Robin Wigglesworth · editor of FT Alphaville (Financial Times), financial journalist and historian — author of Trillions and of A Fabulous Debt: The Epic Story of How Bonds Built the Modern World. A reporter, not a manager: his "views" are assessments of disclosure quality, balance-sheet risk and business durability, never positions.
The AI-displacement de-rating misreads the product: nobody pays Moody's for the credit work (investors automate that themselves), they pay for the letter — mandates, insurance capital rules and the NRSRO designation written into US law. "People will be shocked at the resiliency of their business model."
The same durability case, stated as a moat: rating letters are "the language of credit," a shorthand all three agencies deliberately share because standardisation is the product. "One of the most stubborn oligopolies in the history of business probably" — with the caveat that private-label raters serving PE-owned insurers are "a lot iffier."
Grouped with Meta and Alphabet as a genuine cash machine that outgrew self-funding. "I'm not worried about Facebook and Alphabet or Amazon going bust" — with the caveat that a chunk of reported net income is "other income," i.e. mark-ups on stakes in Anthropic, OpenAI and SpaceX.
"Broadly understood that Anthropic looks financially a lot healthier than OpenAI" — one reason it is moving more aggressively toward an IPO. Also one of the private stakes whose mark-ups flatter hyperscaler "other income."
Marc Rowan's argument that private credit makes the system safer is one Wigglesworth is "quite in agreement" with — the idea is "fantastic," the current vintage isn't: too much money, standards slipped, PIK up "massively," losses coming but "not systemic." An asset-class view, not a company call.
Fink was right about bond ETFs: the Icahn warning that a hyperliquid wrapper on illiquid credit "is going to be lethal" didn't come true — fixed-income ETFs are "ameliorating" liquidity and pushing credit toward equity-like trading ("fascinating and a little bit scary"). Earlier: Fink's "AI securities" as compute futures — plausible, but "just because you say something is an asset class doesn't make it so." No company view.
Jon Gray's case that private credit de-risks the system gets his agreement — lending in locked-up funds is "way safer" than in banks — alongside a near-term call that the asset class is frothy, PIK-masked and headed for losses (not systemic), and that semi-liquid retail wrappers are "dumb." Earlier: an Nvidia chip-financing MOU signatory, discounted as a press release. No company view.
The best discloser and the biggest number — $800bn of purchase commitments (chips, memory, cooling, power), almost half the hyperscaler total, with ~$200bn flagged short-term. If short-term means the next twelve months, published capex expectations are too low.
Credited as the source of the lease numbers — "they did God's work in going through all the filings." Wigglesworth's own contribution was the parallel purchase-commitment tally. Attribution, not a view.
Illustration only — "high-grade corporate bonds like IBM or Microsoft" as the rung on the collateral ladder before the market moved on to asset-backed securities and pushed the money-like assumption too far.
A yardstick only: the 1873 failure of Jay Cooke & Co. — the man who financed the Union's war effort — was "the equivalent of JP Morgan going bankrupt today overnight." Historical analogy, no view.
The template for the whole $1.5trn: Hyperion is a Blue Owl JV where Meta takes 20% equity and guarantees a 20-year lease that covers the vehicle's costs — a real, near-inescapable obligation that never appears as Meta debt. Not a solvency worry; a disclosure one. He'd rather see plain-vanilla bonds and let the market price it.
The counterparty-transformation case: an AAA/AA credit routing spending through far weaker data-centre developers and neoclouds, whose "backlog" is really Microsoft's own off-balance-sheet commitment. A plain-vanilla corporate bond would be cheaper — so "flexibility is probably a convenient excuse" for optics.
Picks-and-shovels, but the price embeds the assumption that nobody else can build GPUs at Nvidia's scale and quality — and a millennium of financial history says pricing power erodes. "Monopolist-like pricing power tends to not last very long… it never lasts forever as far as I know."
He hasn't reported on it directly but relays the FT house view that Anthropic looks financially healthier, and questions how much private "revenue" is actually cash. Waiting on the filing: "I'm really looking forward to the S-1s… that's going to be a popcorn moment."
Meta's JV partner on Hyperion: it owns the other 80%, funds itself with bonds sold to outside investors, and is repaid out of Meta's 20-year lease guarantee. The counterparty in the template deal, not rated.
Cited twice: Google is buying compute from it "for a super super high amount of money," and orbital compute raises an unpriced lending question — how do you maintain or replace chips that burn out in space? Also a marked-up stake inside hyperscaler earnings.
A one-line aside: asked whether a credit mania is even possible, he says "there aren't any meme bonds around today. I guess maybe TLT is the closest. Or the levered version of TLT." A quip about retail behaviour in long duration, not a duration call.
"The real king of debt" — the biggest EBITDA-vs-net-income-loss gap he has ever seen, a backlog that is really the hyperscalers' off-balance-sheet commitments, and delayed-draw term loans dressed as prudence. Fails the fallback test: Google still has YouTube if AI disappoints; a neocloud has nothing.
Saylor's "bitcoin yield" and income-fund pitch is "basically just a return of capital, which is not the same thing" as bond income — the host's charge, endorsed ("No, agree"). Part of a wider warning about "dressing up a fox to look like a goose." A labelling critique, not a bitcoin call.
The named exception to "the big hyperscalers are probably okay" — the most indebted relative to revenue, without the others' cash ballast. "In hyperscaler terms there's the rest and then Oracle… clearly the weakest of the litter."
