Episode 24 of Cloud Computing with Fexingo: Lucas and Luna break down the shifting economics of cloud GPU rental in mid-2026. After years of shortages and sky-high prices, the market is turning. Lucas explains how NVIDIA's Blackwell supply ramp, AMD's MI400 launch, and a wave of new AI-specific cloud startups are driving GPU costs down by 30% year-over-year for certain workloads. Luna asks whether the old 'reserved vs on-demand' calculus still applies when hardware generations cycle every 18 months. They look at a real mid-sized AI startup's bill: spending $240,000 a year on A100s versus migrating to newer H200 instances at half the cost per token. The episode also touches on the hidden gotcha — egress fees still eat 10% of total GPU costs if you move training data across regions. Plus a natural chat about why this show stays ad-free and how listener support keeps it that way.

#CloudComputing #GPU #NVIDIA #AMD #AIInfrastructure #CloudPricing #H200 #MI400 #Blackwell #TechPodcast #Business #Technology #StartupCosts #CloudMigration #EgressFees #FexingoBusiness #BusinessPodcast #CloudEconomics

Keep every episode free: buymeacoffee.com/fexingo