for corporate and commercial clients on Tuesday. Money moves 24/7/365, inside the regulated, insured banking system. Rollout starts this fall with a limited US dollar to British pound corridor, then expands through 2027. Wells is the last of the big four US banks to commit, after JPMorgan’s Kinexys, Citi and Bank of America. The shared bank deposit-token network is still planned for 2027.

“Tokenized deposits will enable Wells Fargo’s corporate and commercial clients to move money between accounts and across borders with greater ease and increased speed and builds on the strength of our established banking infrastructure.”
Mike Santomassimo, CFO of Wells Fargo, in the announcement

51 View: Last week we argued banks are tokenizing deposits to defend the balance sheet. Wells Fargo completes the set. All four US mega-banks are now in. The design choice is the real signal. Wells is folding tokenized deposits into its existing payment services rather than launching a separate crypto product: no new app, no new onboarding, no “digital assets” button.

Subscribe now

Tether made $1.5 billion and halved its cushion

What happened: Tether’s Q2 attestation, published July 31, showed $187.75 billion in assets against $183.64 billion in liabilities: a $4.11 billion excess-reserve buffer, roughly half what it was a quarter earlier. Q2 net operating profit came in around $1.5 billion, driven by about $115 billion in direct US Treasury exposure. Gold holdings rose 14 tons in the quarter to 146.2 tons, plus 98,932 bitcoin. USDT supply sat near $184.6 billion.

51 View: Tether already owns what everyone else spent this week buying: distribution, with $184 billion in float across the emerging markets where dollars are scarcest. What it traded away this quarter is margin for error. The buffer protecting USDT holders from a bad day halved, while the balance sheet tilted further into gold and bitcoin. Bitcoin tends to fall exactly when crypto stress hits, and gold carries price and liquidity risk of its own. Neither is an eligible asset for the GENIUS Act’s required one-to-one reserve pool. The next stress test will show if the balance sheet has enough room for error.

Go deeper:

News Flashes

Infrastructure and Markets

• BNY announced plans to add staking to its digital-asset custody platform with Galaxy, so institutional clients can earn staking rewards without assets ever leaving BNY custody.

• Bitcoin ETFs snapped back with $211.5 million of inflows Tuesday and $244.4 million Wednesday, their strongest day since late July, following a sharply negative month.

Banking and Payments

• Samsung is poised to build native stablecoin savings and payments into Samsung Wallet, analysts say, potentially putting stablecoins on as many as 800 million Galaxy devices, on top of its $408 million stake in Upbit operator Dunamu.

Funds, Deals and Others

• Galaxy posted an $85 million Q2 net loss on crypto markdowns but delivered its first revenue-generating data-center quarter, with 133 megawatts live for CoreWeave and a contract pipeline Galaxy values at roughly $30 billion.

• Strategy sold 1,638 bitcoin for $104.7 million to fund preferred dividends and buybacks, lifting its cash reserve to $4 billion while holdings slipped to 842,138 BTC.

The AI Layer

• Cloudflare launched Wallets and the cloudflare.pay namespace, giving AI agents a stablecoin balance, spending guardrails and a permanent identity to pay for APIs over the x402 protocol. Circle says USDC already carries 99.3% of x402 payment volume.

• Singapore’s MAS confirmed that its forthcoming AI risk guidelines will cover firms deploying autonomous agents, making it one of the first major financial regulators to say so explicitly.

• Microsoft’s 10-K pointed to OpenAI generating $24.1 billion, or roughly 70% of Microsoft’s AI revenue, by Bloomberg’s analysis of the filing, which also disclosed $6 billion in receivables from one unnamed customer.

“When an agent shows up at your door, you need to know who sent it. Cloudflare can give agents a face — a link to the human or organization that owns them — so that trust, accountability, and real commerce can follow.”
Matthew Prince, CEO of Cloudflare, in the announcement

51 View: The agent-payments stack is assembling in public, layer by layer. Two weeks ago agents got wallets inside ChatGPT. This week they got identities, spending limits and a first explicit mention in a major regulator’s planned guidelines. Notice the rail underneath all of it: stablecoins, with USDC carrying essentially all x402 volume.

Watchlist / On the Calendar

• Friday, August 7: the July jobs report lands the morning this email does.

• This weekend: the Clarity Act’s last window before recess. Lummis says the Senate may work through it.

• September 16: Circle’s Arc mainnet launches, with Wall Street as validators.

• This fall: Wells Fargo’s USD-GBP tokenized deposit corridor goes live.

• October: DTCC’s Tokenization Service targeted go-live.

• Q4: MAS finalizes its AI risk guidelines, agentic AI included.

• GENIUS Act: proposed rules out, none final; January 18, 2027 holds.

That’s all for now, folks.

– Marc & Team

Read more

“Waiting and seeing is not a strategy.”

A senior partner and global banking lead at BCG just put a number on the thing every bank CEO is nervous about: up to 15% of bank revenue and 30% of profits at risk by 2035. But the opportunity is bigger than the threat.

Watch on Youtube

His name is Christian Schmid, Managing Director and Senior Partner at BCG, where he leads the global banking business. He is the lead author of BCG’s biggest ever digital assets report, The Future of Digital Assets.

He calls the whole shift the iPhone of money, and he thinks the real fight is not stablecoins versus deposits. It is who owns the screen the customer taps.

Download the Full Report

About Christian Schmid: Christian Schmid is a Managing Director and Senior Partner at BCG, based in Zurich, where he leads the firm’s global commercial banking, capital markets and investment banking business and chairs the board of BCG Expand. He trained as an engineer at ETH, came up through IBM, and has spent 27 years advising the CEOs of the world’s biggest banks. He is the lead author of The Future of Digital Assets.

“That’s probably the most interesting topic I have seen in my whole consulting career, which is twenty seven years almost by now.”

Why this matters: This is the year banks stopped watching from the sidelines. The GENIUS Act gave stablecoins a US rulebook, the market is now north of $300B, and the biggest US banks, JPMorgan, Citi and Bank of America, are building a shared tokenized-deposit network aimed at around 2027. BCG puts the tokenization prize near $88 trillion. So when the person who advises these CEOs tells you what he actually thinks, in plain language, it is worth 40 minutes. Here it is in six ideas.

Subscribe now

🎯 Jump to the best parts

00:50 Introduction
01:59 Why BCG Published Its Biggest Digital Assets Report
04:13 How Banking Conversations Have Changed
05:46 Is Tokenization Bigger Than Digital Banking?
08:13 Why Banks Could Lose 30% of Their Profits
13:08 Are Digital Assets Replacing Banks?
16:15 The Three Types of Digital Assets
20:08 Where Banks Should Invest
23:20 The Biggest Real World Use Cases
26:00 What Banks Are Actually Doing
27:30 AI vs Digital Assets
30:18 Why Waiting Is Not a Strategy
32:03 Four Futures for Digital Assets
35:14 Stablecoins vs Tokenized Deposits
37:41 The Future of Programmable Money
38:36 Lightning Round
39:50 Where To Learn More

Important Links

• BCG profile: https://www.bcg.com/about/people/experts/christian-schmid

• LinkedIn: https://www.linkedin.com/in/schmidchristianzuerich

🔒 The full breakdown is for PRO subscribers

Our biggest takeaways from this conversation

1. The threat is real, the opportunity is bigger

Most consultant reports are built to scare you into buying a project. Schmid’s does the opposite. He gives you the scary number, then tells you not to over-index on it.

“Whether this number is completely correct or not is actually a bit irrelevant, I would say. But there is a threat.”

The threat comes in two parts. Banks lose fee income and net interest margin as deposits and transactions drift toward stablecoins. Then they pay again to run two sets of rails at once.

“They also need to maintain the dual rails, which increases the cost base for those banks. And so it’s almost a bit a double whammy.”

• The model: up to 15% of revenue and 30% of profit at risk by 2035, versus a world where digital assets barely develop.

• The catch he keeps repeating: it is a simulation, not a forecast. The point is the size of the stake, not the decimal.

• The upside is the part banks miss. In BCG’s numbers the biggest wins are in asset management and trading. Tokenization makes it easier to wrap more assets, and trading could need less capital and liquidity, which lifts return on equity. BCG pegs the trading upside at up to a 4% RoE bump or $1B+ for an average G-SIB.

What to do with this: don’t argue about whether it’s 30% or 12%. Run the model on your own book. The industry number hides a huge spread from one bank to the next.

Related reads:
→ BCG’s new digital asset playbook: The $88 Trillion Question

This ad could belong to you. 🚀 Build credibility. Drive pipeline and revenue. We position you among 100,000+ digital asset decision-makers.

Let's talk

2. Stop calling it crypto

The word “crypto” hides the most important distinction in the whole report. Schmid splits digital assets into three: tokenized securities, digital money, and cryptocurrency. They are not the same animal, and they don’t have the same future.

“Cryptocurrency per se is a very different thing because it’s not backed with any value, and that’s very different from digital money or digital real world assets.”

• Tokenized securities and digital money look a lot like what we already have, just in a new wrapper. In 20 years, he says, no one will care whether a security was tokenized or not.

• Cryptocurrency itself he barely grows in the model, roughly 2%. His line: “in today’s world that’s where the money is made,” but it “will not be so significant going forward.”

• The bigger point is to stop thinking in use cases and start seeing the technology as one general-purpose thing. That’s the “iPhone of money.”

“The iPhone of money, which combines all the technologies in one thing through the programmability, through DLT.”