In one line: Wigglesworth is a reporter, not a manager — his edge is filing archaeology and a thousand years of bond history, and his organising claim is that debt, not equity, is where the system breaks: the AI build-out is a debt cycle (~$1.5trn of hyperscaler lease commitments plus ~$1.5trn of purchase commitments that walk, talk and quack like debt), and the bond market is the disciplinarian that governments fear — now less of a safe haven because it sits in more leveraged hands. Read every "view" here as an assessment of disclosure, structure and balance-sheet durability, never as a position.
The bond market is the real disciplinarian. "The price of bonds is the price of money." Trump shrugged off a ~20% S&P drawdown but paused Liberation Day after "two bad days on the bond market"; the gilt market removed Liz Truss in 50 days. Historically the same power won wars — Britain beat Napoleon because it "could fund itself almost infinitely at 3%" through consols (2026-SEP-11).
Sovereign debt: trajectory, not level. Still "on the relaxed side," but "10% less relaxed now than I was 5 years ago" because deficits "we have not seen outside of severe wars" are now entrenched in a strong US economy — Powell's "the level is not unsustainable but the trajectory is not sustainable." The fix is political (tax a bit more, spend a bit less), not arithmetic.
The safe haven has leveraged holders. The Treasury basis trade is levered "10, 20, 30, 50" (anecdotally 100) times through a ~$12trn repo market — "the dark matter of finance," and what killed Lehman and Bear Stearns. Treasuries are "where you hide when things are on fire elsewhere. Well, that's become a little bit more unsafe." The Truss gilt crash is the template: LDI "kindling" turned a sell-off into a margin-call doom loop.
"Safe" and "inflation-proof" bonds carry duration. TIPS were "taken to the woodshed" in 2022; Austria's century bond fell ~80% while Argentina's defaulted one did better on its coupon; a 40-year UK linker did worse than Bitcoin.
The headline number, and how he got it. Hyperscaler off-balance-sheet lease commitments went from ~$1trn to ~$1.5trn in a single quarter — ~$500bn of leases already commenced (visible as payment obligations, never as debt) and ~$1trn not yet commenced, living only in a footnote. Goldman Sachs did the lease sweep; he built the parallel purchase-commitment tally out of the 10-Qs himself, and got the same shape: ~$1trn → ~$1.5trn.
Debt cycles end differently from equity cycles. Equity-financed booms break benignly — the dot-com bust halved the market and was economically "a nothing burger." Debt-fuelled capex cycles are different: "even when the underlying premise comes true and AI transforms the world, quite often they end in tears." The technology being real is not the question; the funding mix is.
Disclosed is not the same as transparent. Nobody is hiding the capex — but preserving "the optics of pristine balance sheets" through non-uniform wording destroys quarter-to-quarter comparability. His preference throughout: issue plain-vanilla bonds "and let the debt markets do the talking," because for an investment-grade issuer that is also the cheapest route — so "flexibility is probably a convenient excuse."
Crises come from assets believed safe. Junk defaulting is fine, "that's just risk and reward." Catastrophe arrives when a business model rests on something treated as money that turns out not to be — government bonds, then high-grade corporates, then asset-backed securities. Long calm since the last crisis is the precondition, not the comfort.
Private credit: long-run bull, near-term bear. He wants the asset class to grow and thinks it genuinely de-risks the banking system, but too much money arrived too fast on backward-looking data and an "illusion of safety" that is really the absence of mark-to-market. A bad — not catastrophic — default cycle "has started already, but it's getting masked" by PIK and refinancing inflows, on recovery assumptions he calls "fantastical" for asset-light borrowers. Public BDCs could trade to 30–40¢ of NAV in a crisis. Repeated in September: frothy enough that lenders cold-call journalists, PIK up "massively," heavy software exposure — "lots of people are going to lose money," but "not systemic," and he agrees with Apollo's Marc Rowan and Blackstone's Jon Gray that moving lending out of banks into locked-up funds "is actually safer for the system."
Structure rules he'd impose. No semi-liquid wrappers for illiquid assets ("if you invest in loans with a 5-year tenor, then you should be locked up for 5 years" — by September simply "dumb"); longer lockups even for mutual funds, because same-day or one-day liquidity is "a massive systemic danger"; and no leverage stacked on leverage — a vehicle lending to highly levered companies should ideally carry none itself. The exception that surprised him: fixed-income ETFs, which Carl Icahn warned would be lethal, are instead making credit more liquid — and more equity-like, "fascinating and a little bit scary."
Watch words. "Democratization" is "usually a code word for jamming something down the necks of retail investors." Financing MOUs are marketing until the structure exists: "take press releases with a pinch, maybe a fistful of salt." And "income" is not income when it is return of capital — Saylor's "bitcoin yield," covered-call funds and ZIRP-era structured products are "dressing up a fox to look like a goose."
What he defends. The rating agencies — customers buy the letter (mandates, insurance capital rules, the NRSRO designation in US law), not the analysis, and the letters predict default better than their reputation suggests. Credit needs "a language." His worry sits one layer down: private-label ratings inside a tangle of private credit, private equity, and PE-owned insurers.
Timing: he claims none. "Capex bubbles can continue for a long time until it becomes very obviously unmanageable," and the industry's "remarkably incestuous" tangle of financing agreements, co-investments and supplier/customer relationships both worries him and can extend the cycle. He also keeps a 2008 cartoon on his desk to remind himself that not everything is 2008.
Transcripts
One dated page per appearance — each has its stock table, talking points, and the saved transcript. Newest first.