There is a story that sums up decades of asset-backed lending. A fund that lends against private jets needs to value its portfolio every quarter to report to investors. They call the usual expert: a charming man in his seventies who shows up with stacks of binders, spreads them across the boardroom table, digs through comparable transactions for days, and delivers a number. He charges the equivalent of two months of work for it. And the fund, which pays without complaint, doesn't fully believe him: it is quietly accepting the risk that his number is wrong.
Thomas, co-founder of Barkr, told us that story on the latest episode of All You Can Debt. But the point is not the man with the binders. The point is what his method reveals about how trust works in private credit.
An industry of opinions
Asset-backed lending carries two risks: the borrower and the asset. For the first one, we built half a century of infrastructure: KYC, scoring, increasingly refined underwriting. The second one we left to opinions. Ask five experts to value the same asset and you get five prices, all delivered under the same unspoken clause: this is what we believe, and you can't sue us for it.
That asymmetry has a cost the market pays every day without seeing it: excess capital reserves, rates inflated by uncertainty, deals that never close because no one can say with authority what the collateral is actually worth.
And it has a bigger cost when something breaks. The sector's major recent blowups were not borrower underwriting failures. They were collateral events: assets that didn't exist, that were already pledged to another facility, or whose cash flows never reconciled against the source. Thomas put it on the episode with a precision worth framing: First Brands was not a marks problem. It was a fraud problem.
What changes when someone stands behind the number
The interesting thing about Barkr is not just that it uses AI to value assets. It's that it backs the price with an insured contractual guarantee. If the asset liquidates below the promised value, someone pays. The price stops being an opinion and becomes a commitment with consequences.
That detail reorders the incentives of the entire chain. A firm that opines can be wrong for free; a firm that guarantees has to be precise, because every error costs it. And the lender holding a backed price can recognize risk transfer, free up capital, and lend cheaper. Trust stops being a courtesy between parties and becomes a piece of infrastructure with a price, a contract, and someone on the hook.
The GPU debate shows why this matters right now. Michael Burry argues that hyperscalers are inflating earnings by stretching the depreciation schedules of their chips. Thomas, whose business is promising prices and answering for them, disagrees with data: real life cycles are closer to five to seven years, older generations don't die but cascade down to companies that need compute without needing the latest chip, and no one has been able to show him a single data center that has gone dark. Who is right is almost beside the point. What matters is that one of them bets his P&L on his answer, and the other publishes on Substack.
The price is half the question
A guaranteed price answers exactly one question: what is this asset worth. Three remain: does it exist, who owns it, and do its cash flows land where the contract says they should. A flawless price on an asset already pledged to another facility is still worth zero.
Those three questions are the ones Vaas answers by connecting to the source, the originator's LMS, the bank accounts, the payment rails, instead of trusting what the borrower reports. Valuation and verification are complementary layers of the same move: taking trust out of the territory of opinions and reports, and putting it in the territory of evidence with someone who answers for it.
Private credit grew. Its infrastructure didn't. The good news is that you can already tell who is building it.
The full conversation with Thomas, including his thesis on the future of compute and how he convinced one of the most conservative insurers in the world to stand behind model-generated prices, is on the new episode of All You Can Debt, Season 2.
🎧 Listen on Spotify 👉 https://shorturl.at/uWtXB
⏯️ Vaas YouTube 👉 https://shorturl.at/Ew1oK
Learn more about Vaas: https://getvaas.com
