Real share gains from Intel. Real share losses to Nvidia. The middle ground has a name: 50¢ gross margin.
↓ scroll to read
✦ The bottom line
AMD has taken share from Intel for 11 years running. Against Nvidia, the story is different — they're competing on price, not pricing power.
↓ the brief below
✦ Teach me
Pricing power
Companies with a real moat can charge more than their costs without losing customers. The simplest measure: gross margin — how many cents of every revenue dollar are left after the direct cost of making the chip.
In chips, gross margin is the cleanest single signal of where you sit in the food chain. 75¢ = your customers have nowhere else to go. 50¢ = you're competing on price.
Wall Street calls this
Gross margin / pricing power
AMD's 50¢ gross margin is *fine* in absolute terms — it's better than Intel and most of the industry. But against Nvidia's 75¢, it tells you that AMD is the *value alternative,* not the must-have.
Cents kept per dollar of sales · full year 2025
50
¢
Up from 49¢ in 2024 — small improvement despite a $440M hit from US export controls on the MI308 China-specific chip. Healthy. But not Nvidia healthy.
50¢ is a good gross margin in chips — most consumer-electronics companies don't get there. But in AI accelerators specifically, gross margin is how you tell whether you have the must-buy product or the also-acceptable product. So how does AMD's 50¢ stack up against the company they're trying to catch — and the company they've already beaten?
Gross margin · the three big US chip rivals
Cents kept per dollar of sales — most recent reported period
Nvidia
75¢
AMD
50¢
Intel
32¢
Nvidia keeps 75¢; AMD 50¢; Intel 32¢. The order matches the AI accelerator pecking order. Nvidia's CUDA software lock-in is the why. AMD's ROCm software is catching up but isn't there yet.