Investors urged to scrutinize bonds funding AI’s $5.8T data center investment

1 hour ago 1



The AI boom needs power, cooling, and an almost incomprehensible amount of money. Columbia Threadneedle Investments estimates that cumulative global investment in AI data centers will reach approximately $5.8 trillion between 2025 and 2030, a figure the firm itself suggests could be conservative.

A bond issuance spree with few historical parallels

Tech hyperscalers, the usual suspects of Amazon, Alphabet, Meta, Microsoft, and Oracle, have been issuing debt at a pace that makes previous cycles look quaint. In 2025 alone, these companies collectively issued around $121 billion in bonds. That’s more than four times the five-year average.

Oracle alone raised $18 billion in a single bond sale in September 2025, earmarked for data center buildouts tied to its partnership with OpenAI. Total hyperscaler debt issuance recently reached approximately $244 billion globally.

Columbia Threadneedle’s October 2025 analysis describes this as a “multi-year demand shock” for capital.

Former Bitcoin miners and the pivot to AI infrastructure

TeraWulf, a company that built its reputation as a Bitcoin miner, has issued $3.2 billion in 7.75% senior secured notes maturing in 2030 to fund its data center expansion. The company has plans for an additional $3.5 billion financing package in 2026, targeted at a Kentucky campus leased to Anthropic, the AI safety company behind Claude.

The revenue lag problem

The core tension in AI infrastructure financing is timing. Companies are spending billions now on construction, equipment, and power infrastructure. The revenue those assets generate won’t fully materialize for years.

Analysts are urging investors to conduct rigorous due diligence on bond covenants, credit quality, and the specific revenue assumptions underpinning each issuance. The gap between capital expenditure and revenue generation also creates potential credit rating pressure. If companies are burning through cash to build while waiting for AI workloads to ramp, their leverage ratios deteriorate in the interim.

What this means for investors

When $121 billion in new issuance hits the market in a single year, some of those bonds will be rock-solid investment-grade paper from cash-rich hyperscalers. Others will be high-yield instruments from companies making aggressive bets on AI demand that hasn’t fully materialized.

For crypto-native investors watching companies like TeraWulf pivot from mining to AI, the transition raises questions about what these companies actually are. A Bitcoin miner issuing high-yield debt to build AI data centers for Anthropic is a fundamentally different business proposition than it was 18 months ago.

Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.

Read Entire Article