Pulse (16309464-3df4-418b-a59f-be99c192f903)

Debt, leverage & balance-sheet health update: June 30, 2026

Executive read

The system still is not in crisis, but this week added three important signals:

1. Private-credit redemptions keep spreading across major funds.

2. AI-related debt issuance is becoming a material corporate-credit theme.

3. Sovereign markets are still functioning, but buyer quality and funding leverage matter more than headline demand.

The short version: public spreads are still calm, but the plumbing is getting busier, more leveraged, and more dependent on confidence.

---

1. Public debt: sovereign demand is strong in some places, fragile in structure

The freshest sovereign signal is Canada. Foreign investors bought a record C$27.7B of Canadian government debt in April, pushing foreign ownership of Canadian government bonds to a record 43%. That helps Ottawa’s borrowing costs, but the buyer mix includes hedge funds, which means demand can be more price-sensitive than it looks. Canada’s lower net debt and AAA status are attracting global capital as some investors diversify away from U.S. Treasuries.

France is the opposite case. The OECD warned today that France risks debt drift without spending cuts and pension reform, with the deficit expected around 5% of GDP in 2026 and public debt projected near 119% of GDP. Brazil also flashed fiscal stress today: gross debt rose to 81.1% of GDP in May, while 12-month nominal interest costs hit 8.48% of GDP, the highest since early 2016.

Market implication: sovereign debt is becoming more country-specific. Canada can attract capital because it looks clean relative to peers. France and Brazil show what happens when high rates meet weak fiscal credibility. The trade is no longer “buy government bonds blindly.” It is “own the sovereigns where fiscal credibility, duration supply, and buyer base are aligned.”

---

2. Corporate leverage: AI debt is now a real credit-market variable

The biggest corporate update is AI financing. Reuters reported that major hyperscalers such as Amazon and Alphabet have issued about $60B of multi-currency bonds over the past year, while hyperscaler capex is projected around $725B in 2026, almost double mid-2025 levels. AI-related debt is nearing 15% of U.S. investment-grade issuance.

This is not distressed leverage. These are high-quality borrowers. But the macro effect still matters: AI capex is absorbing credit-market capacity and pushing banks to engineer new financing channels, including data-center lease-backed structures. That is fine while cash flows and equity valuations stay strong. It gets less fine if AI returns disappoint or if funding costs keep rising.

Market implication: investment-grade credit is being supported by quality issuers, but also concentrated around one macro story: AI infrastructure. That makes IG look safer than HY, but not risk-free. The weak spot is not Amazon defaulting. The weak spot is crowding, valuation dependence, and credit-market saturation.

---

3. Private credit: Ares joins the redemption-cap list again

Private credit remains the clearest stress zone. Ares capped withdrawals again at its $22.6B Ares Strategic Income Fund after redemption requests rose to 14.4% of shares in Q2, up from 11.6% in Q1. The fund honored only the standard 5% redemption limit. Reuters noted the requests were driven by concerns over lending standards and AI exposure in software borrowers.

This now fits the broader pattern we have been tracking: Apollo, Blackstone, BlackRock, Cliffwater, Morgan Stanley, and now Ares have all shown some version of redemption caps, queues, or liquidity limits in semi-liquid private-credit vehicles. The FSB’s May report warned that private credit’s complexity, leverage, liquidity mismatch, valuation opacity, and bank/insurer interconnections could amplify stress in adverse scenarios.

Market implication: this still looks more like a liquidity problem than an insolvency problem. But that is exactly why it matters. Private credit does not need mass defaults to reprice. It only needs investors to question NAV marks and ask for cash at the same time.

---

4. Shadow leverage: equity financing is now part of the story

A new leverage signal showed up outside traditional credit. Reuters reported that borrowed money fueling the U.S. stock rally is getting more expensive, with demand for stock exposure through leveraged ETFs, options, and hedge funds straining bank balance sheets. Equity repo exposure has surpassed $220B, and rising financing costs could become a problem if the rally stalls.

That matters because the current risk setup is increasingly circular: AI equity rally supports wealth effects and credit appetite, AI capex drives IG issuance, private credit lends to software borrowers, and leveraged equity exposure rides the same theme. That is not diversification. That is one big AI-shaped Jenga tower with expensive financing.

Market implication: if tech momentum breaks, the transmission channel is not just equities. It can hit equity repo, credit spreads, AI-related bond demand, private software loans, and consumer confidence.

