Hyperscalers are spending twelve dollars on AI infrastructure for every dollar they earn from it. That ratio is embedded in $575 billion of capital expenditure this year, with Amazon, Microsoft, Alphabet, Meta, and Oracle on track to direct 90% of their operating cash flow into AI data centers in 2026, up from a historical average of 40%.
The debt numbers tell the real story. Hyperscaler bond issuance averaged $20 billion annually from 2020 to 2024. It hit $96 billion in 2025 and will reach $159 billion in 2026. Morgan Stanley projects $1.5 trillion over the next few years. Alphabet already issued a century bond, maturing in 2126, the first by a tech company since Motorola in 1997. The original piece walks through the depreciation math that justifies all of it: at 60% gross margins and 5% borrowing costs, a five-year payback on $431 billion in AI capex requires $180 billion in annual revenue. Current AI revenue is $35 billion.
That gap is the bet: 5x revenue growth in five years. It gets worse if Nvidia's annual GPU architecture cycle forces a three-year depreciation schedule instead of five, pushing the required revenue figure to $276 billion, or 7.9x current levels. Read the full piece for the payback period scenarios by revenue trajectory. The numbers are the argument.
[READ ORIGINAL →]