US Life Insurers' AI-Linked Investments Reach $85.2 Billion, Up 24% in Six Months, as Moody's Warns Its Figure Likely Understates the Exposure

US life insurers have been lending steadily into the artificial intelligence build-out, and a new Moody's Ratings analysis puts a number on it. According to Reinsurance News, which reported the analysis on October 23, 2026, life insurers held $85.2 billion of AI-linked investments at the end of the second quarter of 2026. That is about 1.4% of their cash and invested assets and 2.1% of their fixed-income holdings, up from 1.7% at the end of 2025. It is also 24% higher than the $69 billion held at the end of 2025, a rise over six months. Bloomberg's headline on October 22 carried the same $85 billion figure. An aggregator summary of the report says positions are as of June 30, 2026.

The exposure falls into three groups. The first is debt issued by major technology companies, which stood at $32.2 billion, up 21% from the end of 2025 and equal to about 6.6% of those companies' outstanding debt. Reinsurance News names Oracle, with about $10 billion, and Amazon, with $8.5 billion, and says purchases in the first half of 2026 favoured the higher-rated Alphabet, Amazon and Meta Platforms. The second group, and the fastest growing, is digital-infrastructure financing, at $39.5 billion, up 34%. This funds data-centre buildings, electricity infrastructure and the chips that run AI, including graphics processing units and tensor processing units, and includes asset-backed securities, commercial mortgage-backed securities and other financing. The third group is investments linked to semiconductor manufacturers, which the article names but does not size.

The quality of the newer holdings is the detail to watch. Moody's reports that the below-investment-grade share of the digital-infrastructure financing rose from 4.7% at the end of 2025 to 6.1% in the second quarter of 2026. Below-investment-grade debt, often called junk, pays higher interest because the chance of default is higher. A rising share can mean insurers are accepting more risk in exchange for yield, although the Reinsurance News article does not draw that conclusion; one aggregator summary does, as its own interpretation and not as a Moody's finding.

Moody's is open about the limits of its numbers. It says the total is hard to quantify because of the structures involved: special purpose vehicles, joint ventures, alternative investment funds, commercial mortgage loans and collateralised loan obligations can all channel money into the sector without being labelled as AI investments. As a result, its figures likely underestimate the true exposure. It also points to indirect links, such as guarantees, leases and credit support, through which an insurer can be exposed to a technology company without owning its debt directly. The aggregator summary adds that suppliers of cooling, power and networking equipment, and data-centre real estate, could see credit deterioration if spending slows.

The commitments behind the borrowing are large. Reinsurance News reports that the disclosed future commitments of five major technology companies, Alphabet, Amazon, Meta Platforms, Microsoft and Oracle, are about $2.8 trillion, including leases not yet commenced, purchase agreements and guarantees, compared with $1.2 trillion in 2025, an increase of around 130%. The article also names CoreWeave, OpenAI, Anthropic, NVIDIA, Broadcom and AMD among other participants in the AI sector, and says investment in the global AI ecosystem could total at least $3 trillion over five years, with broader estimates of $3 trillion to $7 trillion. Moody's expects life insurers to remain an important source of funding for the investment-grade debt used to finance these projects.

Moody's lists the risks. Concentration matters because some insurers hold much larger positions than others. Counterparty risks are interconnected, because trouble in one transaction could affect others; examples given include special purpose vehicles that lease equipment to technology companies and chipmakers that finance their own customers. Physical and regulatory constraints, such as limits on electricity supply, permitting requirements and local opposition to data centres, could delay construction, raise financing costs and reduce the cash flow available to repay debt, and the article cites 2026 regulatory developments in New Jersey and Texas. Funding gaps may force companies to raise more debt where planned capital spending exceeds operating cash flow.

On the question of scale, Moody's conclusion is measured: at current levels, AI-related investments remain a relatively modest component of insurers' overall assets. Looking ahead, Reinsurance News reports that it expects AI-related holdings to increase by $50 billion to $100 billion by the end of 2027, if global digital infrastructure spending reaches $1 trillion that year. That would put the exposure at roughly $135 billion to $185 billion on the starting figure, a calculation from the published numbers and not a Moody's forecast of the total.

This analysis concerns insurers' investments, which is a different question from their use of AI in business. A separate Moody's survey of 58 insurers, brokers and reinsurers, reported by Insurance Business on October 18, found that about two-thirds have deployed AI broadly or in several production use cases, and that Moody's expects a neutral-to-positive credit effect from it. The investment-side findings above are about the risk on the balance sheet. Readers who want precise figures should consult the Moody's report itself, since the numbers here come from press summaries of it.