Blog
Insights, analysis, and updates from the AI agent economy. Browse by tag · Browse the archive.
When us-east-1 Goes Dark: What One AWS Region Under Your Whole Stack Really Costs
AWS us-east-1's 15-hour DynamoDB DNS outage in October 2025 broke Ring, Reddit and United check-in at once. A worksheet prices that failure for small, mid and large teams, maps how hosted PaaS silently inherits us-east-1, and compares the bill to a three-location Hetzner fleet under Cluster API.
Your PaaS Bill Lied to You: The Real Cost of Railway, Render, Fly.io and Vercel in 2026
Tural Allahverdiyev's June 2026 teardown put Railway, Render, Fly.io, Vercel and Cloudflare Workers on one page after four Vercel repricings and Heroku's sustaining mode — then priced the same app on a flat Hetzner box. A line-by-line recompute with egress math and the bandwidth number that actually drives the bill.
Who Backs Up the Management Cluster? Rebuilding the CAPI Brain When It Dies
Your workload clusters keep serving traffic when the management cluster dies — but nothing can scale, heal, or upgrade until you rebuild the brain. A line-by-line walkthrough of three recovery paths with concrete commands, the stale-state trap, and a quarterly drill to prove it works.
vLLM vs Ollama in Production: What PagedAttention's 19x Throughput Gap Really Buys on Owned GPUs
Red Hat's 2026 benchmark put vLLM at 793 tokens per second and Ollama at 41 on the same GPU and model — a 19x gap. The same comparison at one concurrent user shows them within 20%. A concrete breakdown of where PagedAttention earns that gap, where it doesn't, and what each engine costs to run on an owned Hetzner GPU fleet.
Vercel Repriced Four Times in 20 Months: What Your Next.js Bill Actually Costs Now
Vercel changed pricing four times in 20 months, Render cut included bandwidth 100GB to 5GB, and Fly.io added two new billing lines. A timeline and line-by-line cost for the same Next.js app on each platform versus a flat Hetzner box.
Vercel Repriced Four Times in 20 Months — Here's What Each One Did to Your Bill
Vercel repriced four times in 20 months — unbundled bandwidth, Active CPU, credit billing, and Turbo builds — while Render cut egress 100GB to 5GB and Fly.io added new meters. A timeline plus a line-by-line recompute of the same Next.js app under each regime, and why a flat Hetzner invoice is the only one that didn't need a changelog.
The Idle-Compute Tax: What Vercel's Active CPU and Netlify's Durable Functions Reveal About Serverless AI Bills
Vercel's Active CPU pricing and Netlify's Durable Functions both fix serverless billing for AI workloads that idle 90% of wall-clock time — a worked cost recompute shows the idle tax, the fix, and what the same workload costs on a flat Hetzner box that never metered the wait.
Snap Cut Caching Costs 60% With Valkey: What the Number Really Means When You Self-Host It
Snap cut its caching bill from $2.1M to $840K migrating 70% of Redis clusters to Valkey on ElastiCache. A line-by-line teardown of the three stacked discounts behind that 60% drop and what the same workload costs self-hosted on owned Hetzner hardware.
Render's $1.5B Valuation Hides a $0.15/GB Question: What 100% Growth Doesn't Tell You About Your Next Bill
Render hit a $1.5B valuation on 100%+ growth the same quarter it cut included bandwidth to 5GB Hobby and 25GB Pro at $0.15/GB. A line-by-line sweep from 25GB to 2TB shows what that actually costs versus a flat Hetzner box — and why valuation is the wrong signal for your next bill.
Subscribe
New posts land in your reader as soon as they publish. Pick a format — all three carry the same posts.
Current feeds keep roughly two days of posts so daily polling does not miss a burst. Older entries stay reachable from the feed's next-page link in readers that follow it, or from the blog archive.
Following one topic instead? Browse tags