The insurance / Bermuda endpoint — where the risk finally lands

Web-verified 2026-06-09. Structured + edges + sources: spec-insurance-bermuda.json. The terminal node of the "who holds the bag" chain (macro-private-credit-marks). Overlay; not used in the proofs.

Scale — STRONG

• PE firms control >$700B of US life-insurance assets — ~25% of US life insurers (vs <10% Europe, ~5% Asia); Athene ~40% of new PE capital into the sector.
• Bermuda long-term reinsurance: ~$1.52T assets (Sep 2025), >80% ceded from the US; the US Treasury has joined the scrutiny.
• US life/health reinsurance ceded from Bermuda: ~$928B (2025) vs ~$205B (2014) — ~4.5× in a decade.
• In 2025 alone, Apollo's Athene + KKR's Global Atlantic moved >$130B of liabilities offshore.

The captive circularity — STRONG

• ~60% of ~$1T US reinsurance (2024) went offshore, much of it to captives the same insurer/PE group controls — so the "risk transfer" is largely internal, not genuine.
• AM Best: ~⅕ of Athene/Global Atlantic investments are loans to affiliated funds — self-dealing on the asset side too.
• Captive Bermuda/Cayman structures = lighter capital, alternative accounting, fewer disclosures — boosting reported ratios without a real third party bearing risk.
• Net: a retiree's annuity can be backed by manager-marked private credit (often affiliated-fund loans), with the liability ceded to a captive the same group controls, in a reduced-disclosure jurisdiction. Self-marking + circular self-dealing + regulatory arbitrage + opacity — stacked under retirement income.

Regulators are warning — STRONG

IMF (alarm on insurer private-credit exposure), FSB (private-credit vulnerabilities report, 6 May 2026), FIO ("shadow banking / shadow insurance"), NAIC (tougher asset-adequacy testing for offshore-ceded business, 2025), US Treasury (Bermuda-market scrutiny). Counter-view recorded: PE-insurers argue long-dated annuities are well-matched to long-dated illiquid credit.

Takeaway

This doesn't predict failure — it identifies where an unwind would land: retiree annuities, valued at marks the holders can't check, in jurisdictions built to disclose less. It is the same Apollo that recurs in the Epstein-finance thread (spec-crypto-sec-epstein), now the largest force in American retirement income.

Where this sits in the larger pattern (updated 2026-06-11)

The offshore-captive mark is the fourth of five self-marked numbers — and the terminal one, because it is where the other four come to rest: bank HTM cost (macro-bank-htm-marks), AI fair-value marks (fin-google-amazon-anthropic-meta), private-credit NAVs (macro-private-credit-marks), insurance captives (here), and AI-compute depreciation (fin-ai-depreciation-debttrap). The self_marked_value proof (U1–U4) shows the four marked balance-sheet classes are one defect whose gaps correlate; the cross-sectional analysis confirms it empirically (~91% common factor in credit). The danger of the Bermuda endpoint is precisely that it pools correlated, self-marked risk into one lightly-disclosed place while reporting it as diversified — so when the common factor moves, the "transfer" that was internal all along provides no cushion.

← Research index · structured data: spec-insurance-bermuda.json · spec-insurance-bermuda.md