Alphabet’s bond comeback proves hyperscalers can fund AI capex in investment grade—but without covenants, the credit model may stay fragile

Alphabet’s own 2026 guidance frames AI infrastructure scaling as a funding problem solved with multiple sources. In its June 2026 investor materials, the company expects 2026 capex of $180B–$190B and states it will fund investments “in a balanced way, including strong operating cash flow, debt issuances and today’s equity offering.”

This matters for the “funding model” hypothesis: hyperscalers can keep spending (and keep suppliers busy) if capital-market access remains available. Yet the question becomes how they access it—at what risk pricing, with what protections, and with what trigger sensitivity if the capex-to-revenue conversion slows.

In reporting on Alphabet’s bonds, the issue highlighted is not just that debt exists, but that the bonds reportedly do not carry standard protections. Reuters characterizes the “lack of protections” as the missing element investors watch for, including a commonly expected change-in-control covenant.

Financially, the capex ramp is so steep that management leans on capital markets in addition to operating cash. In its June 2026 investor presentation, Alphabet notes it generated $174B operating cash flow over the prior 12 months and ended Q1 2026 with $127B in cash and marketable securities and $81B of debt—while also saying it has issued “approximately an additional $20B” of debt since end of Q1 2026.

The supply-chain implication is direct: hyperscaler AI buildouts aren’t a “small capex adjustment”; they are multi-quarter procurement programs spanning compute, networking, power, data-center construction, and optical/fiber expansion. When funding stays accessible, those supply ramps keep going—unless utilization or margins fail to materialize.

Why would equities question capex while IG investors keep buying? One reason is timing mismatch: debt investors can underwrite near-to-mid-term cash generation capacity (or refinancing ability), while equities price the long-duration outcome (returns on incremental compute, software monetization, and competitive positioning).

When the IG-to-AI-capex channel funds capex without heavy covenants, it can become a bridge: the hyperscaler doesn’t have to convert capex payback into immediate earnings to access capital-market funding.

To test whether IG funding can “crack” like HY funding did for distressed issuers, you have to follow the throughput chain:

Upstream, capex translates into contracts and backlog for data-center construction, power equipment, and networking/compute component demand. Downstream, it translates into more compute available for workloads—ads, cloud services, and AI inference/training consumption.

If compute supply scales faster than workload demand, utilization falls; if workload demand rises but monetization lags (pricing pressure or cost-per-token rising), margins compress; both mechanisms strain credit metrics.

This is why the covenant question matters: without restrictions, management can continue scaling, but the market’s tolerance for that scaling depends on how quickly utilization converts into cash flow.

From financial statements, Alphabet’s operating cash flow and free cash flow remain substantial. In FY 2025, operating cash flow is $164.713B and free cash flow is $73.266B, while capital expenditure is -$91.447B. That gives a concrete base for why bond markets can tolerate capex during ramp phases.

The key is not whether cash flow exists; it’s whether it continues to support the ramp without forcing asset sales or expensive refinancing. That’s where the flexible covenant package can shift risk toward faster spread repricing if the utilization funnel breaks.

But the thesis for a “crack” doesn’t require a downgrade today. It requires a mismatch later:

1) capex continues scaling, 2) utilization/monetization doesn’t keep pace, 3) credit metrics deteriorate, 4) investors demand higher spreads—especially when they can’t rely on covenants for downside protection.

In other words: IG doesn’t collapse; it reprices.

The structural break happens when the AI buildout transitions from a “shortage of supply” framing to a “surplus of capacity” reality. At that point, the market may decide that light covenant packages transfer too much downside to bondholders.

If that occurs, you’ll see refinancing costs rise first, then issuance volumes slow, then procurement schedules tighten—exactly the kind of funding channel contraction that can resemble prior HY episodes, but via IG spread repricing rather than immediate distress.