One Gateway or Fifty Servers? What AWS and Google's Opposite MCP Bets Mean for Your Deploy Surface
AWS ships one MCP server with a fixed tool set while Google ships more than fifty, one per service. This post compares both designs on context cost, auth, versioning, and audit, then recommends a concrete tool vocabulary for a deploy surface agents can operate.
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.
When Hetzner Says No: What Happens When Your Bare-Metal Provider Runs Out of Machines, Not Money
On June 26, 2026 Hetzner started refusing new cloud servers for a subset of customers — not a price hike, a stock-out. What that hard capacity ceiling breaks in a Cluster API fleet, and the concrete multi-location, multi-type playbook to keep your PaaS scaling when one location says no.
Hetzner Raised Prices Three Times in 2026 — The RAM Shortage Behind Your Next Bill
Hetzner repriced cloud servers three times in 2026 — 30% in April, then 113–175% on dedicated vCPU in June — driven by a DRAM shortage as HBM for AI accelerators crowds out DDR5. Here is the timeline, the wafer math behind it, and what it does to a Cluster API fleet's real bill.
Hetzner Just Tripled Its Best VMs: What the 2.7x Price Shock Does to Self-Hosting Math
Hetzner raised CCX and CPX cloud servers 2.1x to 2.75x on June 15 on top of April's 37% hike. A line-by-line repricing of the same Node.js plus Postgres workload against Railway, Render, and Fly.io's 2026 meters shows the gap narrowed — but the flat box still wins once egress is counted.
R2 at $0, B2 at $0.01, S3 at $0.09: What Egress Really Costs Your Object Store
The same 500 GB of object storage costs $7.50 on R2, $3.48 on B2, and $11.50 on S3 — but egress flips the bill from $0 to $441 at 10x read volume. A line-by-line recompute across R2, B2, S3, and self-hosted MinIO, plus where a PaaS should draw the line.
Railway's $5 Hobby vs. a €10 Hetzner Box: What a Solo Dev Actually Pays for an API and a Worker
Railway's $5 Hobby plan covers one tiny service — add a background worker and the same always-on workload costs $40.50 in metered RAM, vCPU, and egress against €10 flat on a Hetzner box, but per-minute billing still wins when average utilization drops below 30%.
Railway vs Render vs Fly.io on July 19, 2026: Same Stack, Three Bills, and Where the Meter Moves First
Techsy re-audited Railway, Render, and Fly.io on July 19, 2026 after all three shuffled plan names — here is a line-by-line recompute of an identical Node.js+Postgres workload, where per-service multiplication and egress tilt the cheapest-to-most-expensive gap, and what the bill collapses to on a flat Hetzner CX22.
Your Cloud Account, Their Control Plane: Why Edka's BYO-Hetzner Model Is Winning
Edka hit 437 points on HN spinning up production-ready Kubernetes on Hetzner in ~2 minutes inside your own account — you own the servers, the bill, and the exit. How the BYO-account pattern beats both closed PaaS and DIY, how it compares to Cluster API Provider Hetzner, and what trust actually requires before you hand over an API token.