Goldman Sachs strategist Ryan Hammond estimates that hyperscalers including Amazon, Oracle and Microsoft need about $300 billion in AI revenue over the next few years to break even on their investments; he says they remain well short of that level.
Hammond also estimates AI users would need to spend roughly $1 trillion a year on AI applications for hyperscalers to earn solid returns on their investments and for application providers to achieve strong margins on computing costs.
In Q2 2026, hyperscaler cloud revenue was annualizing about $70 billion above its pre-AI trend, and announced backlogs for the group exceeded $1.5 trillion. The Roundhill Magnificent Seven ETF, which tracks top AI hyperscalers, rose 8% in a month, compared with a modest gain for the S&P 500.
