From reviewing specialty cargo filings at Lloyd’s over the past three years, AI hardware insurance represents the fastest path from market gap to structured product we have tracked in the London market. On June 1, 2026, Overhaul and Navium launched the Helix Consortium through Lloyd’s of London, offering the first dedicated cargo insurance product for AI infrastructure: GPU clusters, AI chips, liquid-cooled servers, and networking equipment in transit. The product provides transit limits up to $75 million under a single agreement party, with storage coverage up to $25 million, across all transport modes globally.

The timing is significant. Hyperscaler capital expenditure is projected to reach $700 billion in 2026 across Amazon, Microsoft, Alphabet, and Meta. Gartner forecasts worldwide AI spending at $2.59 trillion for the year, a 47% increase over 2025. That spending translates into physical hardware moving through supply chains: pallets of GPUs worth tens of millions of dollars each, liquid-cooled server racks requiring specialized environmental controls, and networking equipment feeding data centers under construction across six continents. Traditional cargo insurance was not built for shipments of this value, concentration, or sensitivity. The Helix Consortium is the market’s first attempt to close that gap.

What the Helix Consortium covers

The consortium is structured as a collaboration among three parties, each contributing a distinct capability. Overhaul provides its cargo risk management technology platform, which the company reports protects $1.4 trillion in cargo value across 150-plus countries and processes billions of IoT events globally. Navium, a specialist cargo insurance underwriter and Lloyd’s coverholder, leads the underwriting. The Fidelis Partnership (TFP) participates through its Lloyd’s Syndicates 3123 and 2126, with ten additional Lloyd’s syndicates also on the consortium.

The coverage specifics distinguish this from standard marine cargo policies in several ways.

Feature Helix Consortium Standard Cargo Policy
Transit limit Up to $75 million, single agreement Typically $5M-$25M, multi-party
Storage limit Up to $25 million Often sublimited or excluded
Transport modes Road, rail, ocean, air (all modes) Mode-specific endorsements common
Monitoring IoT label tracker per pallet, 24/7 SOC GPS on container, periodic check-in
Onboarding White-glove setup in 15 days 30-60 day underwriting cycle
Incident response 24/7 Global Security Operations Center Carrier claims team during business hours

The single-agreement structure is particularly notable. Standard high-value cargo shipments often require separate policies from multiple underwriters, each covering a portion of the total value. The Helix Consortium consolidates coverage under one agreement party, reducing documentation friction and eliminating the coverage gaps that emerge when multiple policies interact.

“For the first time, shippers, brokers, and underwriters have a solution truly reflecting AI infrastructure cargo risk profiles,” said Ronald Greene, Overhaul’s EVP. The official launch event is scheduled for June 25 at Navium’s London headquarters, with the consortium already accepting broker inquiries.

The cargo risk profile of AI infrastructure

AI hardware creates a cargo risk profile unlike any asset class that specialty markets have priced before. The core challenge is the convergence of extreme per-unit value, environmental sensitivity, and concentrated supply chains.

Value density. A single pallet of NVIDIA H100 or Blackwell GPUs can carry a value exceeding $10 million. A full container load of AI servers may represent $50 million or more. For comparison, a standard container of consumer electronics typically carries $2 million to $5 million in value. The value-per-cubic-foot ratio for AI hardware exceeds almost every other commodity in transit, including pharmaceuticals and precious metals.

Environmental sensitivity. Liquid-cooled server racks, increasingly the standard for high-density AI compute, require precise temperature and humidity controls during transport. A temperature excursion that would be irrelevant for a container of smartphones can destroy the thermal interface materials in a liquid-cooled GPU cluster. Vibration thresholds are lower than for conventional electronics. The environmental monitoring requirements alone justify a dedicated product class.

Theft targeting. The value density makes AI hardware a priority target for organized cargo theft. Federal prosecutors in March 2026 unsealed an indictment alleging the smuggling of approximately $2.5 billion in Supermicro servers containing restricted NVIDIA GPUs to Chinese buyers. At least $510 million in hardware reportedly reached its destination before the scheme was disrupted. The conspirators allegedly used hair dryers to transfer serial numbers between real and dummy servers. This case illustrates both the theft incentive (individual servers worth tens of thousands of dollars each) and the sophistication of the criminal networks operating in this space.

Supply chain chokepoints. NVIDIA controls approximately 90% of the market for AI training semiconductors. TSMC fabricates nearly all advanced AI chips. The supply chain runs through a small number of facilities in Taiwan, followed by assembly in a handful of locations across Asia. A single disruption at any chokepoint can halt the entire downstream shipment pipeline, creating correlated loss exposures that traditional cargo policies are not structured to aggregate.

The addressable market: why $75 million limits matter

The scale of AI infrastructure spending in 2026 contextualizes why this product emerged now rather than two years ago.

