KM4BI

Nvidia’s $96.2B Quarter, Fake Accounts, and OpenAI’s Chip

The panel examines Nvidia, X’s investigation into roughly 200,000 inauthentic accounts, and OpenAI’s Jalapeno inference chip.

Published

Nvidia reportedly generated $96.2 billion in quarterly revenue as the debate over AI infrastructure intensifies. The speakers discuss both the strength of compute demand and the possibility that some of it could be amplified by financing arrangements connected to data-center deployment.

The episode also reviews an investigation by X’s Global Affairs team, which said it identified a farm of roughly 200,000 inauthentic accounts. As quoted in the discussion, some accounts posted claims that could influence the US debate around AI data centers, including household electricity costs and grid pressure.

OpenAI is another focal point through its custom inference chip, Jalapeno, developed with Broadcom. The performance figures cited in the episode claim 1.5–1.9 times more AI work per watt, up to 3.6 times lower end-to-end latency than Nvidia’s GB200 and GB300, and 700-watt operation versus 1,400 watts for the GB300. The panel sees this as evidence that inference is diversifying beyond Nvidia, even as Nvidia remains central to model training.

The conversation then broadens to AI-enabled surveillance, AI companions, nuclear and solar power, local model execution, preventive health, robotaxis, and space infrastructure.

Source