The packet.ai blog

Notes from the GPU frontier.

Deep dives on GPU infrastructure, model performance, and the economics of compute, from the team building software-defined GPUs at packet.ai.

All

Announcement

Engineering

Guide

Industry

Infrastructure

Product

Technical

Announcement

Why We Raised $19M to Fix GPU Infrastructure

70% of GPU capacity sits idle. packet.ai raised $19M from Creandum to fix that — and we're just getting started.

packet.ai TeamMarch 19, 2026

Latest

Announcement

Why We Raised $19M to Fix GPU Infrastructure

70% of GPU capacity sits idle. packet.ai raised $19M from Creandum to fix that — and we're just getting started.

packet.ai Team

March 19, 2026

Infrastructure

8 Large Language Models on a Single NVIDIA Blackwell Server

8 LLMs, 8 GPUs, one server. From Qwen 3 30B MoE at 33.4 tok/s to Mistral 123B at 18.5 tok/s — and every model available via Token Factory at $0.10/M tokens.

packet.ai Team

February 13, 2026

Industry

30 Years Ago Today, a Computer Beat the World Chess Champion

Deep Blue beat Kasparov with 11.38 GFLOPS in 1996. A B200 GPU today is 4,500,000× more powerful — and rents for $3.75/hr. Here’s what that compression means.

packet.ai Team

February 10, 2026

Engineering

GPU Snapshots Demystified: What Actually Survives Pod Termination

GPU pod snapshots save configuration, not disk images. Here’s exactly what survives termination and how to architect persistent storage so you stop losing pip packages.

packet.ai Team

February 5, 2025

Engineering

GPU Utilization: The Lie Your Dashboard Tells You

GPU_UTIL in nvidia-smi measures if any kernel is running — not whether your GPU is computing. Here’s why SM Activity is the metric you should actually watch.

packet.ai Team

February 3, 2025

Announcement

Packet.ai + SkyPilot: Run ML Workloads with One Command (Alpha)

Native SkyPilot integration for packet.ai: <code>sky launch --cloud packet --gpus H100:1 train.yaml</code>. Same YAML, multi-cloud fallback, auto-stop, H100 from $0.65/hr.

packet.ai Team

January 31, 2025

The Packet

Infra notes, in your inbox.

Benchmarks, architecture deep-dives, and pricing breakdowns — roughly twice a month. No fluff, unsubscribe anytime.

Submit

January 18, 2025

How Dynamic GPU Placement Enables Lower Prices

GPU pricing assumes 30% utilisation — you pay for the idle 70%. Here’s how dynamic placement achieves 5× better utilisation and why that drives packet.ai’s pricing.

January 15, 2025

Why We Only Use Infrastructure Powered by hosted.ai

Every GPU on packet.ai runs on hosted·ai-powered infrastructure. Not some. All of them. Here’s why that matters for reliability and what it means when something goes wrong.

January 12, 2025

GPU Pricing Models Compared 2026: On-Demand vs Reserved vs Spot

GPU pricing models compared: on-demand, reserved, and spot explained with real break-even math. Know which model fits your workload before committing to hardware.