aiqbench analysis
Anthropic details its strategic shift to a heterogeneous compute architecture, leveraging software agents like Mythos to automate kernel porting across AWS Trainium, Google TPUs, and NVIDIA/AMD GPUs. This multi-chip approach effectively decouples model training from specific hardware vendors while introducing "compute multipliers" that significantly increase total effective compute efficiency. The video highlights how Anthropic has built a specialized software stack capable of managing diverse chip families with minimal human intervention, offering a rare glimpse into the operational realities of large-scale LLM deployment.
Anthropic's IPO roadshow video
From Simon Willison
ExcerptAnthropic's IPO roadshow video. Anthropic kicked off their IPO roadshow yesterday, and their roadshow video has some previously unreleased information about how their business works.
The section on "Approach to compute" with co-founder Tom Brown is particularly interesting, describing how Anthropic train and serve Claude across AWS Trainium, Google TPUs, NVIDIA GPUs, and AMD GPUs.
I'm not sure why there's no official video transcript, but here's one that I extracted using MacWhisper.
Excerpt shown under the publisher's RSS terms · 72 words · full text and figures are on the publisher's site
Other coverage of this story
Related coverage

[AINews] Mercor $750m Series D at $20B valuation, $3B run rate
Latent Space · Nov 11, 2026
Anthropic launches IPO roadshow with $64B fundraising goal
SiliconANGLE · Nov 10, 2026

Anthropic launches roadshow for proposed initial public offering
Anthropic · Nov 10, 2026
Ex-Anthropic researcher Jacob Coxon says a roadshow chart of Anthropic's compute multipliers helped drive him out
Dealroom.co · Nov 14, 2026