Nippon Life gives itself ten years to lend $13bn into US data centers

Grow a 1 trillion yen book at 11 percent a year and it doubles in about six and a half years: the log of two over the log of 1.11 is 6.6 ^[1]. Nippon Life has given itself until fiscal 2035 ^[5]. That is ten fiscal years, about $1.3bn a year, for a doubling the most recent year already outran ^[2].

How much of that is new money depends on which sentence of the account you read. Crypto Briefing reports that the insurer plans to allocate roughly 2 trillion yen, approximately $13bn, to infrastructure financing weighted toward US data center construction ^[1]; the same report says the $13bn target amounts to doubling the total project finance balance to 2 trillion yen by fiscal 2035 ^[5]. Under the first reading, $13bn of new commitments. Under the second, the book travels from about $6.8bn to $13bn, so roughly $6.2bn is incremental ^[3]. The 1 trillion yen figure is described as the overseas portfolio and the 2 trillion as the total ^[4][5].

The dollar numbers are a function of the yen as well. Two trillion at $13bn implies about 154 to the dollar; 1 trillion at $6.8bn implies about 147 ^[4]. Nippon Life funds in yen and lends in dollars, exposure the report says hedging can mitigate to varying degrees without putting a number on the cost ^[10]. That cost shows up in the gap between a spread above two points and the realised return ^[3].

Repayment comes out of project cash flow instead of a corporate borrower's balance sheet ^[2], and the cash flow is a long-term power purchase agreement plus a lease from a creditworthy tenant, the match Crypto Briefing describes against insurers' multi-decade liabilities ^[9]. Rising costs for semiconductors and servers have complicated the traditional financing routes and opened room for project-specific institutional loans ^[7]. Roughly 70 percent of the first-half 2025 growth in global data center IT capacity was American ^[6].

Two points on $13bn is more than $260m a year above the reference rate ^[5], and the pitch is explicitly comparative: US project finance pays more than what domestic Japanese alternatives deliver ^[3]. I expect the money to behave like that pitch, arriving while the hedged gap holds and slowing when it closes. The price of these loans then tracks Japanese yields more than American capacity demand. The counter-thesis is duration. An insurer with multi-decade liabilities will pay up for a twenty-year lease because yen paper of that length is scarce ^[9], and on that reading the book keeps growing through a spread compression. What separates the two is a year in which US spreads narrow and the balance rises anyway. Whichever reading is right, at about $1.3bn a year ^[2] this lender is a price taker.