Apollo and SpaceX test the limits of pension fund credulity
By Klaus Berger · Reporting from Frankfurt ·
By leveraging a borderline investment-grade rating to finance rapidly depreciating Nvidia chips, SpaceX shifts the downside of Silicon Valley's infrastructure race onto regulated retirement funds.
When an enterprise asks lenders for $40 billion to buy rapidly depreciating silicon, traditional credit analysis expects a dense prospectus. It should detail cash-flow amortization and collateral salvage value. Instead, prospective lenders received a brief memo. It featured pictures of outer space and an arrow pointing to data centers somewhere in the cosmos. On the tech podcast TBPN, hosts John Coogan and Jordi Hays reported on SpaceX’s pursuit of bank loans and investment-grade bonds. Apollo would steer the debt, with PIMCO in discussions, to finance an enormous order of Nvidia semiconductors. As Hays observed, Silicon Valley treats this casually. Musk has simply “earned the right” to hand institutional committees a short sketch. Trusting his process has paid off before.
Shifting Silicon Depreciation onto Regulated Savers
That blithe dismissal of credit discipline mistreats what fixed-income underwriting actually does. In the ordoliberal tradition of Walter Eucken, the foundational principle of a functional market is liability—Haftung. Whoever commands the investment must bear the downside. What SpaceX and Apollo are engineering is a classic regulatory arbitrage. It is designed to shift capital risk away from equity holders and onto regulated institutions.
SpaceX carries a BBB rating, the second-lowest rung of investment grade. That rating is the magic key. Regulated retirement vehicles, life insurers, and public pension funds operate under strict fiduciary covenants. They are largely barred from speculative junk paper. But a BBB label permits them to allocate billions of civil-servant and pensioner savings. Those institutions are being asked to fund Nvidia’s Vera Rubin architecture. This is an asset class whose economic half-life is remarkably short. By the time a bond matures, the underlying graphics processors are little more than museum pieces. Meanwhile, the debt remains a hard, cash claim on the issuer’s balance sheet.
The Broken Analogy of the Historical Infrastructure Boom
The most capable defenders of this financing argue that transformational infrastructure always requires unconventional underwriting. Lenders who backed early transatlantic telegraph cables or early railway trunk lines took enormous speculative leaps. Those who hesitated missed the industrial transformation. Musk has repeatedly defied traditional automotive and aerospace debt models. He turned seemingly reckless capital expenditure into monopolistic cash generators like Starlink.
Yet history reveals a vital distinction between rights-of-way and computational hardware. When railway developers overleveraged themselves, the roadbeds, bridges, and steel tracks remained permanently in the dirt. They retained residual utility through restructuring. Data center hardware offers no such permanence. High-performance computing clusters burn power and require relentless cooling. They face immediate economic obsolescence the moment a denser, more energy-efficient architecture enters production. Pledging a BBB balance sheet to absorb the depreciation risk of bleeding-edge silicon leaves bondholders underwriting rapid technological churn. They get fixed-coupon yields that offer no equity upside.
The Unforgiving Ledger of the Credit Cycle
On the same broadcast, Coogan cited reflections from artificial-intelligence researcher François Chollet. Chollet noted that verifiable domains like code and mathematics scale at breathtaking speeds. However, the broader economy remains firmly anchored to physical constraints and human data. The financial system is bound by an even stricter anchor. That anchor is the compound math of debt service.
Treating a $40 billion capital call as an act of faith is the classic hallmark of late-cycle euphoria. It should be an exercise in balance-sheet solvency. Apollo and PIMCO market BBB debt stuffed with short-lived computing clusters to municipal pension boards. By doing so, they socialize the downside of private industrial ambition. A brief memo with arrows pointing toward the stars may suffice for venture capitalists playing with surplus liquidity. But pensioners whose capital is quietly pledged to absorb the wreckage of the next chip cycle deserve protection. They deserve the protection of the rulebook.