Episode
How Cloud Bills Are Adding AI Inference Surcharges
- Published
- Jun 26, 2026
- Duration seconds
- 412
- Processing state
not_requested
Actions
POST https://stenobird.com/v1/public/podcasts/cloud-computing-with-fexingo-aws-azure-gcp-and-modern-infrastructure-conversations-7871918/episodes/how-cloud-bills-are-adding-ai-inference-surcharges/transcription-requests
Idempotently request low-priority transcript generation for this episode.GET https://stenobird.com/podcast/cloud-computing-with-fexingo-aws-azure-gcp-and-modern-infrastructure-conversations-7871918/how-cloud-bills-are-adding-ai-inference-surcharges.md
Read the agent-friendly Markdown representation of this episode resource.
Summary
In this episode, Lucas and Luna dig into a new line item showing up on enterprise cloud invoices: the AI inference surcharge. Amazon, Microsoft, and Google are now charging a premium per million tokens when customers use their managed inference APIs on AWS Bedrock, Azure OpenAI Service, and Vertex AI. The hosts break down why the premium ranges from 30% to 120% above base compute costs, how it's tied to NVIDIA's H100 GPU scarcity and the cost of high-bandwidth memory, and what it means for startups building AI features. They also discuss the fine print: some providers waive the surcharge if you commit to reserved GPU instances, while others levy it even on spot usage. Real numbers from a real bill show a 47% increase in monthly spend for a mid-stage startup. This episode is a practical guide to understanding and negotiating the newest cloud cost. #AIInference #CloudCosts #AWSBedrock #AzureOpenAI #VertexAI #NVIDIAH100 #GPUScarcity #EnterpriseBilling #CloudPricing #Technology #TechPodcast #FexingoBusiness #BusinessPodcast #LucasAndLuna #CloudComputing #InferenceSurcharge #TokenPricing #StartupCosts Keep every episode free: buymeacoffee.com/fexingo