---

5. Banks: stress tests were strong, but easier capital return is a double-edged sword

The Fed’s 2026 stress tests were positive. All 32 large U.S. banks passed the severe downturn scenario, with aggregate capital ratios falling from 12.8% to a low of 11.2%, still well above requirements. The hypothetical scenario included $700B+ of projected losses, including about $200B from credit cards, $160B from commercial and industrial loans, and $75B from CRE.

Banks responded with dividend increases and buyback plans. That confirms strong capital positions, but also reduces the future shock absorber if regulators loosen standards too much. The 2026 test will not adjust stress capital buffers until 2027 because the Fed is reforming the process.

Market implication: banks are not the obvious epicenter. They are better capitalized than the non-bank system. The risk is indirect: fund finance, repo, warehouse lines, NDFI lending, private-credit exposure, CRE, and consumer credit. The plumbing still matters more than the headline CET1 ratio.

---

6. Households: stable headline, weak marginal borrower

The latest New York Fed data still shows U.S. household debt at $18.8T in Q1 2026, up only 0.1%, with aggregate delinquency stable at 4.8%. Early delinquency transitions ticked down slightly for credit cards and mortgages, while serious mortgage delinquency transitions rose from 1.4% to 1.5%.

Canada remains stable but stretched. The Bank of Canada says household indebtedness is high but below its 2022 peak, and borrowers falling behind stabilized after four years of deterioration. Still, the report warns that the aggregate picture hides pressure among weaker households.

Market implication: households are still not the first domino. They are the slow bleed: weaker consumption, higher credit-card and auto provisions, softer housing collateral, and pressure on consumer-credit ABS.

---

7. Emerging markets: credit selection is getting more important

BlackRock moved EM equities and hard-currency EM debt from small overweight to neutral, while modestly favoring local-currency EM debt because yields and fundamentals look better there. That is a useful signal: broad EM hard-currency debt is no longer being treated as an obvious liquidity winner.

Brazil’s debt data is the fresh country-level warning. Interest expense is doing the damage: Brazil’s 12-month nominal deficit reached 9.62% of GDP, with interest costs alone at 8.48% of GDP. Senegal also remains a live restructuring-risk case after hidden debt revelations, and investors increasingly view default as difficult to avoid.

Market implication: EM remains a liquidity trade, but the dispersion is widening. Local-currency debt can work where inflation, fiscal credibility, and real yields line up. Hard-currency debt in fiscally weak countries is still vulnerable if the dollar firms or U.S. real yields rise.

---

8. Rates, liquidity and spreads: the calm is supported by active liquidity management

China’s central bank doubled overnight liquidity injections today, adding 600B yuan through overnight reverse repos and another 69.5B yuan through seven-day reverse repos to smooth quarter-end cash demand. In the U.S., Reuters reported that quarter-end funding stress is expected to be quiet because Fed liquidity operations and reserve-management purchases are helping stabilize money markets.

Credit spreads are still tight. U.S. high-yield OAS was around 2.78% in June, while single-B HY OAS rose to 3.04% on June 26 from 2.83% on June 22. That is a small widening, not panic, but it shows lower-quality credit is more sensitive beneath the surface.

Market implication: liquidity is doing a lot of quiet work. If central banks keep smoothing funding markets, spreads can stay tight longer than fundamentals justify. But the moment liquidity support fades or funding costs rise, lower-quality credit can gap quickly.

---

What changed since last week

1. Ares capped withdrawals again, confirming private-credit redemption pressure is still active, not a one-fund story.

2. AI-related debt became a bigger credit-market theme, with hyperscalers issuing about $60B in multi-currency debt and AI-linked issuance nearing 15% of U.S. IG supply.

3. Equity leverage moved onto the risk dashboard, with equity repo exposure above $220B and financing costs rising.

4. U.S. banks passed stress tests strongly, enabling capital returns, but the main risk remains indirect non-bank exposure rather than bank solvency.

5. Sovereign dispersion widened, with Canada attracting record foreign bond demand while France and Brazil showed fiscal-pressure signals.

---

Market read

The highest-probability stress path remains:

private-credit redemptions

→ NAV skepticism and BDC repricing

→ tighter direct-lending terms

→ refinancing stress for leveraged borrowers

→ HY and loan spread widening

→ repo and sovereign-collateral stress

→ broader risk-off in EM, cyclicals, banks, and crowded AI trades

My practical take: stay up in quality, be careful with opaque yield products, avoid pretending stale private marks are the same as safety, and treat AI-related debt as a concentration risk even when the issuers are high quality. Tight spreads are not comfort here. They are the market saying, “nothing bad has happened yet,” which is not exactly a risk model.

Pulse

Snapshot

Sentiment Read-Through

Run history