[Week 19 of 2026] Prompts, Peers, Plumbing

Welcome back to Price and Prejudice with a few musings from Week 19 of 2026.

The Balance Sheet's Blind Spot

This WSJ article walks through how publicly traded private-credit funds, known as BDCs, carry meaningfully more leverage than their balance sheets show. Reported debt averages 1.2 times equity across the group, but estimates are closer to 1.5 times on a full look-through basis, and 3x or higher at some smaller funds. This discrepancy apparently is driven by an accounting rule that lets a BDC avoid consolidating the lenders it owns, even at 100 percent stakes.

We are trained to read a balance sheet as the picture of what a firm owes. So a leverage cap written in balance-sheet terms (BDCs are capped at 2 times debt-to-equity) creates a sharp incentive to push borrowings into entities that do not consolidate. The fair-value accounting rule that lets a BDC record each investment as a single line was meant to keep the balance sheet readable, but in practice it migrates the real leverage one layer down. This is probably a good reminder that a balance sheet is a model of a firm rather than the firm itself, and that the discounts now spreading across the BDC sector are partly the market's way of pricing what it cannot see directly.

The Terminal's Two Jobs

What is the role of a Bloomberg terminal? The first is information retrieval, i.e. pulling up filings, prices, charts, and news. The second is being the social network where the buy side, sell side, and corporates talk to each other through Instant Bloomberg chat. There are a lot of new startups that seek to disrupt the Bloomberg terminal, which means it needs to disrupt both aspects of the terminal.

This WIRED interview with Bloomberg's CTO Shawn Edwards lays out the rollout of ASKB, a chatbot interface to the Terminal that lets users ask things like "how is the war in Iran going to affect my portfolio?" in plain English. ASKB is squarely aimed at the first role of the terminal and it looks like a genuine upgrade: natural language becomes the medium, and the script does what an analyst used to do. So the bottleneck here is human – your ability to come up with good questions and prompts, since information retrieval is much quicker than before.

The more interesting question is what this does to the second role of the terminal. The chat network is widely considered the real moat behind the pricey subscription, and part of what bound that network together was a shared craft. Knowing the terminal is a kind of professional identity, so likely the network loses one of the cultural ties that kept it sticky.  

Single Points of Failure

Yesterday Canvas – the learning platform used by about 8,000 schools and roughly half of all colleges in North America – went dark due to a hack. The contents of the hack aside, the timing was also very unfortunate: students across the country are in finals, and I happened to be running a final exam on Canvas the morning the outage hit.

The Canvas episode is informative because the same architecture sits under finance. For example, most US equity clearing flows through DTCC, most cross-border bank messaging through SWIFT, and most futures clearing through CME and ICE. Each is convenient, well-run, and a single concentrated target. Given this backdrop, the risk of a hack has also been on the minds of regulators. Cyber risk briefly held a top slot on the financial-stability agenda after the 2017 NotPetya attack but it then slid down the list as nothing catastrophic followed.

What an episode like this exposes is that we usually only learn which nodes really matter after one of them goes down. A Canvas outage during finals is probably recoverable, and the worst that happens is a delayed exam (and a ton of complaints) and a long week for the IT help desk. The financial analogue is not. If DTCC or SWIFT had its own hacked morning, there is no professor on the other end improvising a workaround, and the cost of redundancy that nobody wanted to pay for in the calm years suddenly looks cheap. The unsettling part is that this is the kind of risk that does not show up in any return until the day it shows up in all of them at once, so it's probably not a bad idea to keep tabs on such related concerns.