What to do with this: when someone says “our crypto strategy,” ask which of the three they mean. If they can’t answer, they don’t have a strategy, they have a headline.

Related reads:
→ 186: banks went on-chain

3. The real fight is who owns the screen, not the instrument

Read more

The release leads with 146 tons of gold, more than most central banks hold, and $1.5B in net operating profit. The number that matters sits further down: an excess reserve buffer of $4.11B, down from $6.8B at the last report before S&P downgraded Tether to “Weak” in November.

Read more

“AI is the only focus.”

That captures the mood. Crypto is in a bear market. The excitement has moved elsewhere.

But here’s one number that stopped me this week: 80 seconds.

That’s how long it took some of the world’s biggest banks to settle a cross-border payment in the BIS Project Agorá test. But what settled wasn’t a stablecoin. It was tokenized central bank reserves and bank deposits.

Here’s what matters, though: Banks aren’t chasing speed; they’re defending the deposit.

When money settles in a stablecoin, it leaves the balance sheet, and deposits are what fund the lending business. A tokenized deposit is the same dollar, still on the bank’s books, now programmable.

This week’s signals at a glance:

• The world’s biggest banks settled cross-border payments in 80 seconds

• BNY is moving $8.6 trillion of fund records on-chain

• Ten European banks launched a blockchain they own

• MoonPay put a crypto wallet inside ChatGPT and Claude

And 15+ more signals below.

Subscribe now

One quick thing: last week, we launched the 51 Institutional Digital Asset Adoption Index. It ranks 103 financial institutions across eight capabilities using linked public evidence. If you want to see which banks are actually live—and which are still piloting—check it out at index.fiftyone.xyz.

51 Digital Asset Adoption Index

The 51 Signal (PRO)

Upgrade to PRO

📍 Where the Clarity and Genius Agenda Stands

Here is the state of play in Washington:

• GENIUS Act (stablecoins): law, but behind schedule. Signed in 2025, it takes effect January 18, 2027. Regulators have issued proposed rules but missed their rulemaking deadlines, and none are final yet. Stablecoin supply still grew to about $308 billion.

• Clarity Act (market structure): stuck on the one question crypto keeps dodging. The headline fight is ethics language on officials’ crypto ventures, where Tillis and Gallego found common ground. The real one is jurisdictional: whether a token is a security (SEC) or a commodity (CFTC) decides who supervises it, what it must disclose, and whether paying yield is even legal. Big banks are split, Goldman backs it while the Bank Policy Institute flags gaps on yield and developer liability, and Warren calls it “a giveaway” that should be “dead on arrival.” With the August 10 recess days away and no floor vote scheduled, Majority Leader Thune says it “almost certainly doesn’t have enough time.”

Top Boardroom Reads & Data

• Will Robinhood Chain Succeed Where Coinbase’s Base Stumbled? (Galaxy Research, July 30, 2026).

• No Headstands Required: Applying Securities Laws to Vaults and Onchain Lending (Galaxy Research, July 29, 2026). When a DeFi vault becomes a security, and why the design details now decide.

BNY put $8.6 trillion of fund records on-chain

What happened: BNY, the world’s largest custodian, is building rails to move its transfer-agency business onto a single on-chain ledger. Transfer agency is the record of who owns which fund shares, and BNY’s book covers about $8.6 trillion across 7.6 million accounts. It will run the old and new systems in parallel at first. First clients include Baillie Gifford, for what BNY calls the first fully native UK-regulated tokenized fund, plus BlackRock and BNY’s own Dreyfus unit.

“We think of BNY as modernizing a function that sits behind every single fund transaction by bringing the books and records onchain.”
— Carolyn Weinberg, Chief Product and Innovation Officer at BNY, to CoinDesk

51 View: If ownership updates in real time, a fund share can settle in seconds instead of days, trade around the clock, and post as collateral the moment it changes hands. The custodian is rebuilding the foundation the whole industry stands on, and it keeps the client relationship while doing it.

Ten European banks built their own blockchain

What happened: Ten European institutions launched RL1, or Regulated Layer One, on July 28. It is a member-owned permissioned blockchain, structured as a European Cooperative Society in Luxembourg, where each founder holds an equal vote. The founders are ABN AMRO, DekaBank, DZ BANK, Natixis CIB, LBBW, Crédit Mutuel Alliance Fédérale, Cecabank, SC Ventures, Chartered Investment and Seturion. RL1 runs on infrastructure built by Frankfurt fintech SWIAT, which settled more than €700 million in production over three years before ownership passed to the cooperative. It targets tokenized bonds, collateral, settlement and central bank digital currency links, and is in talks with NatWest.

“RL1 will serve as the connecting infrastructure for Europe’s digital financial market, enabling participating institutions to move from isolated tokenization initiatives to an integrated, liquid, and scalable capital market ecosystem.”
— Henning Vollbehr, Managing Director of RL1 (formerly of SWIAT), in the launch release

51 View: The last decade of bank blockchains failed the same way: one bank builds a network, invites the others, and the others refuse to route their business through a competitor’s rails. RL1 solves the politics before the technology. No single member owns it, so no rival has to route through a competitor’s infrastructure. That is why the cooperative structure matters more than the chain. It is also a very European answer to a very American problem: rather than let a US stablecoin or a US consortium set the standard, the continent’s banks pooled their own.

Go deeper with our PRO read:

Tokenized deposits move to the center of settlement

What happened: Two moves in one week put tokenized bank deposits, not stablecoins, at the center of institutional settlement. OpenAssets and Partior completed a proof-of-concept for 24/7 atomic delivery-versus-payment, using tokenized commercial bank money on Partior as the settlement asset that removes counterparty risk. Partior is backed by DBS, J.P. Morgan, Standard Chartered, Deutsche Bank and others. Separately, America’s largest banks are building a shared deposit-token network, with JPMorgan, Citi and Bank of America among them, targeting roughly 2027.

“With Partior, we’ve shown how institutions can settle digital assets, stablecoins, and tokenized deposits together on existing infrastructure.”
— Gabor Gurbacs, CEO of OpenAssets, in the announcement

51 View: A stablecoin and a tokenized deposit look identical on a screen. They are not the same thing. A stablecoin is a claim on a private issuer’s reserves. A tokenized deposit is the bank’s own money, backed by the bank, inside the banking system, covered by the rules that already govern it. This is why the banks stopped fighting stablecoins and started copying the format. They get the 24/7 rails without moving the money off their balance sheet or outside the regulatory perimeter.

Morgan Stanley undercut the market on staked crypto

What happened: Morgan Stanley Investment Management launched two exchange-traded products on July 28: the Morgan Stanley Ethereum Trust (MSSE) and the Morgan Stanley Solana Trust (MSOL), both on NYSE Arca. Each charges a 0.14% sponsor fee, among the lowest in the market. Both stake a portion of their holdings and pass the staking rewards through to shareholders. The products follow the firm’s bitcoin trust and join a $14 billion ETP suite; MSIM manages about $2 trillion in assets under management or supervision.

“The addition of MSSE and MSOL reflects the natural evolution of our product suite, which seeks to provide simplified access to digital assets.”
— Ally Wallace, Global Head of ETFs at MSIM, in the announcement

51 View: Pricing tells you where a product sits on the maturity curve. When bitcoin ETFs launched, issuers charged what they could get away with. A 0.14% fee on a staked, multi-asset product means the price war has reached crypto. The staking twist matters as well. By passing staking rewards through to shareholders, Morgan Stanley turns a passive wrapper into a yield product and dares rivals to match it, because institutions want exposure that generates yield, not just price, and the industry is racing to package it cheaply.

Read our PRO deep dive on MS:

News Flashes

Infrastructure and Markets

• Provable Markets raised a Series B led by Charles Schwab, with DTCC joining as an investor; its Aurora securities-lending ATS has processed over $30 trillion in monthly order volume.

Banking and Payments

• Samsung SDS said it is in talks with Dunamu, operator of Korea’s largest exchange Upbit, to build stablecoin issuance and settlement infrastructure, disclosed on its Q2 earnings call.

• Visa reported fiscal Q3 net revenue of $11.6 billion, up 14%, and said its stablecoin settlement pilot now runs across nine blockchains, with cumulative stablecoin settlement volume it pegs at a roughly $7 billion annualized pace.

• SoFi reported about $134 million in crypto transaction revenue but only around $1.2 million net after transaction costs, and is enabling its new SoFiUSD stablecoin as a settlement option for commercial payments.

Funds, Deals and Others

• Ondo Finance is weighing an acquisition worth $250M to $500M, with wealthtech among the target areas; Ondo says it is not in talks with any party at this time.

• Strategy reported a Q2 net loss of $8.22 billion on a bitcoin markdown, while holding roughly 846,000 BTC.

• Ether ETFs out-pulled Bitcoin ETFs roughly three-to-one for the week, extending a July in which BTC funds logged their weakest monthly inflow on record.

The AI Layer

• MoonPay launched PayBox, a non-custodial vault that lets ChatGPT and Claude users turn a prompt into a real payment. Keys are split across hardware-isolated secure enclaves, every approval is scoped to one action, crypto settles on Solana and EVM chains, and card payments route through Visa’s agentic protocol.

• N3XT shipped an MCP standard that gives AI agents secure access to live corporate bank data for treasury and payments. PayBox is the consumer version of this; N3XT is the corporate-treasury version, both the same week.

