Nvidia shares slip after Microsoft unveils a new AI chip, raising questions about demand for GPUs

Nvidia’s shares fell about 1% on October 12, continuing a theme that has hovered over the AI chip market: whether hyperscalers will increasingly rely on their own custom silicon instead of buying graphics processing units, or GPUs, from Nvidia.

In trading reported by Yahoo Finance via Proactive Investors, Nvidia shares dropped $2.73 to $226.55. The move was tied to the release of a new artificial intelligence chip from Microsoft, which the market viewed as a potential headwind to Nvidia’s ecosystem.

Microsoft’s announcement, as described in the market coverage, suggested that the new chip could reduce the company’s reliance on GPUs for AI workloads. Nvidia has been the default supplier for training and many forms of inference, but the idea of software and cloud operators shifting more compute to in-house accelerators has repeatedly come up during earnings and product discussions across the sector.

For Nvidia, the concern is not that GPUs disappear, but that the incremental demand from large customers could grow more slowly than it otherwise would. Microsoft is a major cloud and AI platform player, and when it indicates new chip capability, investors often reassess how much of the next wave of spending may flow through Nvidia’s supply chain.

The key uncertainty is scope. Market reaction alone does not clarify whether Microsoft’s new chip is intended for training, inference, or specific internal services only, nor does it show when broader availability to partners or developers might begin. Without additional disclosure, traders typically anchor on the direction of travel rather than the timing or the size of the potential GPU substitution.

In AI infrastructure, custom chips matter because they can be tailored to reduce power consumption, improve throughput for particular models, or integrate more tightly with a provider’s software stack. If Microsoft’s new accelerator performs effectively for its own workloads, it can change procurement decisions across its cloud environment, influencing how often customers need Nvidia GPUs for comparable tasks.

At the same time, GPUs still offer flexibility across model types and software ecosystems. Custom silicon can coexist with Nvidia hardware, especially if companies adopt a mix of accelerators for different workload categories. That means a near-term share price move does not automatically translate into a durable long-term loss of market share, but it does reflect heightened sensitivity to customer concentration risk.

What to watch next is whether Microsoft provides technical benchmarks, deployment timelines, or details about where the chip will be used. For Nvidia, investors will likely look for updates on how major cloud customers plan their AI infrastructure roadmaps and whether any customers publicly quantify GPU spending changes tied to new in-house chips.