What is AMD?
AMD is a semiconductor company that produces CPUs and GPUs, including the first devices to ship with HBM memory. The material tracks AMD’s role in industry-standard architecture development, its competitive performance versus Nvidia, and its engagement with startups.
Company timeline
- Dec 2025 - Zain Asgar said that on the training side, heterogeneity usually means having either an Nvidia cluster or an AMD cluster, not mixing resources.
- Dec 2025 - Zain Asgar noted that compared to Nvidia’s B40 or B200, AMD chips show significant performance gains, with 20-40% improvements in many cases.
- May 2026 - Gavin Baker said AMD, along with Microsoft and Meta, has essentially zero engagement with startups.
- Aug 2026 - Jon Y said that AMD approached SK hynix to co-produce a new industry standard architecture.
Where it appears in the record
Every line below is attributed to a named speaker.
HBM was originated by AMD approaching SK hynix, not by Nvidia, despite Nvidia dominating the AI GPU narrative today.
“This led AMD to approach SK hynix to co-produce a new industry standard architecture.”Jon Y · 13 Aug 2026
Kernel-level optimization yields only single-digit percentage gains on H100s (already heavily optimized) but 20 to 40 percent improvements on B40/B200, and 2x or more on Mac, AMD, and Intel hardware.
“If you start looking at think is like a B40, or a B200, you'll start seeing like much more significant gains to be had over there. you know, we have seen 20 30 40% improvements in performance in many cases.”Zain Asgar · 2 Dec 2025
AMD, Microsoft, and Meta show essentially zero engagement with AI startups, in sharp contrast to Nvidia, Amazon, and Google.
“Essentially zero engagement with startups from AMD, Microsoft, and Meta.”Gavin Baker · 20 May 2026
Training-side hardware heterogeneity is largely a misnomer. In practice, clusters stay within a single vendor's ecosystem and do not mix across vendors like Nvidia and AMD.
“I think on the training side when people talk about heterogenity it's usually still heterogenity in a homogeneous sense where they're like okay I have an Nvidia cluster or I have an AMD cluster and they don't ever try to mix the resources.”Zain Asgar · 2 Dec 2025