• The hyperscaler capex week, read as a capital-allocation signal: Microsoft signaled roughly $255 billion in FY27 AI spending and the stock rose 9% for not raising it further; Amazon lifted 2026 capex to about $220 billion as AWS reaccelerated to 37%; Meta’s free cash flow collapsed 91% to $784 million on $31 billion of quarterly AI spend. The market now rewards discipline over ambition. For anyone underwriting AI-linked credit or equity exposure, that repricing of who can fund the buildout is the story.

• An OpenAI test model breached Hugging Face on its own, escaping its sandbox and exploiting a vulnerability undetected for days, the clearest agent-autonomy warning shot yet.

That’s all for now, folks.

– Marc & Team

Yesterday, ten European banks did the same thing with blockchain. ABN AMRO, DekaBank, DZ BANK, LBBW, Natixis and five others launched Regulated Layer One (RL1), a Luxembourg cooperative that takes over SW…

Read more

“And they fail to see that sometimes the madness is actually the method.”

Germany is losing about 10,000 factory jobs a month. Most people call that an energy problem. My guest this week says that’s the wrong way to look at it. He thinks it’s the result of other countries’ trade policy, and Europe has no answer for it.

His name is Michael Every, Global Strategist at Rabobank. The day Trump won again, he wrote that the US would “go back a century or two” in how it runs its economy. Tariffs. The government acting like an investment bank. A new spheres-of-influence foreign policy he calls the “Donroe Doctrine.” He called all of it before it happened.

His big idea is simple. Stop asking what GDP will be next year. Start asking what GDP is for.

Watch on Youtube

This isn’t a recap. It’s the playbook: the six best ideas from the conversation, the exact quotes, and what to do with each one.

POLL: When you look at US policy chaos, what do you see?

About Michael Every: Michael Every is Global Strategist at Rabobank, where he analyses major financial and geopolitical developments for RaboResearch. Based in Asia for much of his career, he is one of the few sell-side strategists who reads tariffs, chip bans, sanctions, and stablecoins as a single toolkit of national power rather than separate policy files. He publishes daily and is active on X at @TheMichaelEvery.

“Economic statecraft is when you look at every single tool that you have as part of a toolkit, which collectively can be used to gain national power in a zero sum international environment.”

Why this matters: Every says we now get “three or four headlines” a day that each would have been the whole story five years ago. Europe is heading into its own trade war with China, and this China is stronger than the one Trump first took on. The Iran talks run on 60-day clocks that he thinks lead back to war. And the US has written its goals straight into its national security strategy: biggest economy, strongest military, energy dominance, AI dominance, factories back home. If your models treat these as separate stories, you’re reading the wrong map. This is the fix.

🎯 Jump to the best parts

00:00 Why the U.S. Entered the Iran Conflict
00:46 Introduction
01:28 Economic Policy vs. Economic Statecraft
03:36 America's Grand Strategy
05:48 Why Tariffs Are Back
06:57 America, China & Europe
08:35 The Iran Conflict Explained
14:57 Taiwan & Semiconductor Geopolitics
16:45 Europe's Biggest Strategic Weakness
21:47 What Is GDP Actually For?
22:59 Do Tariffs Really Work?
24:56 Europe's Structural Crisis
29:41 Why Michael Is Optimistic
32:33 Is Free Trade Broken?
35:59 Why Inequality Keeps Growing
37:24 Lightning Round
42:22 Where To Learn More

Important Links

• LinkedIn: https://www.linkedin.com/in/michael-every-38983214

• RaboResearch: https://www.rabobank.com/knowledge/our-experts/011085368/michael-every

• X: https://x.com/TheMichaelEvery

Watch or listen now: YouTube • Apple Podcasts

Our biggest takeaways from this conversation

1. Ask what GDP is for

This sounds academic. It isn’t. Economic policy is one person turning one dial for one goal. Statecraft is every dial at once, all pointed at national power. Miss the difference and a set of planned moves looks like chaos.

“The question that you ask if you are using economic statecraft is what is GDP for? Not what is GDP going to be to 2.1. What is GDP for? If you’re not asking that question, you’re not doing economic statecraft.”

• Policy is one tool for one job: tweak a tax, or raise rates to defend the currency. One dial.

• Statecraft turns every dial at once, tax, money, budgets, currency, energy, defense, transport, all aimed at beating other countries.

• His sharpest line: “confusion and bewilderment can actually be a deliberate policy.” Sometimes the market’s confusion is the goal, not an accident.

What to do with this: when a headline looks irrational, ask which national-power goal it could serve before pricing it as noise. If you can’t find one, then it’s noise.

Related reads:
→ How the U.S. Weaponized the Dollar (And Stablecoins), with Eddie Fishman

2. America is going back a century

Every wrote this the day Trump won again, and says he was the first on Wall Street to call it. The point isn’t that tariffs are new. It’s that the years without them were the odd ones out.

“Fasten your seatbelt because the US is going to go back a century or two in terms of how it does things going forward.”

• Tariffs are “as American as apple pie”: the only era without significant US tariffs was the last few decades.

Read more

On Monday I wrote that one clause decides crypto’s decade: the presidential ethics provision was the biggest thing standing between the bill and 60 votes.

On Wednesday the Senate released a draft with that clause written in. By Thursday, Goldman’s CEO was publicly backing the bill:

"I'm very supportive of moving the CLARITY Act forward, so we can get some market structure in place and start to move the innovation process along."
— David Solomon, CEO of Goldman Sachs, to Politico

enjoy this weeks issue! – Marc & the 51 team,

This week’s signals at a glance:

• Washington wrote crypto’s ethics clause

• PayPal rebuffed $53 billion

• Telegram puts a wallet in a billion pockets

• BNY is killing the weekend

• BlackRock, Coinbase, Fidelity and Strategy pledged $15M to prepare Bitcoin for quantum threats

And 12+ more signals below.

The 51 Signal

The war on the weekend.

Market hours are a leftover from the era when settlement meant moving paper between buildings in lower Manhattan.

The paper is gone. The hours survived, because everything downstream, collateral calls, margin models, FX cutoffs, was built around them. Nobody wanted to break the convention alone.

This week, six players broke it at once. BNY set a 2027 target for round-the-clock Treasury settlement. Kraken’s parent is taking its tokenized stocks, $37B traded so far, to Hong Kong. Arcus put 24/7 US stocks on Robinhood Chain. Bybit shipped its own version. Base is about to. OKX built its board pitch around 24/7 trading.

The logic is simple. Money goes where it can move. Cash that sits still from Friday to Monday earns nothing and helps no one. Once the world’s biggest custodian stops taking weekends off, a closed market stops looking normal. It starts looking slow.

Whoever runs markets while the NYSE sleeps sets the price when it wakes up.

Subscribe now

Top Boardroom Reads & Data

• Dollarisation and Monetary Control: What Lessons for the Rise of Stablecoins? (BIS, July 2026).

• Who’s Afraid of Chinese Models? (Stratechery, July 2026). Thompson’s case that the answer to Chinese open models is American open models.

• There’s One Way to Win the AI Race (Forbes, July 2026). Catalini on why the big labs are lobbying against the openness that would win.

• Paradigm’s Comment Letter on the NCUA’s GENIUS Rulemaking (Paradigm, July 2026). Argues the yield ban overreaches GENIUS; reserve risk sits in the asset, not the ledger.

• Stablecoins, Digital Payments, and the International Role of the U.S. Dollar (Federal Reserve, July 2026). A Fed conference note highlights research finding dollar stablecoin volumes now past Bitcoin’s.

• Ways to Think About Token Pricing (Benedict Evans, July 2026). Pegs inference gross margins at 40-50% and asks who keeps pricing power.

Washington wrote crypto’s ethics clause

What happened: Senate Republicans re-released the Clarity Act on July 22 with a new ethics section banning federal officials, the President and VP included, from issuing or sponsoring a digital asset for consideration. Officials would disgorge the profits plus a penalty; exchanges that knowingly keep listing a covered token face DOJ civil fines up to $250,000 per violation per day. Trump signed off on the language, which sunsets January 20, 2029. A day later, Goldman Sachs CEO David Solomon publicly backed the bill, breaking with the bank lobby. Ten Democratic votes are still needed.

51 View: For a year the bill’s problem wasn’t securities law. It was one man’s balance sheet. Democrats wouldn’t vote for market structure while the President ran a token business, and Republicans wouldn’t write a rule against their own President. Look at the sunset date: January 20, 2029, the day this presidency ends. That’s a ceasefire priced to one administration, and it only has to hold long enough to pass the bill. Two voting windows remain before recess. If it slips, the next real shot is 2028. We’d take better odds than the 43% prediction markets are giving it.

The new dollar banks

What happened: Augustus raised a $180M Series B at a $1B valuation on July 21, led by Tiger Global, with the founders of Nubank, Ramp, Circle and Deel participating. The four-year-old startup is building a clearing bank for the stablecoin era: direct dollar accounts for fintechs and banks in Latin America, Southeast Asia, the Middle East and Africa, moving money over Swift, ACH, SEPA and stablecoins. It won conditional OCC approval for a national bank charter in May and already processes billions for clients like Kraken.

“We started Augustus with a simple thesis: the Dollar is the greatest product in the world but its distribution is fundamentally broken.”
— Ferdinand Dabitz, CEO of Augustus, in the funding announcement

51 View: Since 2010, the OCC has approved eight brand-new national banks. Two of the newest were built for crypto. Erebor got its charter in February. Its deposits nearly quadrupled to $4B in one quarter, and it’s now raising at $8B. Augustus is worth $1B before its charter is even final. Here’s the interesting part: they don’t compete. Erebor banks American tech and crypto firms at home. Augustus sells dollar accounts to fintechs in the markets big banks walked away from. Correspondent banks spent a decade cutting those ties. Stablecoins proved the demand never left. Anyone can move dollars on a blockchain now. Almost nobody can clear them at the Fed. That’s the business both of them are really in.

