90 days, 5 support cases, zero answers: how Azure handles H100 capacity requests

I run a production GPU workload on Azure. In late June 2026 I opened quota requests for non-Spot H100 (ND/NC H100 v5) in five regions: UK South (2606280050000772), Australia East (2606240050000059), East US (2606280050000769), West Europe (2606280050000770), North Europe (2606280050000771).

Over 90 days later: no approval, no denial with a reason, no ETA, no named owner. What I have instead, in writing from Microsoft:

• "All GPU SKUs are globally capacity constrained" (technical lead, case 2606240050000059), no ETA.
• Support cannot see or reach the capacity decision-makers (support engineer).
• "Severity B is the maximum for a backlogged request" - refusing Severity A on a total production outage.
• Two contradictory promised update dates, both missed.
• Eight different engineers rotating across the cases.

Meanwhile my Spot H100 production VM was evicted five times, four of them within 48 hours.

My question: has anyone actually obtained non-Spot H100 capacity from Azure in 2026? Through which path - support case, capacity reservation, account team, partner? And who at Microsoft actually owns capacity decisions, when support says they cannot reach them?