Measure 02 · Pressure
The more of the world's digital operations pile onto a few providers, the larger the damage when any one of them fails. This measure puts a number on that blast radius.
The Herfindahl-Hirschman Index of the Big Three, 2017–2023, with the fitted OLS line (+68.9/yr, R² = 0.991). Because seven data points is too few to trust OLS significance testing, the trend is confirmed with the non-parametric Mann-Kendall test (τ = 1.00, p ≈ 0.003). The hollow 2024 point is excluded from the fit — Synergy reclassified some Azure revenue that year, which lowers the measured share without any real drop in concentration.
Synergy Research Group · OLS + Mann-Kendall / Sen's slope
The stacked areas show each provider's share; the dashed line is their combined total. Notice the rise comes almost entirely from Azure — but the combined line keeps climbing regardless of how the split shifts between AWS and Azure. That combined exposure is what actually matters for systemic risk.
Synergy Research Group quarterly releases, 2017–2024
Concentration measures fragility at the infrastructure level rather than counting outages directly. Amazon Web Services held roughly steady across the period; the rise came almost entirely from Microsoft Azure doubling its share. That matters for how you read the risk: the danger isn't any one company's dominance, but the share of the world's infrastructure resting on a small handful of them together. The combined top-three share rose steadily no matter how the split shifted between AWS and Azure — which is exactly the exposure that turns one provider's bad day into everyone's.
One honest caveat sits on the last year. In 2024, Synergy reclassified some Azure revenue from infrastructure to software, which makes concentration appear to dip even though Azure's actual revenue kept growing. That year is shown for completeness but left out of the trend estimate.