Explained
What Is Cash Capex? The Cash That Leaves Before AI Hours
Cash capex is cash spent to buy halls and chips before those assets bill. Amazon, Microsoft, Meta and Alphabet — who gets paid, and why hours wait.
Explained
Explained
Mechanism explainers for the AI stack — what a part is, why it binds, who gets paid. Free teaching desk on Second Order.
Get the Wire
Explained
What Is GPU Depreciation? How AI Servers Wear Out on Paper
GPU depreciation is how a company spreads the cost of AI servers over the years it expects them to work. The cash left on day one. The useful life decides how fast that cost hits profit.
Explained
What Is Hybrid Bonding? Copper-to-Copper Joins for AI Stacks
Hybrid bonding stacks dies copper-to-copper. Logic is in volume; HBM4 still ships on microbumps. Who gets paid, and which clock is stuck.
Explained
Who Finances GPU Clusters
Who finances GPU clusters is the capital map on the Neoclouds layer — GPU-backed lenders, OEM lessors, residual offtakers, and customer prepay. Roles only. No tip sheet.
Explained
RPO vs Take-or-Pay: What's the Difference?
RPO is booked cloud not yet recognised as revenue. Take-or-pay bills reserved GPUs after delivery, even idle. Same stuck step: the hall, not the signature.
Explained
GPU Cloud Providers
GPU cloud providers are the roles along the Neoclouds bind — fleet owners, hour sellers, offtake buyers, and power+hall suppliers. Roles only. No tip sheet.
Explained
What Is EUV? Extreme Ultraviolet Lithography for AI Dies
EUV lithography prints the tightest layers on a leading-edge die. Who gets paid when Low-NA scanners are booked, and why High-NA is a later clock.
Explained
What Is a Take-or-Pay GPU Contract? Offtake That Bills Idle Hours
A take-or-pay GPU contract is a committed offtake for reserved accelerator capacity: the customer pays for the booked cluster whether or not every hour is used, and the operator can only start that bill once the contracted capacity is delivered.
Explained
HBM vs DDR5
HBM stacks beside the GPU on advanced packaging; DDR5 sits as host memory on DIMMs. Same DRAM family tree — different stuck step and who gets paid. AI servers usually need both.
Explained
What Is Remaining Performance Obligation? Cloud Backlog That Is Not Cash Yet
Remaining performance obligation is contracted revenue that a cloud operator has not yet recognised: the customer has promised to buy, the operator has promised to deliver, and the cash line waits until the service is performed.
Explained
Who Makes AI Chips
Who makes AI chips is the map of roles along the stack — designers, foundries, packaging attach, and HBM. Roles only. No tip sheet.
Explained
What Is GPU as a Service? Billed Accelerator Hours for AI Work
GPU as a service is billed accelerator time: a cloud operator puts GPUs in a power-ready hall and sells hours — on-demand, reserved, or as a dated capacity block — instead of the customer buying the cards.