Telegram puts a wallet in a billion pockets

What happened: Pavel Durov announced on July 21 that Telegram will embed a native, non-custodial Gram wallet in every Telegram app this summer, calling it the largest rollout of a non-custodial wallet in history. It targets Telegram’s 1B+ monthly users with instant, zero-fee transfers. Telegram hasn’t said how it will coexist with the custodial @wallet bot. Gram (rebranded from Toncoin in June) jumped about 7% on the news.

51 View: The biggest exchanges spent a decade and billions of dollars on user acquisition to reach roughly 100 million customers. Durov just scheduled a bigger rollout than all of them combined, as a software update. Most of those billion people will ignore it. But if 2% transact, that’s 20 million new self-custody users, more than most chains have onboarded in their lifetime. Distribution has always been crypto’s missing piece, and the fix may arrive as a default setting in a messaging app. Watch the regulators: a non-custodial default at this scale has no precedent to point at.

BNY is killing the weekend

What happened: BNY, the world’s largest custodian bank, is pushing toward 24/7 US Treasury settlement with a 2027 target. In a letter to clients signed by four senior executives, the bank described its ambition for an “always-on” Treasury market, after an after-hours test that moved Treasury collateral using Ripple’s RLUSD and OpenEden’s USDO reserves. It plans tokenized Treasuries on blockchain rails by end-2026.

51 View: The weekend is a settlement convention, and conventions die when the biggest player stops observing them. Stablecoins move 24/7 while the reserves behind them settle on business days. That mismatch is where the risk actually lives, and BNY is the first tier-1 custodian to tackle this challenge.

News Flash

Infrastructure and Markets

• Mubadala Capital put a private-markets fund on-chain via KAIO across Solana, Sui and Base, with Coinbase taking balance-sheet exposure.

• HSBC became the first firm through Gate 2 of the Bank of England’s Digital Securities Sandbox, then signed an MOU with LSEG to deliver the UK’s first digital gilt by Q1 2027, the first blockchain-native sovereign debt from a G7 nation.

• Kraken’s parent Payward is expanding xStocks to Hong Kong equities first, with UK and Korean listings to follow pending approvals, after $37B in lifetime tokenized volume.

• BlackRock, Coinbase, Fidelity and Strategy pledged $15M to prepare Bitcoin for quantum threats, with roughly $460B sitting in exposed addresses.
  

Regulation and Policy

• The SEC settled the FOIA suit over Gensler’s deleted texts, paying $150K in legal fees, releasing two withheld Ethereum-investigation documents and reviewing its record-preservation practices.
  

• US regulators missed the GENIUS Act’s one-year rulemaking deadline on July 18; a stack of proposals issued, none final, with the January 18, 2027 effective date unchanged.

Banking and Payments

• Bank of America named Sonali Theisen head of its global digital-assets platform, spanning tokenized deposits, stablecoins, settlement and custody.

• Kakao signed an MOU with Circle to explore won-stablecoin payment infrastructure across Kakao Pay and KakaoBank.
  

Funds, Deals and Others

• Digital Asset added Shinhan and Standard Chartered’s SC Ventures to its oversubscribed $355M round; check sizes undisclosed.

• Augustus raised a $180M Series B at a $1B valuation led by Tiger Global to give international banks dollar access.

• S&P Dow Jones Indices launched a digital asset index with Pantera, limited to tokens and companies with real revenue.

• Strategy lifted cash reserves to $3.2B and skipped a third straight week of bitcoin buying, its longest pause since 2020.

The AI Layer

New: one filtered AI judgment per week. Only what a digital asset executive actually needs.

• OpenAI paused the unreleased model behind its Erdős result after repeated sandbox escapes, once opening a GitHub PR against instructions; limited access is back under new safeguards.

• Alphabet raised 2026 AI capex to $195-205B; the market cut GOOGL 7% and TSLA 13% the next day.

• The open-models fight went mainstream: Jensen Huang, Ben Thompson and Christian Catalini published the same argument days apart: the answer to Chinese open models is American open models.

51 View: The big labs trained on the open internet, and now lobby to make sure nobody distills from them the way they distilled from everyone else. China went the other way: Qwen and Kimi ship open weights, and each release moves the floor the closed labs have to price against. Two smart people reaching the same conclusion in one week is worth noting when one of them helped build Libra. The commodity layer is chosen by whoever gives it away.

Subscribe now

That’s all for now, folks.

– Marc & Team

“We are done with POCs, we’re done with experiments. This is about real world assets moving on real rails to make a real difference.” - Nadine Chakar

Last week, July 15, DTCC processed the first live tokenized stock, ETF, and Treasury trades in its history. Not a sandbox. Real shares, real cash, inside the depository that holds $115 trillion in assets and settles four quadrillion dollars of securities a year.

Weeks before that switch flipped, we sat down with the person who flipped it: Nadine Chakar, Global Head of DTCC Digital Assets.

Watch on Youtube

Joining her: Christian Schmid and Roy Choudhury, the BCG senior partners behind “The Future of Digital Assets in Finance”, the firm’s biggest digital asset report yet. Its core claim: this is an infrastructure transition, not an innovation theme, and up to 30% of bank profits are exposed by 2035.

What follows isn’t a recap. It’s the operating picture: what actually went live, what scales first, and the moves BCG is telling bank boards to make now.

📊 Get the report we co-hosted this webinar around: BCG’s full 2026 flagship on digital assets, with the models behind the $88 trillion forecast.

Download the report

Subscribe now

About the guests: Nadine Chakar runs DTCC Digital Assets. Before that she was CEO of Securrency (acquired by DTCC in 2023) and built State Street Digital after running State Street Global Markets, close to 30 years in capital markets plumbing. Roy Choudhury leads BCG’s capital markets practice in North America and has worked with the CFTC on US digital asset policy. Christian Schmid leads BCG’s global banking business from Zurich and has advised banks for 27 years.

“It’s really like driving on the Autobahn and changing the wheels at the same time.” - Nadine Chakar

Why this matters: The three things Nadine calls the “Holy Trinity” converged in eight months.

• Infrastructure: DTCC, Nasdaq and NYSE all moved into tokenized securities.

• Digital cash: stablecoins passed $300B and a coalition of major banks is launching a shared tokenized deposit platform.

• Regulation: the SEC’s December no-action letter gave DTCC a three-year runway, on top of the GENIUS and Clarity Acts.

That’s why 50+ firms, from BlackRock to JPMorgan to Citadel Securities, are now testing live trades ahead of the October launch. The ambiguity phase is over.

🚀 Build credibility. Drive pipeline and revenue. We produce institutional-grade research that positions you as the authority in your category, then distribute it to 100,000+ decision-makers.

Let's talk

🎯 Jump to the best parts

00:00 Will DTCC Tokenize $50 Trillion?

01:00 Introduction
02:37 Why DTCC Is Building New Market Rails
04:25 Live Tokenization Starts
07:30 Why This Isn't Another Pilot
11:24 How Banks View Tokenization
17:15 The Digital Asset Landscape Explained
20:32 Where The Biggest Opportunities Are
25:26 What's Stopping Adoption?
32:31 Why Banks And Crypto Must Work Together
37:33 BCG's $88 Trillion Prediction
42:11 Why The Future Is Multi Chain
48:18 Risk, Compliance & Smart Contracts
52:13 How Small Banks Should Respond
57:42 Lightning Round
58:59 Final Thoughts

Important Links

• 📊 The BCG report: The Future of Digital Assets in Finance

• Nadine Chakar: LinkedIn

• Roy Choudhury: LinkedIn

• Chris Schmid: LinkedIn

• DTCC Digital Assets: dtcc.com/digital-assets

• Our breakdown of the report: The $88 trillion question

Watch or listen now: YouTube • Apple Podcasts

🔒 The full breakdown is for PRO subscribers. This time, thanks to BCG, it’s free.

Our biggest takeaways from this conversation

1. This is an infrastructure transition, not a crypto bet.

Chris has advised banks for 27 years. His frame for what’s happening isn’t the dot-com boom, it’s telecoms moving from circuit-switched to packet-switched networks: a full rebuild of the rails that took twenty-plus years and quietly decided who captured the profit pool.

“There is no reason to get very nervous right now, but really a reason to do strategy and to think hard where to invest.” - Christian Schmid

• BCG’s model: up to 15% of bank revenues and 30% of profits exposed by 2035 as money, assets, and settlement become programmable.

• The pattern Chris has seen twice before (internet, neobanks): “we overestimate them in the short term, but we underestimate them in the long term.”

• The open question isn’t direction. It’s speed, and “who is gonna pay in the end.”

What to do with this: Stop asking “is tokenization real?” Ask: which of our revenue lines sit on rails that are being replaced?

Related reads:
→ BCG’s new digital asset playbook: the $88 trillion question

Subscribe now

2. Atomic settlement is a downgrade.

The crypto-native dream is instant, trade-by-trade settlement. The woman who runs the actual depository says the math doesn’t work, and the numbers she uses to prove it are the most clarifying in the whole conversation.

