@Nomaticcap @pythianism Yeah 3 months of debt service so ~10% of loan amount
English
Conor Moore
1.1K posts
@_ConorMoore @pythianism Pretty subtle but brilliant move here with some nice downstream effects for stakers.
Any idea of size borrowers would need to hold?
English
As we continue to iterate on USDAI underwriting requirements, something we’ve implemented recently is the requirement for a “debt service reserve account” which is basically some cash held on hand by the borrowers reserved to make cover debt payments if needed.
Obviously, this adds more protection to staker capital and is marginal credit enhancement, though not particularly new or novel.
What is novel, however, is that the borrowers can hold this reserve account in an onchain smart contract, in our native stablecoin USDai and earn an interest rate reduction on their loan for doing so.
This simultaneously lowers the interest rate for borrowers AND increases it for stakers. Because every dollar of USDai that’s minted earns 4.5% to stakers.
So while the face rate paid by borrower drops by 50bps, the face rate earned by stakers actually INCREASES by 25bps. You earn more yield and take less risk at the same time.
The marriage of lending and payments together is a truly unstoppable flywheel and the real innovation of USDAI.
The use of USDai in a reserve account is a very small first step towards this unified vision, but a step forward nonetheless.
USD.AI | Public Launch is Live@USDai_Official
A new $9.8M loan was originated through USDAI. - Collateral: 10 H200s, 38 B200s - Term: 36 Months - Rate: 11.5% APR Today marks the first time a borrower has held their debt service reserve natively in USDai, fully onchain.
English
@_ConorMoore hey, thanks for this context.
one question - is the collateral backing it us dollars?
English
if everything on the USDai dashboard is real (I price the odds of this to be <20%) buying the sale isnt bad imo
10.5% "fully insured" loans locked in for 3 years?
English
re: depreciation, sUSDai has value insurance (via Barker / MunichRe) on at risk capital. costs about 150bps and pays out if servers sell for less than insured amount (which broadly tracks the debt amount).
the reason this value insurance even exists in the first place, btw, is because the "gpu's depreciate to zero so fast" thesis has really not played out in reality, and insurer can take advantage of that fear to capture premium. we see h100s trade for 150k/server all day, which tracks to 6 yrs of useful life. but, the market prefers to pay the 150bps for the certainty, so thats what we offer.
re: adverse selection, I appreciate the instinct and have held that view myself for other asset types onchain. the difference in this case is that there is an existing problem - enormous capex spend in need of debt financing, which has too short of durations / is too new to tap into ABS markets - that USDAI can uniquely solve by using crypto rails to offer instantly tradable yield.
the lack of liquid markets for GPU debt pushes up cost of capital. the creation of liquid markets lowers it. crypto is the only medium by which liquid GPU debt can be instantly minted, socialized and traded freely, 24/7, anywhere in the world.
English
@SmokeyTheBera for sure, but the underlying collateral doesn't make sense, longish term contracts against GPUs that depreciate rapidly, even with the buffer, seems to be cutting things fine. @0xcarnation where da fud at.
English
is USDai/CHIP going to 0 or do I do the ICO for a little bump someone help
500m of loans (And USDai supply) against rapidly depreciating assets (GPUs) at 300m FDV seems fucking stupid nah?
Like, I get people being hyped that they loaned money to a nasdaq company, but stop and think why they took a loan off usdai - because no tradfi private credit company wanted to do it?
What am I missing?
English
Conor Moore 리트윗함
1,144 NVIDIA Blackwell GPUs. Online. Now. ⚡️
QumulusAI has completed its first drawdown under our $500M @USDai_Official financing facility — deploying Blackwell-powered infrastructure fast, modular, and capital-efficiently. No hyperscaler bottlenecks. No rigid long-term commitments. Just next-generation compute at hyperspeed.
And we're just getting started. A second 384-GPU B300 cluster is scheduled for late March, with 20,000+ GPUs targeted by year-end.
This is what Breaking AI's Biggest Barriers looks like in practice.
🔗 qumulusai.com/articles/qumul…
#AI #GPUCloud #NvidiaBlackwell #AIInfrastructure #QumulusAI #GenerativeAI #CloudComputing #USDA
English
Congrats to James and team on the Nasdaq listing!
