RAM chip manufacturer SK Hynix says its memory is sold out through 2029, and memory prices nearly doubled in a single quarter this year. So why don’t memory makers just build more factories? Alex from Speedify explains the machine at the center of the RAM crisis: ASML’s EUV lithography systems, which cost up to $430 million, weigh about 150 … Read More

Your GPS doesn’t actually know where you are. GPS navigation knows which signal is loudest, and that’s the opening someone can exploit… even on aircraft. Alex from Speedify explains how GPS works on planes 20,000 kilometers up, why the signal reaching your phone is so faint it sits below the noise floor, and why a delay of a single millisecond … Read More

A PHP software bug sat in the code behind 70% of the websites on the internet for 14 years because every person who looked at the software bug called the flaw too hard to exploit. Alex from Speedify breaks down why PHP, a 1994 personal home page project that critics have called a hack from the start, still holds up … Read More

Last week Ukraine’s military satellite terminals stopped working on anything moving faster than 100 miles an hour, but the outage wasn’t Starlink. Alex from Speedify explains why it was actually Starshield, the separate SpaceX constellation built for the US military, and why that distinction decides who controls the switch. He also gets into the part almost nobody noticed: ordinary Starlink … Read More

QNX runs in 275 million cars, and almost nobody who drives one has heard of it. Alex and Brian get into what it actually took to move Speedify onto a real-time operating system: why “real time” means guaranteed rather than fast, what happens when your process has to yield to the one controlling the brakes, and why porting from Linux … Read More

An old M1 Mac putting out 16 gigabits per second of packets, all built in Rust with LLM help. Alex and Brian from Speedify dig into how to actually get Claude to write the fastest Rust code possible, including why you should build your test harness in a completely separate LLM session (so the LLM doesn’t make the same bugs … Read More