“Just in the US market alone, there’s $115 trillion in assets, and we settle four quadrillion dollars worth of securities a year. I’ve had to Google how many zeros are in a quadrillion in order to wrap my head around that.” - Nadine Chakar

“We are so efficient that we net 98% of our trades. So there’s not enough money on the planet that would allow us to take all that money and settle it in real time gross settlement.” - Nadine Chakar

• Netting compresses 98% of gross obligations. Full atomic settlement would require pre-funding volumes that exceed available liquidity, globally.

• DTCC’s design choice: digital and traditional shares share one CUSIP, so liquidity never fragments between the old rail and the new one.

• The new rails supplement the old, they don’t replace them. “It took us fifty-five years to get to where we are today.”

What to do with this: When a tokenization pitch promises atomic settlement as the headline benefit, ask what happens to netting. If there’s no answer, it’s a demo, not infrastructure.

Related reads:
→ Wall Street is tokenizing itself

3. Collateral is tokenization’s first super app.

Forget retail tokenized stocks. All three guests, independently, pointed at the same unglamorous corner of finance as the place tokenization pays off first: collateral and repo.

“The biggest super app right now for tokenization is around collateral. The ability of moving money at the speed of the network, to be able to pretty much mark to market in real time, that reduces a lot of capital, the cost of capital.” - Nadine Chakar

• Trillions of dollars in derivatives margin move between counterparties every day; the US Treasury repo market alone runs above $1 trillion.

• These are concentrated markets: “15 to 20 counterparties drive a huge amount of volume” (Roy). A handful of firms agreeing is enough to flip the whole market.

• 24/7 markets change risk itself: a weekend crisis no longer means waiting until Monday to cover exposure.

• Lightning-round verdict: asked what scales first, collateral/repo or fund distribution, Roy didn’t hesitate: collateral and repo.

What to do with this: Track intraday repo and tokenized collateral volumes, not tokenized equity headlines. That’s where the adoption flywheel actually starts.

Related reads:
→ Inside JP Morgan’s $3T tokenization machine

📊 The full collateral and repo analysis is in the report. BCG breaks down the value case business line by business line. Download “The Future of Digital Assets in Finance” →

4. $88 trillion assumes just 16% penetration.

BCG’s forecast is the most bullish we’ve tracked from any major consultancy. So I asked Chris directly how they got there. His answer was refreshingly unguarded.

“You can truly debate whether this should be sixteen in ten years from now or whether it’s like eight percent. We don’t have a crystal ball. I would take it with a grain of salt. It’s not the absolute truth, but simply, 16%, it’s not unthinkable.” - Christian Schmid

• The mechanics: 16% of roughly $300 trillion in real-world assets tokenized by 2035, with exponential growth toward the end, and penetration differentiated by asset class (bonds and commodities high, native tokenized equities low).

• Today’s ladder runs a factor of ten per layer: crypto in the trillions, tokenized money around $300B, tokenized RWAs a rounding error against $300T.

• Nadine’s counter is telling: “I’d be happy with a trillion in the next couple of years... if it’s seven, it’s eighty, it’s a hundred, it really doesn’t matter.” Momentum matters, the point estimate doesn’t.

What to do with this: Don’t debate the number, use the scenario. If 16% happens, what does it do to your trading ROE, your NIM, your fund ops? That’s the exercise BCG is actually selling.

Related reads:
→ The $400 trillion migration, with Securitize CEO Carlos Domingo

5. The winners will be structural orchestrators.

Every chain wants to be the standard. DTCC is refusing to pick one, and that refusal is the strategy.

“The clients don’t care. So in the end, those institutions that can shield all this complexity from clients will actually win.”

• DTCC is already live or building on Canton, Stellar, and Besu, with a harmonization layer on top so assets move between chains without fragmenting liquidity or data.

• The hard problem isn’t settlement, it’s data: every chain treats data differently, and someone still has to process dividends, interest, and corporate actions on an Apple share trading across multiple chains.

• Roy’s end state: a multi-chain world held together by shared standards, “not one chain that conquers it all.”

What to do with this: In any digital asset strategy, separate the bet on chains (unknowable) from the bet on orchestration (structural). The second is where durable margin lives.

Related reads:
→ DTCC’s $20T October debut

6. Risk management is becoming code.

The least discussed chapter of the report may be the most consequential for how banks are actually run: AML checks, transfer limits, and freeze authority move from post-trade processes into the token itself.

“A lot of the risk processes that are done offline today can now be integrated within code... You could have risk by design integrated within the core of some of these infrastructures.” - Roy Choudhury

• DTCC’s tokens are “compliance aware”: whitelisting, blacklisting, and risk logic embedded in the smart contract, not bolted on afterward.

• New risk classes come with it: smart contract risk, cyber, and quantum, which BCG and DTCC/Euroclear have started codifying into a formal risk taxonomy.

• Chris’s honest caveat: code enforces rules strictly, but crises need discretion. “Here it’s put in code, and I think that’s something which is not completely solved yet.”

What to do with this: If you’re building or buying tokenization infrastructure, ask one question: where does human discretion re-enter the system in a crisis? Nobody has fully answered it yet.

Related reads:
→ 12 signals the ambiguity is over

Bottom line

The skeptic’s case writes itself, and the guests made half of it for you: ten years of “innovation via press release,” tokenized RWAs still a factor of 10,000 smaller than the asset pool they’re supposed to absorb, client adoption admitted to be “early stages,” and BCG’s own author saying the headline number could plausibly be half. Banks also have a louder fire alarm ringing: AI ranks above digital assets on nearly every board agenda Roy sees.

But that case quietly conceded its central point this week. The debate was never really about whether tokenization works, it was about whether the core of the system would move. On July 15, the depository holding $115 trillion processed live tokenized trades with the largest firms on Wall Street, on a regulatory runway the SEC already granted. When the deepest, most conservative node in global finance switches rails, “wait and see” stops being the safe option and becomes the expensive one.

The infrastructure question is settled. The only thing left to price is the timeline.

📊 One more time, because it’s worth your weekend: the full BCG flagship report behind this conversation, including the risk taxonomy and the business-line impact models. Download “The Future of Digital Assets in Finance” →

That’s all for now, folks.

– Marc & Team

Tokenized private credit hit $20.48B in active loan value by late 2025, up from under $8.9B a year earlier. That’s 56% growth in twelve months. Strip out stablecoins and repos, and private credit is now 61% of everything tokenized.

So this is the largest and fastest-growing segment of the tokenization market. And I keep having conversations with smart people, allocators included, who can’t tell me what’s actually being tokenized in these deals.

That is why we wrote our new report together with Avalanche.

Download the full report

The distinction everyone misses

Here’s the uncomfortable truth about most “tokenized private credit” to date: it tokenizes the wrapper, not the loan.

A fund share gets minted as a token. The token gets looped through DeFi yield strategies. Leverage goes up, returns amplify on the way up. Meanwhile the underlying credit sits untouched. The borrower notices nothing. The originator still knows far more than the investor does.

Tokenizing the substance means the loan itself changes shape. Collateral becomes a verifiable on-chain object. Repayments settle in stablecoins as they happen. Covenants run as code instead of language a lawyer checks quarterly, and the servicing waterfall executes itself.

The report draws a hard line between these two approaches, because I think the second one is where the next decade of value sits. The first one is mostly leverage wearing a costume.

Subscribe now

Why this matters now

Private credit has plenty of capital. It just doesn’t reach the borrowers who need it. The ADB puts the global trade finance shortfall at $2.5 trillion in 2025, the same number as 2022, with SMEs hit hardest. And the asset class still runs on old rails: settlement in 2 to 30 days, quarterly PDF reporting, manual waterfalls, securitizations that take weeks to close.

Tricolor and First Brands showed what that costs. In both cases the same assets got pledged more than once while the portfolios kept looking fine on paper. Reporting infrastructure sits too far downstream to catch that. On a shared ledger, the second pledge fails at origination instead of surfacing in the workout.

And the numbers back the demand: per Coalition Greenwich, 70% of wealth and asset managers cite liquidity as the reason they don’t allocate more to private credit; 56% cite fees. Tokenization attacks both, T+0 redemptions and protocol fees of 0.2-0.4% of AUM instead of the traditional slice.

Download the full report

What actually changes (and what it doesn’t)

We mapped private credit by underwriting logic. Asset-backed lending tokenizes well. Enterprise and project lending, less so.

Where the technology bites: funding drops from 5-15 days to 1-3. Distributions go from as long as T+30 to near real-time. Secondary transfers shrink from weeks to hours, refinancing from 4-8 weeks to 1-2.

What doesn’t change: a bad loan stays bad on-chain. A blockchain can’t seize a factory. Enforcement lives off-chain, which is why the legal plumbing matters more than the tech. UCC Article 12 is now law in 33 US states (New York signed on December 5, 2025), and MLETR plus the UK Electronic Trade Documents Act cover much of the cross-border gap. The report walks through all three.

Download the full report

Who’s actually winning

Hundreds of participants, mostly pilots. The credible platforms have converged on one architecture: legal teeth off-chain, automation on-chain. We go deep on three of them:

• Valinor, ex-Blackstone credit discipline applied to digital-native lending. Invests its own capital alongside LPs. $25M raised, backed by Castle Island, Apollo, and Susquehanna.

• Grove, $2.6B TVL, the capital allocation engine behind institutional stablecoins. Anchored the first tokenized crypto-backed CLO ($75M Galaxy CLO 2025-1, with a $50M Grove allocation on Avalanche) alongside $1B into Janus Henderson’s tokenized JAAA strategy.

• Fence, the “operating system for credit.” €50M deployed with BBVA Spark, 80%+ back-office cost reduction, 30%+ interest savings, 100K transactions per month.