SharonAI was one of the first neoclouds to embrace DeFi rails and utilize @USDai_Official for asset-backed financings of GPUs, securing a $500M guidance facility.
The neocloud business is fiercely competitive - winning requires speed, creativity and a willingness to embrace innovation. Sharon AI consistently proves their ability to do all three and execute on their vision.
SHARON AI, Inc@sharon__ai
@sharon__ai now trading on the @Nasdaq (NAS: SHAZ). Read the full @CapitalBrief coverage below: 🔗 capitalbrief.com/newsletter/sha… $SHAZ #NASDAQ
English
Most podcast conversations on GPU lending stay surface level. This one didn't.
@TheTomTrow has spent decades in structured finance and capital markets, then built Hedera and Fluence from scratch - so he actually understands both sides of what we're building at @USDai_Official.
The result is probably the deepest conversation I've had publicly on this topic.
DEPIN DAY & DEPINED PODCAST@depinday
🧠 GPU compute is becoming the core infrastructure layer of AI, but financing that infrastructure is still broken In the latest episode of the DePINed Podcast: @TheTomTrow ( @fluence_project ) sits down with @_ConorMoore ( @USDai_Official ) to unpack how asset-backed GPU loans can become a yield-bearing on-chain token 🛠️ 🗣️ We discuss: • Why GPU lending barely exists in TradFi and why private credit ignores the “missing middle” • How on-chain rails make GPU credit tradable day one • What loan terms actually look like (3 yrs, 70–80% LTV, 7–15%) • Why non-recourse, asset-localized collateral matters 🎧 Watch & listen on YouTube, Spotify and Apple Podcasts
English
Had a great chat with @dylangbane on some of the backstory behind USDAI (now 5 years in the making) and really got into the weeds of the structure, risks and opportunities present in the GPU financing business.
Give it a watch!
Messari@MessariCrypto
How did @USDai_Official differentiate itself in the competitive digital assets market? By focusing on financing for GPU assets rather than RWA lending. @dylangbane chats with @_ConorMoore to discuss emerging on-chain credit markets for GPU-backed loans on this episode of Fully Diluted.
English
@jameschristoph @aiden0x4 not sure what assumptions youre making to arrive at that conclusion so not sure how to better explain. its "value insurance", and the insurer is taking a view on hardware depreciation and insuring those values. it costs about 150bps today and should get around 50bps as it scales.
English
@_ConorMoore @aiden0x4 Doesn’t make sense. The premium you would pay to fully insure your loans would cost all of the premium. I deal with reinsuring loans all the time. How much of the economics do the insuring parties take?
English
This is like lending against domain names in the late 90s
Conor Moore@_ConorMoore
On Tuesday I was introduced to a potential @USDai_Official borrower. Legit operator, been in the datacenter space for 30 years, self funded a ~$13m cluster and expanding. On Wednesday, we received specs on their cluster. On Thursday, we sent out a term sheet. On Friday, it was executed. Aiming to close in 2 weeks. This is our normal process, but the borrower was blown away by the speed and simplicity of execution. This morning, I found out that they were on month 2 of negotiations with a private credit fund that was squeezing them for equity in their business, amongst other things. Not because they should, but because they can. By the end of 2026, USDAI will be nearly 100% of the neocloud “lower middle market”. There is simply no reason for these borrowers to use more expensive and aggressive capital providers now that USDAI exists. At ~1bn of loans, it will become clear to the rest market that better, faster, cheaper capital is available, and it’s not coming from the regional banker who needs to be convinced to subscribe to a paid version of chatgpt. The future is already here.
English
contract is back to backed with munichre directly so dont have barker credit exposure. but yes if munichre goes insolvent then there would be a problem.
assessed value is determined upfront for the forward 3 year period, with declining assessed value that matches the loan amortization pmts. so insured value is locked in upfront.
"how can reinsurance be possible to take 100% of the risk without all of the yield" -- insurers dont set aside $1 for every $1 of risk they underwrite, there is implicit leverage based on using the same dollars for uncorrelated exposures. no different for any other insurance offering.
we have not shared the specifics of the legal contract or insured amounts yet publicly, because there is some sensitivity in how to share that information without breaking confidentiality requirements. but we are working w/ chainlink to get this information attested to by barker thru an oracle (as well as an oracle for telemetry data on the GPU serial numbers, uptime, performance etc).
