Nvidia (NVDA) is introducing a cut down version of the DGX Spark with half the memory and storage, The Register reports

Newsquawk StaffPublished
Newsquawk headlinesUTC

Every headline is on the live feed 20 minutes before this site.

Nvidia (NVDA) is introducing a cut down version of the DGX Spark with half the memory and storage, The Register reports

A booming gray market of token resellers in China is providing local AI Cos. with access to US models including Anthropic’s Claude, reports The Information

Tesla (TSLA) Q3 total deliveries 486,532 (exp. 456, 896); Q3 total production 464, 391 (exp. 486, 761)

On the Newsquawk feed at , 20 minutes before this page.

Use the PlatformFree. No signup, no card.

Context

A cut-down SKU of an existing product is classic range-filling rather than a demand signal: semiconductor vendors have long introduced lower-memory, lower-price variants to broaden the addressable segment at the bottom of a line while preserving the flagship configuration and pricing. The mechanism to watch is mix, not units: a cheaper variant typically carries lower average selling price but can run at similar or better margin, so the revenue read depends on cannibalisation of the full-spec version versus incremental volume. Note the sourcing: this is a single-outlet media report rather than a company announcement, and in past episodes of this kind the detail has occasionally been revised or corrected once the vendor confirms specifications, so the half-memory framing should be treated as provisional. The tag inclusion of an unrelated diagnostics name appears to be a tagging artefact with no bearing on the story. The follow-ons are any company confirmation, positioning of the variant relative to the existing Spark, and whether channel checks show it displacing full-spec orders or expanding the developer and desktop-AI installed base.

Related headlines

The feed had this first.

Use the Platform