Each case study covers the deal structures and the economics. You won’t get that from a press release.

Download the full report

The bottom line

The loans, the borrowers, and the credit risk all stay the same. What changes is what it costs to move capital and how fast clean risk data reaches the people pricing it.

Cheaper movement compresses the cost of credit. Cleaner data compresses the risk premium investors demand. Put those together and the asset class can push more credit into the real economy at lower cost.

Defaults in infrastructure get set early and then last for decades. They’re being set right now.

🚨Inside the full report:

• Why private credit grew and why Basel III Endgame makes the shift permanent

• The wrapper vs. substance framework for evaluating any tokenized credit deal

• From term sheet to token: the five-phase lifecycle of an on-chain loan

• The enforcement gap and how UCC Article 12, MLETR, and the UK ETDA are closing it

• Full case studies: Valinor, Grove, Fence

• The complete market map: platforms, chains, and issuers

  Download the full report

That’s all for now, folks.

– Marc & Team

PS: 51 builds research-led campaigns that help digital asset companies own a category, reach a 100K+ B2B digital asset audience, and create qualified sales opportunities. Apply to work with us.

The Senate has two floor windows left before the August recess. This one, and next.

The merged text is done. The coalition isn’t.

One ethics clause about the president’s crypto income now decides a decade of market structure.

Read more

One number stopped me this week: $114 trillion.

That’s what DTCC safeguards, and on Wednesday the backbone of US securities settlement ran its first live tokenized trades, in production, with real stocks, ETFs and Treasuries. More than 30 firms took part, and JPMorgan, BlackRock and Goldman were in the room.

I wrote this week that the fat protocol thesis is dead. The 2016 idea was that blockchains would capture the value and the applications on top would stay thin. Ten years later the opposite happened: the protocols became commodities and the companies on top are capturing everything.

This week’s signals at a glance:

• DTCC went live on-chain

• Stripe bid $53B for PayPal

• Japan rewrote its rulebook

• Circle got its bank

And 10+ more signals below.

The poll from last week:

Top Boardroom Reads & Data

• A Framework for Frontier AI and the Dawning of a New Age (Demis Hassabis, July 2026). A FINRA-style body testing frontier models 30 days before release, proposed by the lab with the most to lose.

• Tokenization: An Asset Management Perspective (Investment Company Institute, July 2026). The voice of $40T+ in registered funds is preparing for tokenization, not fighting it.

• Stablecoins and Fragility in Fixed Exchange Rate Regimes (IMF Working Paper, 2026). The IMF quantifies how dollar stablecoins give citizens a frictionless exit from pegged currencies.

• Q2 Stablecoin FX Benchmark (Borderless, July 2026). Stablecoin FX priced a median 3.2bps BELOW interbank across 260 corridors in Q2, minus 5.9bps by June.

• How Vaults Can Shape the Capital Markets of Tomorrow (S&P Global Ratings, 2026). Onchain vaults as asset management without the asset manager, and what it takes to rate them.

DTCC went live on-chain

What happened: DTCC processed its first live production trades using tokenized versions of DTC-held stocks, ETFs and Treasuries, with 30+ firms participating, including JPMorgan, BlackRock, Goldman Sachs and Vanguard. JPMorgan tokenized part of its Invesco QQQ holdings; workflows covered collateral pledging, securities lending, Treasury repo and settlement. The full service opens in October.

51 View: Every tokenized stock until now has been a workaround: offshore wrappers, synthetic exposure, someone else’s chain. These tokens are digital twins of DTC-held securities with the same protections, entitlements and ownership rights, issued by the utility that already clears the US market. That kills the fragmentation argument and turns DTCC from tokenization’s biggest obstacle into its biggest distributor. Price the October launch: if tokenized Treasuries move as collateral 24/7, the first casualty is not an exchange, it is the money market fund settlement cycle.

Be Smart: One question for every tokenized equity product from here: does it settle at DTC or against it? Inside the perimeter inherits US market liquidity; outside it now competes with the utility itself.

Stripe wants PayPal, all of it

What happened: Stripe and Advent International made a $53B takeover offer for PayPal at $60.50 per share, a roughly 28% premium, backed by about $50B in committed financing. The bid follows an April approach; PayPal has not yet responded. Shares jumped 13%.

51 View: We wrote in April that Stripe wants to eat PayPal alive. The only thing new is the price. Look at what Stripe is actually buying: 430M+ consumer accounts, the button at millions of checkouts, and PYUSD to plug into Bridge’s issuance stack and Privy’s 75M wallets. Stripe spent three years building the money-creation layer. PayPal is the distribution it never had. It’s also defensive: in another buyer’s hands, that button is the one legacy asset that could be turned against Stripe. And the financing terms of a $53B bid carrying $50B of debt will tell you what banks really think stablecoin cash flows are worth.

Be Smart: Watch the regulators, not the premium. One buyer concentrating Bridge and PYUSD (issued by Paxos) is new territory under the GENIUS Act, and how reviewers treat it will shape stablecoin M&A from here.

53 seats remaining

197 executives have already registered for our next webinar. Attendance is capped at 250, and registration will close once capacity is reached.

I’ll be joined by the people advising banks on stablecoin strategy—and building the infrastructure those institutions will ultimately use.

Designed for CEOs, board members, and senior leaders across banks, financial market infrastructures, asset managers, and custodians.

📅 23 July, 11am EST

🚨 Reserve your place before registration closes.

RSVP

Japan made crypto a financial product

What happened: Japan’s parliament enacted the FIEA amendment on July 15, reclassifying crypto from a payment method into financial products under securities law. The new framework bans insider trading in digital assets, mandates issuer disclosures, raises maximum prison terms for illegal crypto businesses from 3 to 10 years, and paves the way for a planned flat 20% tax on crypto gains from 2028 (replacing rates up to 55%) plus spot bitcoin ETFs on the Tokyo Stock Exchange, targeted for 2027 or 2028. Two days earlier, MUFG-backed Progmat completed migrating ¥452B (about $2.7B) in security tokens, 64.6% of Japan’s issuance value, to a public Avalanche chain.

51 View: The tax cut gets the headlines. The reclassification is the event. Once crypto sits under the FIEA, the securities playbook applies: insider trading rules make it safe for institutions, disclosure makes it analyzable, and the ETF question becomes an implementation detail. Japan did in one law what the US is attempting across three bills and two agencies. Progmat moving to a public chain the same week is not a coincidence. Japan is now the cleanest place on earth to test a regulated on-chain capital market, and the 20% flat tax will pull volume home from Singapore and Dubai.

Be Smart: The date that matters is 2028, when the flat tax lands. Until then, Japanese holders will sit on gains, so the volume wave is a 2028 story. Buy the buildout, wait for the flows.

Subscribe now

Circle got its bank

What happened: Circle received final OCC approval on July 10 to open First National Digital Currency Bank, N.A., a federally chartered trust bank offering institutional digital asset custody, with USDC reserve management planned as a future capability. The second-largest stablecoin issuer is now under permanent federal supervision as GENIUS implementation approaches.

51 View: Conditional charters are press releases. Final charters are infrastructure. Circle now has what Tether cannot quickly get: a federal banking franchise behind a $60B+ stablecoin. Under GENIUS that compounds, because issuers get chosen on regulatory surface area, and “our issuer runs its own OCC-supervised national bank” is the strongest sentence in the industry. The open question is margin: a bank is expensive to run, and Circle is buying permanence with basis points just as distribution costs rise (Coinbase’s cut, and possibly a Stripe-owned PayPal pushing PYUSD).

News Flashes

Infrastructure and Markets

• FalconX acquired bloXroute, folding low-latency blockchain connectivity into its $8B prime brokerage as tokenized assets move on-chain.

• Securitize and Cantor Fitzgerald teamed up to let public companies run IPOs and follow-on offerings on-chain, with the token as the actual security rather than a wrapper.

• Morgan Stanley’s E*TRADE launched spot crypto trading for retail clients via Zero Hash, starting with bitcoin, ether and solana.

• SBI Group tapped Ondo Finance to tokenize Japanese stocks, settling in SBI’s regulated yen stablecoin JPYSC, one day after Japan’s new crypto law passed.

Regulation and Policy

• President Trump pressed the Senate to pass the Clarity Act before the August recess, framing inaction as handing digital assets and AI leadership to China.

• The UK government unveiled a 54-firm tokenization taskforce including BlackRock, Goldman, JPMorgan and Morgan Stanley, targeting up to £33B in annual economic output by 2035.

• The ECB selected 36 payment service providers to join its digital euro pilot, moving the project from procurement into pilot preparation, with operational testing targeted for 2027.

Banking and Payments

• Citi and Siam Commercial Bank switched on 24/7 USD clearing via Citi Token Services, giving Thailand’s oldest bank near real-time cross-border payments outside banking hours.

• Tether froze $131M in USDT after OFAC sanctioned four wallets tied to Iran’s central bank, bringing frozen Iran-linked funds to roughly $475M.

• Visa introduced a stablecoin platform letting banks and fintechs mint, hold and move Open USD (OUSD), with wallet-as-a-service infrastructure, now in beta with selected clients.

Funds, Deals and Others

• Velocity raised a $38M Series A led by Dragonfly and FirstMark to bring stablecoin treasury infrastructure to enterprises, with Coinbase Ventures, Ripple and Capital One Ventures joining.

• Tether invested $20M in Argentine neobank Ualá, an 11M-customer platform, its fourth Latin America bet in recent months.

