Nvidia’s shares fell 4.2% yesterday, caught in a broad selloff across AI names as investors revisited their doubts about the scale of artificial intelligence spending and how data centers and chips get financed. The retreat, as AP reported, was a sector-wide repricing rather than a verdict on any single Nvidia announcement. That distinction matters, because the company spent Wednesday rolling out a stack of initiatives that say far more about execution than about demand.
The announcements spanned telecom networks, autonomous driving, drug discovery and manufacturing capacity. Taken together, they sketch a company trying to move from selling chips to keeping AI factories productive — and to securing the physical supply chain that has to deliver them.
Minerva shifts the focus to operations
At the center of the software side sits Minerva, a platform Nvidia unveiled yesterday. It links operating data from infrastructure and AI workloads with purpose-built models designed to flag operational problems and keep AI factories running. For investors, the question is whether the prospect of more usable AI infrastructure justifies the risk of a large capacity expansion. Minerva is, for now, a performance promise rather than evidence of improvements already achieved.
The same test applies to the partnerships. Nvidia reported telecom projects and trials with Nokia and several carriers, among them T-Mobile, Deutsche Telekom, Elisa and BT. According to the company, T-Mobile completed the world’s first AI-RAN call on a live network — a concrete implementation step rather than a statement of intent. Its reach is plainly limited, though: one successful call is not a nationwide rollout.
Beyond networks, Volvo Cars is adopting Nvidia’s Hyperion and Alpamayo Stack for automated driving, and Bolt has chosen Alpamayo Stack for its European robotaxi program. In healthcare and life sciences, Nvidia pointed to an open protein design model developed with the University of Washington’s Institute for Protein Design and work with Siemens Healthineers on adaptive radiation therapy. A wider spread of use cases would support the case that Nvidia technology is finding practical work across industries — but that is a prospect, not a settled result.
The supply chain gets the money
Hard commitments are being made on the hardware side. Together with IBIDEN, Samsung Electro-Mechanics and Unimicron, Nvidia plans to raise substrate capacity 2.5-fold over three years. It has also disclosed a $1.1 billion advance payment to KYEC to reserve capacity for new test facilities in the United States.
Further out, Amkor is targeting more than 300% higher interposer wafer capacity by the end of 2028, and SPIL aims to expand capacity for optical engines 15-fold by mid-2027. Such build-outs can ease supply bottlenecks, but only if technical execution and customer demand move in step.
Should investors sell immediately? Or is it worth buying Nvidia?
The demand picture offers support. Reuters reported Tuesday that ByteDance, Alibaba and Tencent together ordered more than three million Nvidia RTX PRO 5500 graphics cards, with a combined list value above $35 billion. Nvidia could not guarantee full delivery, according to the same report, because demand exceeded planned production. That points to genuine need for extra supply — though the list value is neither realized revenue nor a secured cash inflow.
Prepayments front-load the risk
Here lies the counterargument, and it is one of timing. Nvidia is committing funds and locking in capacity while the eventual utilization of that capacity depends on demand. High demand for compute does not automatically mean every planned investment stays economically viable, and the financing concerns AP described go straight to that link between technical demand and available capital.
Minerva addresses operational difficulties; it does not remove customers’ financing risks. If customers plan spending more cautiously or use infrastructure more slowly, additional capacity alone will not solve the problem. The varying maturity of the projects reinforces the point. At Bolt, data collection in Madrid is scheduled for 2027 and remains subject to regulatory approval. Choosing a technology, testing it and running it at scale are separate stages.
What to watch
The decisive question for investors is not how many initiatives Nvidia announces, but whether deliverability and productive use grow together. As long as demand outruns planned production and Nvidia can turn extra capacity into usable compute, the expansion remains coherent. Should customers’ willingness to spend falter, or supply chain plans slip, the upfront payments would weigh more heavily — and Minerva’s value would have to show up unmistakably in day-to-day operations.
Progress at SPIL and Amkor offers the next concrete checkpoints, alongside further advances in the announced telecom projects. A convincing record of implementation would give the market a sturdier argument against its financing worries than any additional letter of intent.
Ad
Nvidia Stock: Buy or Sell?! New Nvidia Analysis from October 23 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 23.
Nvidia: Buy or sell? Read more here...
