Nvidia spent the week rolling out AI on three fronts at once — onto consumers’ desks, into the software that designs chips, and across the optical plumbing that links data centers — a reminder that the company’s ambitions now stretch well beyond selling the processors that power AI.
The highest-profile move came Friday with the global launch of RTX Spark notebooks, which Nvidia says can run AI applications directly on the machine. The devices reach shelves through ASUS, Dell, HP, Lenovo, Microsoft and MSI, with entry pricing set at $2,599. According to Nvidia, the systems carry up to 128 GB of shared memory and deliver as much as 1 Petaflop of AI compute in the FP4 format.
Those are manufacturer specifications, not a guarantee that every workload will run at that speed. The real selling point is local execution of AI — keeping processing on the device rather than in the cloud — which puts as much weight on how reliably software works with the new platform as on raw graphics horsepower. Early reviews offered a mixed picture: coverage on Thursday praised the Blackwell GPUs and their local-AI capabilities but flagged weaker results from the main processor and gaps in software compatibility. In one reported test, OpenClaw failed to connect to its gateway.
Software aimed at the fab
Two days earlier, Nvidia had taken the wraps off cuPotential, a new library, and reported that semiconductor companies are already using its CUDA-X libraries for AI work in chip design and manufacturing. The company put the speedup in early-access testing at a factor of two to three.
That figure is a technical benchmark first and foremost, and it does not translate directly into revenue. The strategic weight lies in where the software is applied: at the point where chips are conceived and built, extending Nvidia’s reach past the operation of finished AI systems. Whether it becomes a meaningful additional growth driver depends on how widely the tools are adopted.
Should investors sell immediately? Or is it worth buying Nvidia?
Networking forms the third leg. Nvidia and Cisco unveiled plans on October 12 for an optical module form factor called USFP, specifying signals of 400 Gb/s per lane and capacity of up to 12.8 Tb/s. TE Connectivity and Amphenol are contributing to the supporting hardware ecosystem, tying the proposed specification to partners who can build around it. What exists so far is a plan, not a shipped product.
A split screen for investors
All three initiatives point in the same direction — anchoring Nvidia’s technology at multiple points in the AI stack — yet none of them settles the question that has been weighing on chip stocks. On Friday, according to media reports, shares came under pressure from doubts about the durability of AI infrastructure spending and about projects financed with heavy borrowing, with funding stress at CoreWeave drawing particular attention. The selling was not triggered by anything Nvidia-specific, the reports indicated.
That leaves two distinct layers for anyone valuing the stock. On one side sits an expanding product and software lineup, with RTX Spark offering a tangible example of new AI hardware and cuPotential and USFP widening the addressable terrain. On the other sits investor skepticism over whether debt-funded AI investment earns an adequate return.
New use cases strengthen the long-term case, but their economic value will be settled by how far customers and partners actually take them up — and by whether the software holds up in everyday use, not just in peak specs. The first reviews give some grounds for optimism on that score; a uniformly convincing picture has yet to emerge.
Ad
Nvidia Stock: Buy or Sell?! New Nvidia Analysis from October 18 delivers the answer:
The latest Nvidia figures speak for themselves: Urgent action needed for Nvidia investors. Is it worth buying or should you sell? Find out what to do now in the current free analysis from October 18.
Nvidia: Buy or sell? Read more here...
