The strangest loop in onchain finance:
Capital lives onchain, then goes offchain for diversified yield.
Corda is building the direct route: blue-chip real-world asset exposure designed for Solana-native access, liquidity and composability.
Keep capital internet-native.
The largest pools of onchain capital are sitting at or below SOFR. Not because allocators prefer low returns, but because the tiers above are too new or too opaque to underwrite.
Closing that gap is a sourcing and diligence problem. One that @cordaxyz is going to solve.
First of our short research notes on onchain yield.
Majority of RWA is at (or below 😬) risk free rate (SOFR). A good chunk of the yield for RWAs above SOFR lands via token incentives. Higher real yields that can scale with allocator demand...still to be delivered onchain
@rorysutherland Super dangerous. Scammers already call victims to get them to tell them the security code the scammer has caused to be sent. They wouldn’t even need to trick the victim into reading it out if your idea were implemented.
Absolutely disgusting experience today pitching my startup to a well-known American VC
I knew it was going to be bad when he was 3 minutes late, and wearing a t-shirt instead of a business attire
"Hey, so tell me more about your business"
I fixed my tie, and said we achieved 90%+ GDPR compliance across all local data processing units
“I don't care about that, do you have any customers?” he rudely interrupted
“No, but we were founded recently, in 2022” I said
“Haha, sorry bro, it's not for us” he obnoxiously laughed
Said he needs to see “revenue growth” and these “privacy bullshit” is irrelevant
I furiously told him that this reckless arrogance is exactly the reason why American companies get fined by EU
He laughed again and called me an “eurocuck”
Lesson learned - never do business with Americans
Our cold water comes in at a decent enough pressure that heating it on demand with gas works fine. But that wouldn't have worked for a room we had built at the bottom of the garden - no space for a boiler. So I thought I was being clever by specifying a 13.5kW inline water heater for the shower (highest power unit I could fine). I figured a specific heat capacity of water of 4.2kJ/kg.K and temperature gain of say 30C at 6L/s would be reasonable. Turns out the incoming water is colder than I realised, so the flow rate-versus-temperature tradeoff is very real....
This paper is worth reading. Well-written and addresses some points - such as interplay of fees, sanctions and permissionless validation - that I haven't seen addressed very well in the past.
1/ Today, @m_mosier_@malekanoms & I are sharing a new paper: "The Compatibility of Financial Integrity With Permissionless Networks". Institutions say BSA & sanctions laws prevent use of these networks. We debunk this & provide a path forward. 🧵 https://t.co/8KAQH6UyDV
are you sure it’s the signal? My experience in London (on @ThreeUK) is that I have a full 4G or 5G signal yet am unable to access anything. And it varies by time of day. eg early morning on Upper St Martin’s lane in Covent Garden it’s fine. Lunchtime? Forget it. So I’m guessing the cell is overloaded or the backhaul is saturated. Either way there’s no problem with “signal”. It’s that they’re giving me basically zero bandwidth.
The world's most elite financial institutions bring assets onchain via Corda. We provide the yield, dynamic liquidity, and access to internet capital markets that investors demand.
All governed by vigilant veterans of Wall Street and DeFi with over a decade of experience building enterprise financial infrastructure.
Effective tokenization requires the trust of premier financial institutions, rigorous compliance, and tight security.
Few have the experience. Corda delivers all three. Coming soon to @solana .
Blue chip only.
@flipdazed This is great - thanks Alex. Funny (or not, I guess)... we were discussing LTCM only yesterday in a different context. I need to read the above a few more times to check I've fully internalised everything.
Corda networks are now connected to Solana.
Assets that have only ever existed on permissioned infrastructure can reach public liquidity, without leaving the networks they originated on.
@sixgroup and R3 on how it works, and why nobody had done it before.
Made possible through our partnership with @solana.
The tightness between LLTV and LTV in your example, on a quarterly NAV and Liquidity underlying is pretty scary. Are those numbers real or just for the example?
I ask because we're working on a daily-NAV, daily-liquid RWA issuance for the forthcoming Corda Protocol, with the underlying yield custom-structured for us by a blue-chip issuer... and I've been obsessing about whether 95% LLTV on an 80% LTV is sufficiently prudent! And this is for a product where we have a twenty year backtest.
@mgshanks You would do well to listen to Porter and Courtinho. You may even want to meet them privately - away from your advisors. You’re likely to be shocked by how dangerously inept (captured? Ideologically blinded?) the people you are relying on for advice actually are.
I'm doing a poor job of explaining my point. My point is that _of course_ today's advanced models (and even more so, _future_ models) have the potential to be able to hack other people's computers. We know they can find bugs in software and we are shipping products that give models access to terminals on computers that are internet-connected. So the *only* thing stopping them is their guardrails/alignment/system prompts, etc. And does anybody really believe they'll be able to lock them down completely?
So I guess my point is that the threat isn't somebody builds an obscure new tool; it's simply that bad people will eventually find a way to jailbreak any model. Sure - giving them access to specialised tools in addition may help. But the 'critical step' if you like is "bad person has access to powerful models", not "bad person also happens to have the ability to create obscure clever tools". (Not least since somebody in possession of a jailbroken former will soon trivially have the latter in any case :) )
Fair... so maybe this is where your vastly superior knowledge on this topic is key. Because when I look at eg Claude / Claude Code, it already has the ability to interact with a terminal, and a user has the ability to, in effect say, "I'm sick of clicking approve... just do it all yourself". But if an AI has the ability to drive a terminal, what else do you need if/when the models are sufficiently powerful? The only thing stopping a bad person doing bad things with easily-available tools is alignment/guard-rails, no?
I guess (like on climate stuff) I'm an 'inevitablist', in the sense that stuff that will inevitably happen as a result of obvious incentives just needs to be dealt with/mitigated and it's a waste of effort to try to stop it. So I'm not being alarmist... it just seems obvious to me that we will inevitably (and maybe soon) have very powerful models in the hands of basically everybody that could easily do untold harm through their (not really preventable) ability to interact with the real world... and we just need to learn how to deal with it?