MP Marc Popeinfra · open source
sys: online · EST Let's talk
the log · 159 entries

Log-scale notes on infrastructure.

Analysis on backup, cloud, open source, security and the companies shaping them. One entry a day, filtered through twenty-five years of running the systems involved.

RSS
2026.07.127 min read Leanstral 1.5 Reportedly Found Five Bugs Across 57 Repositories Mistral reports that its open-source Lean 4 model found five previously unreported bugs, including a Rust integer overflow that panics in debug builds and produces incorrect output in… programmingai-machine-learningsoftware-development 2026.07.117 min read GLM-5.2 Makes Open-Weight AI a Stronger Option for Production Coding GLM-5.2's reported coding scores sit close to Claude Opus 4.8, with lower API prices and MIT-licensed weights. The production decision still depends on workload tests, hardware costs, and… programmingtechnologyai-machine-learning 2026.07.107 min read GPT-5.6: Choosing Between Sol, Terra, and Luna in Production GPT-5.6 reached general availability on July 9 with three pricing tiers. Shared context limits make routing easier, while explicit caching changes the cost of repeated prompts. programmingtechnologyai-engineering 2026.07.096 min read Samsung's 1,810% Profit Forecast Shows the Pressure on AI Memory Supply Samsung's Q2 2026 forecast points to pressure on HBM supply. That constraint affects GPU delivery schedules, conventional memory prices, and infrastructure planning through 2027. technologyai-engineeringinfrastructure 2026.07.086 min read JADEPUFFER: Sysdig Reports an Autonomous AI-Driven Ransomware Attack Sysdig says an LLM agent exploited an unpatched Langflow server, moved into production systems, and encrypted 1,342 Nacos records without human tactical input. technologyai-engineeringinfrastructure 2026.07.077 min read Together AI's $800 Million Raise Highlights the Cost of Closed-Model APIs Together AI reports $1.15 billion in annual bookings as it raises $800 million. Its pitch is lower-cost open-weight inference, though the savings depend on the workload and the… technologyai-engineeringinfrastructure 2026.07.067 min read Microsoft's 2029 Post-Quantum Deadline Gives Infrastructure Teams Three Years to Prepare Microsoft has moved its quantum-safe migration deadline to 2029. Infrastructure teams need to locate vulnerable cryptography, plan for algorithm changes, and account for data that must stay… programmingtechnologyinfrastructure 2026.07.057 min read Bad Epoll: A Linux Root Exploit Missed in AI-Assisted Kernel Review Bad Epoll has a reported 99% exploit success rate on tested systems. Its disclosure highlights two gaps: incomplete review of nearby kernel code and delays between upstream fixes and… programmingtechnologyinfrastructure 2026.07.048 min read Microsoft's $2.5 Billion Plan to Help Enterprises Put AI to Work Microsoft's Frontier Company pairs a $2.5 billion commitment with about 6,000 engineers and specialists. Its deployment model addresses the integration work that keeps many enterprise AI… technologyai-engineeringtechnology-leadership
open channel

Have an infrastructure problem that needs an owner?

Backup, recovery, remote access, fleet automation, or a platform that has to stay up. Tell me what is breaking.

Start a conversation