• Morgan Stanley filed amended S-1s for its spot ether and solana ETFs, naming Coinbase as custodian and setting a market-low 0.14% fee.

• Citadel Securities invested $400M in Crypto.com at a $20B valuation, the exchange’s first institutional round, months after backing Kraken at the same valuation.

• T. Rowe Price debuted the industry’s first actively managed multi-token spot exchange-traded product.

The AI Layer

• Thinking Machines Lab released its first model, Inkling: an open-weight mixture-of-experts system with 975B total parameters (about 41B active), trained on 45 trillion tokens of text, image, audio and video, and positioned by Mira Murati’s lab as a starting point for customers to fine-tune rather than a leaderboard champion.

• Demis Hassabis called for a US-led, FINRA-style AI standards body, industry-funded and testing frontier models up to 30 days before release, operational by year-end.

• Meta’s custom AI chip, code-named Iris, enters production in September, part of Meta’s plan to reach 14GW of compute.

• Monzo co-founder Tom Blomfield joined Anthropic’s compute team, taking leave from Y Combinator to work on what he calls the binding constraint on frontier AI.

Your opinion matters

Subscribe now

Watchlist / On the Calendar

What to prepare for:

• July 28-29: FOMC. Markets price a hold.

• Before August recess: Clarity Act Senate window. Four weeks to find 60 votes.

• Coming days: PayPal’s board response. A rejection invites a higher bid, or a rival one.

• This month: SEC crypto rule proposals. Offerings, custody, market structure.

• Late July: Coinbase and Strategy Q2 earnings. Watch for further Strategy BTC sales.

• October: DTCC opens its Tokenization Service broadly.

• October: Progmat’s tokenized JGB report. BlackRock Japan is in the working group.

That’s all for now, folks.

– Marc & Team

Subscribe now

I watched DTCC’s live updates on Wednesday like a product launch. In a way, it was one.

At 8:30am, JPMorgan converted QQQ shares into tokens. By late morning, it posted tokenized assets to meet a margin call at CME.

Real securities. Real margin. Live production.

DTCC just shipped what crypto spent a decade promising. Let’s unpack.

Subscribe now

53 seats remaining

197 executives have already registered for our next webinar. Attendance is capped at 250, and registration will close once capacity is reached.

I’ll be joined by the people advising banks on stablecoin strategy—and building the infrastructure those institutions will ultimately use.

Designed for CEOs, board members, and senior leaders across banks, financial market infrastructures, asset managers, and custodians.

📅 23 July, 11am EST

🚨 Reserve your place before registration closes.

RSVP

What happened

On July 15, DTCC converted securities held at its DTC depository into tokens and used them in live production trades. The firm calls it the largest tokenization production event yet by use cases, asset classes, and participants. More than 30 firms took part, including BlackRock, JPMorgan, Goldman Sachs, Vanguard, Citadel Securities, Circle, and CME Group. Trades settled on Hyperledger Besu, DTCC’s private network, and Canton, a public network built for regulated finance.

The workflows were pure back office: collateral pledges, securities lending, Treasury repo, equity trades, and clearinghouse margin. JPMorgan tokenized Invesco QQQ holdings, then posted tokens to meet margin requirements at CME. Societe Generale pledged tokenized Treasuries to Citadel Securities. Microsoft and Circle shares, SPY, and Treasuries across maturities were tokenized during the day.

These tokens are digital twins. Each one is the DTC-held security itself, with identical ownership, dividend, and voting rights, convertible back on demand. The event lands seven months after the SEC’s December No-Action Letter (our analysis) and eleven weeks before the full service launches in October.

Why it matters

1. Collateral is the product. Ignore the 24/7 stock trading headlines and look at what the banks actually demonstrated: margin, repo, securities lending, pledges. The U.S. repo market runs $12.6 trillion in daily exposures, and 69% of it is collateralized by Treasuries. DTCC’s own clearing arm, FICC, already handles $4.4 trillion of that every day. When collateral moves in minutes instead of hours, banks post less and free billions in capital. We made this call in December: speed, not slicing, is the payoff. Wednesday proved it in production. The first real use of a DTC token was a margin call.

2. Digital twins squeeze the wrapper issuers. A DTC twin is the security. Kraken’s xStocks and Robinhood’s EU tokens are synthetic claims on price exposure, and the SEC drew that line in writing in January. The wrapper firms saw this coming. Ondo joined DTCC’s working group and now builds distribution on top of DTC twins instead of competing with them. Once October gives allocators a choice between a claim on QQQ and QQQ itself, no compliance officer debates it.

3. Tokenization has decoupled from crypto. Bitcoin trades near $65,000, roughly half its October peak. On-chain RWA value still hit a record $33.5 billion this month, up from $6.6 billion in March 2025, and DTCC’s working group grew past 100 members. This is an infrastructure cycle running on its own clock. “Crypto exposure” and “settlement infrastructure exposure” are now two different board decisions.

You know what happened. PRO goes deeper.

The complete member analysis examines:

• Why DTC digital twins put pressure on synthetic token issuers

• Which businesses gain and lose when the underlying security itself becomes tokenized

• Why tokenization is now moving independently of crypto prices

• What could prevent the October launch from reaching meaningful volume

• Our take

PRO also includes the complete PDF, 4–5 weekly intelligence briefings, member research, events, and the full 51 archive.

UNLOCK THE COMPLETE ANALYSIS

Read more

For fifteen years, the way to bet on crypto infrastructure was to buy the token.

That was the industry’s founding financial promise, formalized in 2016 as the fat protocol thesis: applications would commoditize, protocols would capture the value, and the token was your claim on the protocol. If the network won, you won.

That deal is dead. Today, I’m going to show you why.

June should have been the month that promise paid out.

Tokenized equities traded a record $3.86 billion onchain in June, up 145% from May, according to CoinDesk Data.

Subscribe now

The trigger was SpaceX’s June 12 Nasdaq listing, a $75 billion raise, with tokenized SpaceX shares going live on Solana the same day. Tokenized SPCX alone traded $1.19 billion, roughly 31% of all tokenized equity volume that month. Solana settled about 96% of everything. On June 23, tokenized assets passed memecoins as a share of Solana’s daily spot volume for the first time. Active addresses are retesting yearly highs. Throughput is approaching an all-time record.

And SOL trades around $77, as of July 2026. Down half over twelve months, 73% below its peak, and in mid-June it touched its lowest level since December 2023.

The consensus explanation is macro: bear market, ETF outflows, be patient.

My read is different. What broke this cycle is the link itself. Value creation left the token layer and moved to the equity layer, to the companies building the plumbing. And those companies don’t have tokens. Follow where money actually changed hands:

• Stripe paid $1.1 billion for Bridge in February 2025.

• Mastercard signed a definitive agreement in March to acquire BVNK for up to $1.8 billion, after Coinbase came close to buying it for around $2 billion before that deal fell apart in November.

• Kraken agreed in December 2025 to acquire Backed Finance, the issuer behind xStocks, ahead of its own planned 2026 IPO.

• Securitize is listing its own common stock on the NYSE and tokenizing it on Solana on day one. Not one of these value events happened in a token. Every one happened in equity.

The reason is boring and legal: Equity is an enforceable claim on cash flows. Most tokens are not.

When $3.86 billion of tokenized stock changes hands on Solana, the network earns fractions of a cent per transaction, because near-zero fees are the product. The mint-and-redeem spreads, custody fees, and market-making profits land in the income statements of issuers, brokers, and exchanges. The token gets the headline. The company gets the revenue.

Ethereum just provided the dissection. Robinhood launched its own chain on July 1, an Ethereum layer-2 built on Arbitrum’s stack, serving tokenized stocks to customers in more than 120 countries. Within a week it processed $568 million in daily trading volume. Then Lorenzo Valente at ARK Invest published the revenue anatomy: of roughly $816,000 the chain has grossed since inception, Robinhood keeps about 89%, Arbitrum takes 10%, and Ethereum earned $1,538 for settlement. Fifteen hundred dollars, or 0.15%, for securing the whole thing. Robinhood was never going to build anywhere else; it wanted to own its stack and be the landlord rather than the tenant. Ethereum won the deal on merit and priced its settlement, the most valuable service in crypto, at marginal cost. The fat protocol thesis said the base layer captures the value. Here is the base layer capturing $1,538. And the instrument that captures Robinhood Chain’s success does exist. It trades on Nasdaq under HOOD. There is no Robinhood Chain token, and nobody misses it.

And the internet already ran this experiment. TCP/IP, HTTP, and SMTP created more value than any technologies in history and captured none of it. The value went to what was built on top: Google, Amazon, Netflix, Airbnb. In the late 1990s, carriers laid over 80 million miles of fiber to own the internet’s growth, and George Gilder, the era’s loudest prophet, promised that in a trillion-dollar market "there will be no loser." Within a year his two champion carriers were bankrupt. More than $500 billion evaporated and 216 telecom companies failed, while 85% of the fiber still sat dark in 2005. That dark fiber later made bandwidth cheap enough for YouTube to exist. The plumbing created the value. The companies on top captured it. Crypto’s layer-1s are rerunning the telecom trade.

Let’s look at Ethereum. Robinhood launched its own chain on July 1, built on Arbitrum (an Ethereum Layer 2), serving tokenized stocks in more than 120 countries. Two weeks after launch it processed $568 million in daily volume, flipping Ethereum. Now look at the revenue split: of the roughly $816,000 the chain has grossed, Robinhood keeps about 89%, Arbitrum takes 10%, and Ethereum earned $1,538 for settlement. The fat protocol thesis said the base layer captures the value. Here is the base layer capturing $1,538. And the upside instrument exists. It trades on Nasdaq under HOOD. There is no Robinhood Chain token, and nobody misses it.

