Key Takeaways
- • Columbia Threadneedle estimates cumulative global AI data center investment will reach $5.8 trillion between 2025 and 2030, with the firm suggesting this figure may be conservative.
- • Tech hyperscalers (Amazon, Alphabet, Meta, Microsoft, Oracle) issued approximately $121 billion in bonds during 2025 alone—more than four times the five-year average—with total hyperscaler debt issuance reaching $244 billion globally.
- • Oracle raised $18 billion in a single bond offering in September 2025 specifically for data center buildouts tied to its OpenAI partnership, exemplifying the scale of capital deployment.
- • TeraWulf, a former Bitcoin miner, issued $3.2 billion in 7.75% senior secured notes maturing in 2030 for data center expansion, with plans for an additional $3.5 billion financing in 2026 to support a Kentucky campus leased to Anthropic.
- • A critical timing mismatch exists between capital expenditure and revenue generation: companies are spending billions now on infrastructure while revenue streams remain years away, creating material credit rating pressure and covenant risk for bond investors.
Summary
The artificial intelligence infrastructure build-out is generating unprecedented capital demand, with Columbia Threadneedle estimating $5.8 trillion in cumulative global investment in AI data centers through 2030. This projection underscores the enormous financial requirements needed to support AI's computational demands for power, cooling, and processing capacity.
Tech hyperscalers have responded to this capital need with aggressive bond issuance. In 2025 alone, Amazon, Alphabet, Meta, Microsoft, and Oracle collectively raised approximately $121 billion in debt—more than quadrupling their five-year average annual issuance. Oracle's September 2025 bond sale of $18 billion marked a significant single transaction, with proceeds earmarked for data center construction tied to its partnership with OpenAI. Globally, hyperscaler debt issuance has reached approximately $244 billion, demonstrating the magnitude of capital mobilization occurring across the sector.
The transition from traditional sectors to AI infrastructure is also reshaping financing patterns among infrastructure companies. TeraWulf, formerly a Bitcoin mining operation, has pivoted to AI data center development and issued $3.2 billion in 7.75% senior secured notes due 2030 to fund expansion. The company plans an additional $3.5 billion financing package in 2026 to support a Kentucky campus facility leased to Anthropic, the AI safety company behind Claude. This shift illustrates how established infrastructure players are repositioning capital deployment toward AI workloads.
A critical structural tension emerges from the timing mismatch between capital expenditure and revenue generation. Companies are deploying billions in 2025-2026 for construction, equipment, and power infrastructure, yet the revenue streams supporting these assets will not materialize for years. This lag creates material credit risk and potential rating pressure as investors assess whether companies can service debt obligations before generating sufficient returns. Analysts are emphasizing rigorous due diligence on bond covenants, underlying credit quality, and the specific revenue assumptions embedded in each issuance, particularly given the substantial sums involved and the nascent nature of AI infrastructure monetization.
Tech hyperscalers have responded to this capital need with aggressive bond issuance. In 2025 alone, Amazon, Alphabet, Meta, Microsoft, and Oracle collectively raised approximately $121 billion in debt—more than quadrupling their five-year average annual issuance. Oracle's September 2025 bond sale of $18 billion marked a significant single transaction, with proceeds earmarked for data center construction tied to its partnership with OpenAI. Globally, hyperscaler debt issuance has reached approximately $244 billion, demonstrating the magnitude of capital mobilization occurring across the sector.
The transition from traditional sectors to AI infrastructure is also reshaping financing patterns among infrastructure companies. TeraWulf, formerly a Bitcoin mining operation, has pivoted to AI data center development and issued $3.2 billion in 7.75% senior secured notes due 2030 to fund expansion. The company plans an additional $3.5 billion financing package in 2026 to support a Kentucky campus facility leased to Anthropic, the AI safety company behind Claude. This shift illustrates how established infrastructure players are repositioning capital deployment toward AI workloads.
A critical structural tension emerges from the timing mismatch between capital expenditure and revenue generation. Companies are deploying billions in 2025-2026 for construction, equipment, and power infrastructure, yet the revenue streams supporting these assets will not materialize for years. This lag creates material credit risk and potential rating pressure as investors assess whether companies can service debt obligations before generating sufficient returns. Analysts are emphasizing rigorous due diligence on bond covenants, underlying credit quality, and the specific revenue assumptions embedded in each issuance, particularly given the substantial sums involved and the nascent nature of AI infrastructure monetization.
Go deeper than the headline
Research every firm in this story, track the ones that matter and get the daily briefing — free for 7 days.