NavyaAI Research
AI economics, video costs, and infrastructure research.
Practical reports for teams building, buying, and optimizing AI systems, video pipelines, and infrastructure at production scale.
August 2026 - Edge AI
Jetson Edge Inference Benchmark
Measured LLM benchmarks on a $499 Jetson Orin Nano: 158 tokens/sec at 16 users, sub-100ms TTFT, the ~24M tokens/month API crossover — failures included, raw CSVs shipped.
View reportJune 2026 - Cloud cost
Cloud Egress Report 2026
The hidden transfer tax: NAT, cross-AZ, cross-region, CDN, and Supabase realtime fanout — and how to cut it 20–80%.
View reportMay 2026 - Video economics
AI Video Pricing Report
Gemini Omni Flash, Sora 2, Veo 3.1, and the per-second cost model for production AI video.
View reportMay 2026 - AI economics
AI Economics Report 2026
Token price collapse, AI capex, provider margins, hidden costs, and sustainability risk.
View reportToken pricing - AI bills
AI Cost Report
Why token prices fell 99.7% while enterprise AI spend still tripled.
View report