Key Takeaways
- • Big Tech companies are projected to spend approximately $750 billion on AI infrastructure in 2026, with the majority now funded through bond issuance rather than internal cash reserves, marking a structural shift in how the sector finances growth.
- • Meta raised its 2026 capital expenditure guidance to $135 billion while Google issued $55 billion in bonds; Goldman Sachs reports the four largest tech companies have issued over $170 billion in corporate debt, with Wall Street estimating $300 billion in AI-related investment-grade debt supply for 2026.
- • Goldman Sachs analysis reveals Big Tech's free cash flow has deteriorated to its lowest level since the dotcom era, with Amazon's capital expenditure consuming 94% of operating cash flow, creating unsustainable financing dynamics.
- • The 10-year Treasury yield at 4.5% (94th percentile of its 12-month range) directly impacts AI debt servicing costs; the 10Y-2Y spread compressed to 0.29% from 0.74% in February 2026, signaling bond market skepticism about the growth narratives underlying these investments.
- • Investor fatigue is emerging across longer-dated AI debt instruments despite continued appetite, with market spreads widening as high-quality issuers attempt to place increasingly large tranches.
Summary
The aggressive capital expenditure cycle required to compete in artificial intelligence infrastructure has forced Big Tech firms to rely on debt markets at unprecedented scale. According to CNBC's Kate Rooney, the largest technology companies are expected to deploy approximately $750 billion on AI buildout in 2026 alone, a figure that represents a material shift from the self-funded capex model that dominated the sector for the previous decade. Meta exemplifies this trend, raising its 2026 capex guidance to $135 billion, while Alphabet issued $55 billion in bonds to fund comparable infrastructure investments.
Goldman Sachs quantifies the magnitude of this borrowing cycle: the four largest tech companies have collectively issued over $170 billion in corporate debt, with Wall Street analysts projecting $300 billion in AI-related investment-grade debt supply throughout 2026. This volume introduces credit cycle dynamics to what was previously viewed as purely a technology sector narrative. The analyst consensus suggests this AI-related debt could represent 15–20% of most corporate bond indices, indicating that AI financing now constitutes a material component of the broader credit market.
The deterioration in underlying cash generation metrics raises questions about the sustainability of this borrowing trajectory. Goldman Sachs research documents that Big Tech's free cash flow has contracted to its lowest level since the dotcom era, with Amazon's capital expenditure consuming 94% of operating cash flow. This compression leaves minimal margin for error and increases dependence on continuous debt market access, creating potential vulnerabilities if credit conditions tighten.
The Treasury curve has amplified these concerns. With the 10-year Treasury yield at 4.5%—positioned at the 94th percentile of its 12-month range—every basis point increment materially increases the cost of servicing multi-decade bonds issued to fund AI infrastructure with uncertain and distant cash flow horizons. The 10-year minus 2-year spread has compressed to 0.29% from 0.74% in February 2026, a flattening that market participants interpret as the bond market pricing skepticism regarding the growth assumptions embedded in Big Tech's AI investment cases.
Investor appetite for longer-dated AI debt instruments shows nascent signs of fatigue. While institutional demand persists, spreads are widening as issuers attempt to place larger tranches, forcing even high-quality borrowers to pay wider yield premiums. This dynamic suggests the market is beginning to test whether the current volume of AI-related issuance can be absorbed without material repricing of risk.
Goldman Sachs quantifies the magnitude of this borrowing cycle: the four largest tech companies have collectively issued over $170 billion in corporate debt, with Wall Street analysts projecting $300 billion in AI-related investment-grade debt supply throughout 2026. This volume introduces credit cycle dynamics to what was previously viewed as purely a technology sector narrative. The analyst consensus suggests this AI-related debt could represent 15–20% of most corporate bond indices, indicating that AI financing now constitutes a material component of the broader credit market.
The deterioration in underlying cash generation metrics raises questions about the sustainability of this borrowing trajectory. Goldman Sachs research documents that Big Tech's free cash flow has contracted to its lowest level since the dotcom era, with Amazon's capital expenditure consuming 94% of operating cash flow. This compression leaves minimal margin for error and increases dependence on continuous debt market access, creating potential vulnerabilities if credit conditions tighten.
The Treasury curve has amplified these concerns. With the 10-year Treasury yield at 4.5%—positioned at the 94th percentile of its 12-month range—every basis point increment materially increases the cost of servicing multi-decade bonds issued to fund AI infrastructure with uncertain and distant cash flow horizons. The 10-year minus 2-year spread has compressed to 0.29% from 0.74% in February 2026, a flattening that market participants interpret as the bond market pricing skepticism regarding the growth assumptions embedded in Big Tech's AI investment cases.
Investor appetite for longer-dated AI debt instruments shows nascent signs of fatigue. While institutional demand persists, spreads are widening as issuers attempt to place larger tranches, forcing even high-quality borrowers to pay wider yield premiums. This dynamic suggests the market is beginning to test whether the current volume of AI-related issuance can be absorbed without material repricing of risk.
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.