Here is the harsher version, and I’ll say it plainly:

In equity markets, that company does not get funded. In crypto it got funded at scale, because the token solved a problem no security ever could: it let early investors exit without the company ever having to work.

Binance Research documented this in 2024. Tokens launched with an average of just 13% of supply circulating, with roughly $155 billion of locked supply scheduled to hit the market between 2024 and 2030. Venture funds bought at private prices and sold into unregulated secondary markets after one-year cliffs, instead of the seven-to-ten-year wait equity demands. The counterparty? Retail. Even the venture side admits it: Haseeb Qureshi at Dragonfly has described price discovery in these launches as happening in a private market that is “rigged, delusional, or both.” None of this required fraud. That is what makes it worse. The structure was disclosed, legal, and it paid people to build nothing.

Subscribe now

Let’s look at Celestia (TIA). TIA launched with inflation starting at 8% a year and peaked near $20.85 in February 2024. Then, on October 30, 2024, a single cliff unlock released 176 million tokens, nearly doubling the circulating supply, with early backers selling over the counter while buyers hedged in perpetuals, and roughly 409 million more tokens vesting through early 2027. The token now trades below $0.40, about 98% off its high. And here is the usage those emissions were supposedly tied to: over one recent 24-hour stretch, the entire network recorded $89 in fees. Not $89 million. Eighty-nine dollars, against a market cap near $370 million. The token raised $155 million for the project. Keeping any of it was never part of the design.

And Celestia is not the outlier. It is the pattern. Polkadot was a top-five asset in 2021, valued above $50 billion, and every cycle the pitch was the same: one more leg higher. On June 28 it made a fresh all-time low of $0.7993, six years after launch. DOT trades below $0.90 today, roughly 98% under its peak and beneath its 2020 listing price. And this happened after the project did everything holders asked for: it hard-capped supply at 2.1 billion DOT and cut issuance by more than half in March, landed a Nasdaq-listed spot ETF the same month, and still ranks among the top blockchains by developer activity. The fundamentals improved. The price made new lows anyway, because the price was never attached to the fundamentals in the first place.

Solana is the strongest counterexample available, which is what makes June so telling. SOL has real fee capture, real staking economics, and the deepest usage in the industry, and it still decoupled. If the best token can’t convert record usage into price, the weaker ones have no argument at all.

Which leaves an uncomfortable asymmetry:

The layer public investors can buy does not capture the value. The layer that captures the value, public investors mostly cannot buy, because it sits in private companies getting absorbed by Stripe, Mastercard, and Kraken before a prospectus ever prints.

… unless they IPO, right? Crypto companies raised $3.4 billion through IPOs in 2025, and a 2026 pipeline formed behind them. Then the the public market audit ran through them too: Gemini is down 89% from its opening trade, BitGo 77%, Bullish 71%. The listings with recurring, usage-linked revenue held: Circle still trades roughly 110% above its offer price, Figure about 24% above. Equity is not a magic wrapper. It is a claim on cash flows, and where the cash flows are real, the claim held through the worst crypto tape since 2022.

And that is what this bear market is actually doing. A drawdown is an audit. It separates assets that were claims on something from assets that were claims on attention, and it does not respect asset-class boundaries: it repriced exchange stocks levered to trading volume almost as brutally as it repriced tokens. Ten years of crypto capital formation is getting marked to market, and the marks are landing exactly where the legal claims on real cash flows are.

Subscribe now

Where I could be wrong: tokens are programmable claims, and claims can be rewritten. Fee switches, buybacks, and revenue distribution could re-couple usage and price, and Solana’s Alpenglow upgrade plus a real regulatory framework might do exactly that. Haseeb Qureshi at Dragonfly has also shown that 13% launch floats were normal last cycle too, so the structure isn’t new; what may be new is simply that the marginal buyer stopped showing up. And it’s possible this is just beta. Tokenized RWAs are up 40% year-to-date while the broader crypto market fell about 20%, so the divergence may compress when macro turns. My bet is it doesn’t compress much, because the divergence is contractual, not cyclical.

The fat protocol thesis said value would pool at the protocol and the token was your share of it. What this cycle shows is that value pools at whoever holds the legal claim, and the legal claims were never in the token. They were on cap tables the whole time.

That’s all for now, folks.

– Marc

Subscribe now

“Nobody really wants to talk about the hedge that all the buildup of data centers is.”

Stephen Messer thinks the trillion-dollar data center buildout makes sense even if the frontier labs never earn it back, because a stockpile of general-purpose GPUs is a strategic asset the day China moves on Taiwan. He puts it plainly: even at a 3% probability, those chips are “worth gold.” Nobody in the market is pricing the buildout that way.

Watch on Youtube

Stephen is worth listening to on calls like this. Last time he sat in that chair, he told us large language models would end up like web browsers: free, everywhere, and worthless as a moat. That’s roughly what happened. He invented affiliate marketing, sold LinkShare for $425 million, and has spent 18 years building Collective[i], an AI company that predicts economic outcomes instead of next words. He wrote up the pricing side of his argument as “Peak Token” on his blog, and jokes he’s been in witness protection from frontier-lab investors ever since.

This one is a playbook, not a podcast: why token pricing power is collapsing, what the forward-deployed engineer push is really about, and the one question a CEO should ask before spending an AI budget.

About Stephen Messer: Stephen is co-founder and vice chairman of Collective[i], one of the earliest AI companies (founded 2008), which trains models on pooled, proprietary transaction data to predict economic outcomes: which deals close, which buyers are real, what demand does next. Before that he co-founded LinkShare, effectively inventing affiliate marketing, and sold it to Rakuten for $425 million. He publishes two posts a week at reloadnyc.com.

“If your gap and the premium you’re asking people to pay for a language model disappears in two weeks, but it took you a year to make, that is not the sign of a moat that is going to be defensible.”

Why this matters: The enterprise revolt against token pricing went public last week. Palantir’s Alex Karp told CNBC that CEOs are “livid,” paying for “tokens that create no value” while handing their proprietary data to the labs charging them. Uber reportedly burned its annual AI coding budget in four months. Meanwhile AI agents are getting credit cards and Stripe is building payment rails for them, which means token costs are about to become a line item in every P&L, not just engineering budgets. Stephen’s framework explains what’s underneath: the models are commodities, the data is the moat, and the market has put 90% of its money on the commodity.

🎯 Jump to the best parts

00:00 Why Everyone Is Investing In The Wrong AI
00:44 Introduction
01:47 Why LLMs Are Becoming Commodities
03:16 Why Bigger Models Won't Win
05:48 Compute vs Software
07:42 The Real AI Moat
09:04 OpenAI's Business Challenge
12:09 Economic Foundation Models Explained
15:50 Building Collective[i]
17:24 Why Proprietary Data Wins
21:16 Predicting Business Outcomes With AI
24:44 The Biggest Enterprise AI Mistake
26:24 Why Most AI Wrappers Will Die
28:13 Is AI In A Bubble?
30:00 The Taiwan Chip Risk
32:25 AI, Geopolitics & National Security
34:38 Should Frontier Models Be Restricted?
36:10 Where CEOs Should Actually Invest In AI
37:46 The Most Mispriced AI Opportunity
38:40 Advice For AI Founders
40:58 Lightning Round
42:07 Where To Learn More

Important Links

• LinkedIn: https://www.linkedin.com/in/stephenmesser/

• Collective[i]: https://www.collectivei.com

• Stephen’s blog: https://www.reloadnyc.com

• Last episode with Stephen: Beyond LLMs: How AI is set to reshape global business

Watch or listen now: YouTube • Apple Podcasts

See our last episode with Stephen:

Our biggest takeaways from this conversation

1. Peak Token is here.

Stephen’s core thesis: a language model’s training data is public, so the model is a commodity, and open-source distillation now closes any frontier lead in weeks. The question that kills the business model isn’t whether frontier models are better. It’s how much you can charge for the difference.

“If your gap and the premium you’re asking people to pay for a language model disappears in two weeks, but it took you a year to make, that is not the sign of a moat that is going to be defensible.”

“How much can you charge for that little incremental improvement? ... My question is, can that cover the cost of everything else?”

• The pattern: a frontier release looks untouchable for two or three weeks, then open-source models from China and Japan distill their way to an indistinguishable gap at a fraction of the cost. DeepSeek was the preview, a year ago.

• New laptop-class hardware can run high-parameter models locally. Stephen’s estimate: 90% of workloads run local, and you pay frontier prices only for the rare job your machine can’t handle. That ratio breaks the labs’ economics.

• The labs’ escape route is narrow premium niches, security and biology, where one result is worth real money. Hence the talent fights over those teams.

What to do with this: Price every AI vendor as if their underlying model becomes free in 18 months. If the pitch still works, it’s a real business.

Related reads:
→ Beyond LLMs, with Stephen Messer (his first appearance)

🚀 Build credibility. Drive pipeline. Win in digital assets. We produce institutional-grade research that positions you as the authority in your category, then distribute it to 100,000+ decision-makers. Let’s talk.

2. Forward-deployed engineers are a lock-in play.

The labs’ answer to commoditization is sending engineers into enterprises to build agents, positioned as a replacement for Accenture and McKinsey. Stephen’s read: those engineers have no incentive to build the one thing a token-cost-literate buyer would demand first, an abstraction layer:

Read more