What it means
Scale-up and scale-out are the two ways to build a bigger AI machine. Scale-up tightly couples many accelerators inside a single node so they act like one very large chip, using ultra-fast short-range links. Scale-out connects many of those nodes into a cluster over a network fabric, so thousands of chips can train one model together. Both are necessary, and both are bandwidth-hungry: as models grow, more of the engineering — and more of the cost — moves into the links between chips rather than the chips themselves. The networking that stitches a cluster together becomes a first-class part of the system.
Why it matters to investors
As AI clusters grow, a rising share of spend shifts from the accelerators to the fabric that connects them. That is the structural tailwind behind high-speed switching, interconnect silicon and optics — and the reason networking names are an increasingly central, not peripheral, part of the AI trade.
Companies on this part of the chain
Named to show where the term sits in the AI supply chain — research, not advice, and never a recommendation to buy or sell.
Related terms
See Scale-out vs scale-up in the live AI chain.
THE ENTITY maps every constraint onto one live model — which part is tight now, who owns it, and who gets squeezed when it moves. Plain-English reads you can check.
THE ENTITY is an educational read on the AI supply chain — research, not investment advice. It explains how the chain works and who sits where, never price targets or buy/sell calls.