Understand AI compute
From chips and FLOPS to what compute costs and who pays for it.
Your outcome
Explain what AI compute is measured in, how fast demand has grown, and how the deals behind it compare.
Read the linked section, reflect on the question, then mark the stop complete. Section times estimate 200 words per minute, excluding exercises. This is a guided reading list, not a test or certificate. Progress stays in this browser.
Learn the units
Read how compute is counted: FLOP/s, FLOP, GPU-hours, accelerator counts and megawatts.
About 2 min for the linked section and its subsections
Check your understanding: Which unit sizes a training run, and which sizes a data center?
Tell a rate from a count
Separate FLOPS, a speed, from FLOP, an amount, before comparing any two chips.
About 3 min for the linked section and its subsections
Check your understanding: Why can one chip's rating change with the number format?
See how fast demand grew
Follow the measured growth of frontier training runs and what it implies for hardware.
About 4 min for the linked section and its subsections
Check your understanding: What would a fivefold yearly growth rate mean for a 2030 training run?
Read the deals behind the buildout
Learn what each stated value means, then sort the deals by value or capacity.
About 4 min for the linked section and its subsections
Check your understanding: Why can an investment intention not be ranked against a signed contract?