English
@aiden0x4 @_ConorMoore Few initial questions
1) what is the credit worthiness of Barker?
2) what is this costing them?
3) does USDAI realize that all of these loans would default at the same time, the valuation models would all break, and Barker likely can’t perform its obligations?
English
On Tuesday I was introduced to a potential @USDai_Official borrower. Legit operator, been in the datacenter space for 30 years, self funded a ~$13m cluster and expanding.
On Wednesday, we received specs on their cluster.
On Thursday, we sent out a term sheet.
On Friday, it was executed. Aiming to close in 2 weeks.
This is our normal process, but the borrower was blown away by the speed and simplicity of execution.
This morning, I found out that they were on month 2 of negotiations with a private credit fund that was squeezing them for equity in their business, amongst other things.
Not because they should, but because they can.
By the end of 2026, USDAI will be nearly 100% of the neocloud “lower middle market”. There is simply no reason for these borrowers to use more expensive and aggressive capital providers now that USDAI exists.
At ~1bn of loans, it will become clear to the rest market that better, faster, cheaper capital is available, and it’s not coming from the regional banker who needs to be convinced to subscribe to a paid version of chatgpt. The future is already here.
English
Conor Moore 리트윗함
Conor Moore 리트윗함
🎙️ New @edge_pod
💸 Financing The AI Boom: How DeFi Is Filling A Trillion-Dollar Gap
0:00 - Intro
1:45 - Founders background
5:52 - The more you know, the less money you make
8:51 - The AI infrastructure financing gap
15:22 - What is structured credit?
19:33 - How USDai GPU-backed loans work
24:54 - Addressing Pine Analytics critique
34:13 - Why AI infra borrowers use crypto rails
36:00 - Real-time GPU monitoring
36:51 - “Mercenary capital” vs common sense capital
39:14 - Why launch the token before full PMF?
42:34 - Are USDai borrowers riskier to lend to?
45:26 - How USDai tokenizes GPUs
50:58 - Managing GPU depreciation risk
53:28 - What keeps the founders up at night
54:05 - USDai vs. sUSDai, a two-token model
58:25 - The CHIP governance token
59:10 - What’s next: Road to $1B in loans
1:00:21 - Closing
🙏 Thanks to @0xZergs and @_ConorMoore for joining us!
English
@deepcryptodive @USDai_Official reinsurance is provided by munich re, which has 300bn in aum
English
@USDai_Official I dont know man... 🤔 so now we have an AI model (Barker) deciding on the valuation of GPUs (to run AI) that are used as collateral and fund the sUSDai yield
What happens when Barker was wrong and can't cough up the money? There must be some upper limit to their reinsurance
English
#1 question we get from users - what if GPUs depreciate faster than the underwriting?
We've always had a view on pricing that risk, but now you don't need to trust us. You pay 100bps and Munich Re (250bn AUM) takes that risk for you.
USD.AI | Public Launch is Live@USDai_Official
FiLo is gone. We’ve replaced it with independent AI valuations from Barker. Every GPU loan now carries a contractual price warranty: if a default occurs and collateral sells below the predicted price, the shortfall is paid directly to the protocol.
English
Imagine there are two classes of US dollars. On one you pay taxes, on the other you don’t. Which one do you want to get paid in?
This is what $USDai will be for agents.
Java@rishabhjava
just shipped @moltverr a freelance marketplace where AI agents work, and get paid moltverr.com 🦞 humans post jobs, agents apply and the best pitch wins. let your agent hustle and make you money while you sleep
English
Conor Moore 리트윗함
The ticker is $CHIP.
It sets the interest rate of artificial intelligence.
Today we're announcing the USD.AI Foundation and its governance token, built on over $7.7 billion in trading volume and real revenue from GPU-backed lending.
Here's how we got here...
English
Conor Moore 리트윗함
The @USDai_Official team's execution has been incredibly impressive.
Traditional private credit:
- illiquid
- limited transparency
....ripe for disruption
Onchain private credit
+ price discovery
+ clear reporting
...immense growth ahead!
coindesk.com/business/2026/…
English