AXA XL's Galileo Re Cat Bond Prices the First US Terrorism Risk

For 24 years US terrorism risk has sat on a federal backstop rather than in private capital markets. AXA XL has now sponsored Galileo Re Ltd. (Series 2026-1), a $67.5 million catastrophe bond whose single limit covers US named-storm risk and, for the first time in cat bond history, US terrorism.

The size is trivial against the Terrorism Risk Insurance Act. The structure is not, because it prices a peril with no loss history and settles off an index that only pays once the government certifies the event.

$67.5M

Size of the Galileo Re Series 2026-1 notes.

2.16%

Initial combined modeled expected loss across both perils.

Spread paid to investors on the notes.

Key Takeaways

$67.5 million across one shared limit covering two unrelated perils, so a loss from either draws down the same collateral rather than each carrying a dedicated tranche.
2.16% combined modeled expected loss against a 6% spread, roughly two and a half times, which is the price of model uncertainty rather than of modeled loss.
The first terrorism cat bond to use an industry-loss trigger. Every prior terrorism transaction used indemnity, and 81% of H1 2026 risk capital still does, up from 79%.
TRIA expires in 2027, the bond matures in June 2028, and the trigger index only counts losses from events certified under that same statute.
$17.98 billion of H1 2026 issuance across 83 transactions, with the outstanding market at $65.6 billion, is the capital backdrop that made an untested peril placeable.

The First Cat Bond to Carry US Terrorism

Cat bonds have securitized hurricane, earthquake, wildfire and pandemic risk for two decades. Terrorism has been the conspicuous absence, and for structural reasons. The peril has no stable base rate, a single attack can move the whole loss distribution overnight, and severity is driven by human intent rather than a physical process a model simulates from first principles.

Congress created TRIA in 2002 after the private market withdrew from terrorism coverage almost entirely, which removed most of the commercial pressure to look past the backstop for capacity. Galileo Re is the first test of whether capital markets will price the risk directly.

The two perils are modeled by different means under one limit. Verisk modeled the hurricane component with its AIR model, the framework behind most property cat bonds. Moody's modeled terrorism with its RMS model, which simulates attack scenarios across target selection, weapon type and casualty distributions, because there is no loss record to fit a frequency-severity curve to.

Pooling a physically modeled peril with a scenario-modeled one under a shared limit is itself an assumption: that the correlation between a major hurricane and a terrorism event hitting the same exposed portfolio is low enough to make one pool of collateral more efficient than two. No live cat bond has tested it. The notes run to early June 2028.

Pricing a Peril With No Loss Distribution

The combined modeled expected loss is 2.16% and investors are paid a 6% spread, about two and a half times the modeled rate. That multiple is the market's price for uncertainty about the model, not for the loss the model produced.

The asymmetry between the two perils explains the size of it. Wind is modeled from decades of storm track and structural vulnerability data, so a buyer has real confidence in the shape of the distribution even where the point estimate is uncertain. A scenario terrorism model is only as good as its assumptions about who attacks what with which weapon, and those assumptions cannot be back-tested against a record the way hurricane frequency can. The spread is the gap between the model producing 2.16% and anyone believing the tail is bounded there.

The trigger is the more consequential design choice. Galileo Re pays on an industry-loss basis, drawing on a third-party index of market-wide loss rather than on AXA XL's own indemnified claims. Every prior terrorism cat bond used indemnity. An index settles faster, because it does not wait for a claims department to adjust and finalize, and it introduces basis risk: AXA XL's own losses could run above or below its pro-rata share of the index, and the sponsor carries the difference.

That choice runs against the market. Indemnity triggers took 81% of H1 2026 risk capital, up from 79% a year earlier, so the index structure sits in a shrinking minority at the moment it is applied to a new peril. For a proof-of-concept this size, settlement speed and modeling simplicity over precision is a defensible trade.

The timing helps explain the appetite on both sides. Guy Carpenter's global property catastrophe rate-on-line index is down roughly 16% for 2026, so sponsors and investors alike have reason to look for risk that does not trade at softening-market spreads, and terrorism is about as uncorrelated with property catastrophe as a peril gets.

The Trigger Depends on the Statute the Bond Replaces

The index is Property Claim Services' Global Terror Index, and a loss only qualifies once the event has been certified as an act of terrorism under TRIA. The bond's payout mechanism is therefore tethered to the federal program it is offered as an alternative to.

TRIA is due to expire in 2027, roughly a year before the June 2028 maturity. If Congress does not reauthorize on schedule, or reauthorizes with a materially different certification process, the legal machinery the trigger depends on can change mid-term. That is legislative basis risk layered on top of the modeling basis risk the index trigger already carries, and it is not a risk the spread was set against any historical distribution to cover.

The precedent from the last peril priced without a loss history is not reassuring on trigger design. The World Bank's pandemic bonds, issued in 2017 with no frequency-severity data to anchor a model, did trigger and pay during COVID-19 in 2020, and were criticized afterward for trigger mechanics that delayed payment exactly when the money was needed. That experience plausibly shaped the preference here for a faster-settling index. It also shows the pattern: the first transaction in a new peril category gets a workable structure to market, and the mechanics get fixed in the second and third deals.

Scale keeps the stakes contained for now. Against TRIA's loss-sharing structure, which engages only after an insurer's losses clear a deductible tied to prior-year premium and the industry aggregate passes roughly $200 million of insured terrorism losses, $67.5 million is a rounding error. What it establishes is that a named sponsor and two established modeling agencies can package the peril credibly enough for capital to take it, which is a different claim from the one TRIA was built on.

AXA XL's Galileo Re Cat Bond Prices the First US Terrorism Risk

Key Takeaways

The First Cat Bond to Carry US Terrorism

Pricing a Peril With No Loss Distribution

The Trigger Depends on the Statute the Bond Replaces

Further Reading

Sources