Gartner’s May 2026 forecast puts worldwide AI spending at $2.59 trillion, with AI infrastructure alone accounting for over $401 billion. Within that infrastructure segment, spending on AI-optimized servers is the fastest-growing subsegment, projected to triple over five years. The four largest hyperscalers (Amazon, Microsoft, Alphabet, and Meta) are on track to spend a combined $700 billion in capital expenditure in 2026, up 77% from the prior year, with roughly 75% directed at AI infrastructure: servers, GPUs, data centers, and supporting equipment.

IDC forecasts data center semiconductor revenue at $477.1 billion in 2026, growing to $843.2 billion by 2030. Every dollar of that revenue eventually translates into physical hardware that must be manufactured, assembled, shipped, and installed. The logistics chain connecting TSMC’s fabs in Taiwan to hyperscaler data centers in Virginia, Oregon, Singapore, and Dublin is one of the highest-value supply chains ever constructed.

The global maritime cargo insurance market was valued at approximately $22.4 billion in 2025, projected to grow at 5.84% annually through 2032 (MarkNtel Advisors). The broader cargo transportation insurance market, including inland and air transit, stood at $59.76 billion. AI hardware represents a new premium pool within these markets, and the $75 million transit limit signals that the Helix Consortium expects individual shipments to approach that ceiling regularly. For context, most standard marine cargo policies cap at $5 million to $25 million per conveyance.

Actuarial pricing without loss history

The fundamental actuarial challenge with the Helix Consortium is that the product insures an asset class with virtually no credible loss history. GPU clusters at their current scale barely existed five years ago. The traditional actuarial pricing toolkit, which relies on loss triangles, development factors, and credibility-weighted experience, fails when the underlying data simply does not exist.

This is not the first time the specialty market has confronted this problem. We have tracked similar dynamics in Corgi’s approach to pricing AI liability coverage and in the emerging standalone AI liability market following ISO CG 40 47 exclusions. In each case, the absence of historical loss data forces underwriters to rely on alternative pricing methodologies.

Exposure rating from analogous classes. The most likely pricing foundation uses historical loss experience from adjacent cargo classes: high-value electronics, pharmaceutical shipments, and semiconductor wafer transport. Each shares some characteristics with AI hardware (value density, environmental sensitivity, theft exposure) but none is a precise analog. The credibility weighting assigned to these analogs is itself a judgment call without empirical support.

Engineering-informed frequency estimation. Overhaul’s platform provides a data advantage here. The company reports a 98% disruption prevention rate across monitored shipments and processes vast IoT telemetry data from shipments across 150 countries. If the consortium can demonstrate that real-time monitoring reduces loss frequency to near-zero for compliant shipments, the pricing model shifts from historical loss experience to a technology-enabled prevention framework. This parallels the pricing evolution in commercial fleet telematics, where real-time monitoring data eventually supplanted traditional actuarial experience rating.

Severity distribution uncertainty. Even if frequency can be estimated through monitoring data, the severity distribution presents acute challenges. A total loss on a $50 million AI server shipment produces a severity outcome that dominates the loss experience for years. With no credible tail data, the choice of severity distribution (lognormal, Pareto, or a mixed distribution with a mass point at total loss) introduces significant parameter uncertainty. The 98th-percentile loss estimate could easily vary by a factor of three depending on distributional assumptions.

Correlated accumulation risk. The supply chain concentration creates a correlation problem that is unusual in cargo insurance. A single incident at a key logistics hub, whether a warehouse fire in a Taiwan semiconductor district or a vessel grounding on a major shipping lane, could trigger multiple policy responses simultaneously. Traditional cargo portfolios assume independence across shipments; AI hardware’s concentrated supply chain violates that assumption. Actuaries pricing this book must model accumulation scenarios that look more like catastrophe exposure than standard marine cargo.

Lloyd’s innovation precedent: from cyber to cargo AI

The Helix Consortium fits a pattern that Lloyd’s has repeated across multiple emerging risk classes over the past two decades. The London market’s consortium structure, where multiple syndicates pool capacity behind a specialist underwriter, has historically been the incubation mechanism for risk classes that eventually become mainstream.

Cyber insurance (2010s). Before 2016, Lloyd’s annual reports referred to cyber as a “newer or less well understood sector,” noting that many insureds were first-time buyers. Today, Lloyd’s hosts 77 cyber risk insurers, and the global cyber insurance market has grown from negligible premiums to roughly $12 billion annually, with projections reaching $60 billion within the decade. The early cyber products at Lloyd’s shared the same characteristics as Helix: no loss history, concentrated exposures, and a pricing methodology that relied heavily on engineering assessments rather than actuarial triangles.

Satellite and space (1960s onward). Lloyd’s pioneered satellite launch insurance when the commercial space industry was embryonic. Early policies were priced almost entirely on engineering risk assessments, with actuarial experience accumulating over decades of launches. The satellite market proved that Lloyd’s syndication model could absorb the extreme value concentration (a single satellite loss could exceed $300 million) that standard markets could not.

