06-reference/research

13f eligibility bdc feeder interval funds

2026-10-22·research-brief·source: deep-research·by Ray Data Co (deep-research synthesis)
disclosure-surfacesprivate-creditai-capex13fbdc

Listed BDC shares ARE 13(f)-eligible — the parent brief's eligibility premise was wrong, and its conclusion survives anyway because Form 13F has no look-through field

The question

"Are exchange-listed BDC shares, rated feeder notes, or listed interval funds 13(f)-eligible in a way that gives partial visibility into AI-capex private credit — verified against the SEC's Official List rather than the FAQ page?"

Context: this refines the one soft claim in [[2026-08-13-ai-capex-lender-exposure-insurance-pension-vehicles]], which asserted the 13(f)-eligibility boundary from regulatory structure after sec.gov returned HTTP 403, and flagged it in-text as needing verification before becoming load-bearing.

What we already know (from the vault)

What the web says

Convergences and contradictions

Synthesis for RDCO

Verdict: REFINED, not overturned. The parent brief's eligibility claim is wrong; its build recommendation stands, for a better reason. Do not extend /investing-smart-money-watch to alt-managers or listed credit vehicles — but the reason is no longer "there is no 13(f) security to report." The reason is that every eligible instrument in this space is a wrapper whose 13F disclosure terminates at the wrapper. This matters because the old reason was fragile and the new one is not. Under the old reason, any future listing event — a listed AI-infrastructure credit CEF, or a non-traded BDC converting to listed status, would reopen the question. Under the new reason it would not: listing the wrapper makes it eligible and still tells you nothing about its loan book.

On the two questions, answered separately, because they come apart here. Legally eligible? Yes for listed BDC common, listed credit CEFs, BDC ETFs and listed BDC convertibles; no for rated feeder notes; no for non-listed interval funds and non-traded BDCs. Does a reported holding reveal AI-capex credit exposure? No, and the incremental information content is exactly zero. A 13F line for Ares Capital states that a manager held N shares worth $Y on the period-end date. The loan-level detail — borrower, industry, coupon, cost, fair value, non-accrual status — lives in Ares Capital's own 10-Q Schedule of Investments, is public on EDGAR, and is public whether or not any manager files a 13F. The parent brief already ranked that surface correctly at ~45 days. So the 13F route is not merely weak; it is strictly dominated by a surface we already knew about. Reading 13Fs to learn about BDC loan books is reading the shareholder register to learn about the assets, when the assets are itemised two filings over.

Pricing the proxy honestly, because "weak proxy worth pricing" was the backlog entry's actual ask. As a credit-exposure proxy: zero, per above. As a flow or positioning signal on listed-BDC equity: weak and badly confounded — the reported market value moves with the NAV discount, rate beta and BDC-sector sentiment, the register is retail-heavy so institutional 13F filers capture a minority of it, and it prints at 45 days. As a coverage matter it is worse than it looks: the 37 eligible listed BDCs are the minority slice of private credit by AUM, and the non-traded perpetual-life vehicles that dominate the asset class are precisely the 13(f)-invisible ones. Whether the listed 37 hold meaningful neocloud or data-center paper at all is still unresolved and is already on the parent brief's follow-up list. One genuine positive, and it is modest and cheap: the Official List is itself a free, auth-free, quarterly fixed-width file giving CUSIP, issuer and class for every listed BDC and listed credit CEF, with an *A*/*D* status column (1,180 additions and 873 deletions in Q3 2026). That is the cheapest available way to construct and maintain the universe for the screen the parent brief's third follow-up wants. It is a universe artifact, not a signal.

The one surface this brief found that the parent brief did not, and it is the only reason the night was worth spending. The AI-capex borrowers' convertible notes are 13(f)-eligible and sit on the list under their own names: Applied Digital 2.750% (038169AB4), Nebius 2.000% / 3.000% / 1.000% / 2.750% (63954QAB2, 63954QAD8, 63954QAG1, 63954QAH9 — all four flagged *A*, added this quarter), IREN 3.250% and 3.500% (46270CAB5, 46270CAD1), Cipher Digital 1.750% (17253JAA4), Riot 0.750% (767292AB1), with CoreWeave, Fermi, Galaxy Digital and Hut 8 common also listed. This inverts the parent brief's framing. The parent looked down the chain for lenders and found wrappers that disclose nothing. Looking up the chain at borrowers, the instrument is named by CUSIP, the borrower is named by issuer, the amount is reportable in PRN, and the de minimis is only $200,000 of principal. That is a real, free, 45-day window into who holds the convertible paper of the non-hyperscaler AI-capex tier, the exact tier the Nvidia release named as its target borrowers. Calibration, explicitly: this is holder-side information about a dual-natured instrument carrying equity optionality, not origination data on private loans, and it says nothing about the private platforms themselves. Confidence that the instruments are eligible: high (V, read off the list). Confidence that the reporting is actually populated rather than nominally permitted: unknown (U) — convertible reporting practice among 13F filers was not tested, and that is the first follow-up below. Net effect on the thesis: zero. No phase marker, no anchor, no position.

Why this is in the vault

This closes open follow-up #5 of [[2026-08-13-ai-capex-lender-exposure-insurance-pension-vehicles]] against the Official List and Rule 13f-1(c) at primary, correcting that brief's eligibility claim while leaving its /investing-smart-money-watch build decision in force on a sounder mechanism. It also records the operational fix for a repeat failure: sec.gov 403s are a missing-contact-UA problem, and curl -A "...(ben@raydata.co)" clears every SEC and eCFR target on the first attempt — so the vault's cms.gov UA workaround generalises to the SEC and no sec.gov question should be answered from secondary sources again.

Open follow-ups

Related

Sources

Vault

Web — primary (all fetched and verified via curl -A "Ray Data Co Research (ben@raydata.co)", HTTP 200)

Not accessed