Data center construction. The Fidelis Partnership, one of the Helix Consortium’s founding members, already operates a Data Center Construction Consortium through Lloyd’s, having committed $1.6 billion in data center capacity during 2025. That consortium covers construction risks across 12 lines of business for AI data center projects. The Helix Consortium extends coverage from the construction phase into the operational supply chain, creating a lifecycle coverage arc from site construction through hardware transit and installation.

Michael Davern of TFP explicitly positioned Helix as a complement: “The Helix Consortium complements our existing data centre Construction Consortium across 12 lines of business.” This suggests a deliberate strategy to build an integrated AI infrastructure insurance ecosystem at Lloyd’s, covering construction, transit, and eventually operational risks under related consortium structures.

Supply chain concentration as a systemic risk

The supply chain that produces and distributes AI hardware is among the most concentrated in any major industry. This concentration creates systemic risk characteristics that actuaries pricing the Helix Consortium must account for.

NVIDIA holds approximately 90% market share in AI training semiconductors. TSMC fabricates nearly all cutting-edge AI chips on a single island. Assembly is concentrated in a handful of facilities across Taiwan, Malaysia, and Vietnam. Logistics corridors funnel through a small number of ports and airports. A disruption at any single node can cascade through the entire chain.

For cargo insurers, this concentration means that independent loss assumptions break down. Consider a scenario: a typhoon disrupts operations at a major Taiwanese port for two weeks. Every AI server shipment routed through that port experiences delay, potential damage from improvised storage, and increased theft exposure as hardware sits in unsecured holding areas. A single weather event could trigger dozens of claims simultaneously across the Helix book.

The $2.5 billion Supermicro smuggling case illustrates the criminal dimension of this concentration risk. When a small number of suppliers produce hardware worth millions per unit, the incentive for organized diversion is enormous. Export control violations add a regulatory dimension: hardware that crosses borders illegally may be uninsurable, creating coverage disputes that standard cargo policies were never designed to adjudicate.

Overhaul’s IoT monitoring platform is designed to mitigate these concentration risks through per-pallet tracking, route deviation alerts, and unauthorized stop detection. The company’s 24/7 Global Security Operations Centers provide real-time intervention capability. Danielle Basstoe of Navium framed the integration directly: “Advanced cargo monitoring combined with specialist underwriting creates a smarter, more resilient framework protecting the global AI supply chain.” The question for actuaries is whether monitoring technology can sufficiently reduce the tail risk in a supply chain this concentrated, or whether the systemic exposure requires dedicated catastrophe-style loadings.

Why this matters for actuaries

The Helix Consortium creates implications across several actuarial practice areas.

Specialty pricing actuaries will watch how the consortium builds its pricing model over the first 12 to 24 months of operation. The product is a live case study in exposure rating for a novel risk class. The interplay between IoT-derived frequency data and engineering-assessed severity distributions represents a pricing methodology that could apply to other emerging technology risks. As AI hardware shipment volumes grow, the loss experience that accumulates on this book will become the foundation for future pricing across the specialty market.

Reinsurance actuaries should note the accumulation exposure. A $75 million transit limit on a single shipment, combined with the supply chain concentration described above, creates a portfolio accumulation profile that resembles property catastrophe more than traditional marine cargo. Retrocession markets pricing AI hardware cargo exposure will need to develop scenarios that account for correlated losses across the supply chain. The Gallagher Re specialty market analysis has documented similar challenges in pricing correlated specialty risks.

Reserving actuaries face the bootstrapping problem common to new product lines: with no development triangle history, initial reserve estimates must rely on benchmarks from adjacent classes and expected loss ratios set by the underwriting team. IBNR estimation in the first few years will be almost entirely judgment-driven. As the book matures, the transition from judgment-based to experience-based reserves will require careful attention to development patterns that may not resemble any existing cargo book.

Enterprise risk actuaries at carriers and reinsurers participating in the consortium should consider AI hardware cargo as a new accumulation zone in their aggregate risk models. If a carrier already writes data center construction (through TFP’s existing consortium), technology E&O, and now AI cargo transit, the correlation across these coverages during a supply chain disruption event could produce aggregate losses well beyond what independent line-by-line models predict.

Looking ahead

The Helix Consortium’s launch at Lloyd’s signals a broader trend: the insurance industry productizing coverage around the AI infrastructure buildout in real time. The progression from TFP’s Data Center Construction Consortium to the Helix Consortium’s cargo transit product suggests that operational AI coverage (protecting AI systems once deployed) may be the next product development target.

Several developments will determine whether AI cargo becomes a major specialty class or remains a niche product. If hyperscaler capex continues at projected levels through 2028, the volume of high-value AI hardware in transit will create a premium pool large enough to attract additional capacity beyond the initial consortium. If Overhaul’s monitoring platform can demonstrate a sustained disruption prevention rate comparable to its current 98% benchmark, the pricing model may converge toward technology-discounted rates rather than traditional experience rating.

The loss experience that accumulates on this first-generation product will shape how the entire specialty market prices AI infrastructure risk for the next decade. For actuaries working in specialty lines, cargo, or emerging risk, the Helix Consortium is worth tracking from its earliest filings.

Further Reading

Sources