NVIDIA Imposes 15% AI Markup

NVIDIA just doubled down ahead of earnings, hiking AI server prices by about 15% as memory costs surge and supply stays tight. Wall Street expects a staggering $92 billion in quarterly revenue, but the real tension is whether demand holds up as hyperscalers and enterprise buyers face pricier, harder-to-get gear. NVIDIA is betting that efficiency gains with its next-gen Vera Rubin chips—touting up to 30 times the throughput per megawatt—will justify the higher costs, but with memory and packaging bottlenecks at TSMC, price hikes look like a way to ration supply rather than expand it.

But here’s the catch: compliance and geopolitical risks are flaring, especially in China. Shipments of high-end H200 AI chips trickled in under special licenses, only to get stuck in Hong Kong due to power limits, while homegrown Chinese accelerators eat into NVIDIA’s share. Adding to the headache, Taiwanese prosecutors just flagged an NVIDIA manager in a server smuggling probe, raising the stakes on internal controls and casting a shadow over future sales channels.

Featuring reporting and insights from Tom’s Hardware and PBS NewsHour for the full picture on pricing, supply, and the high-stakes China drama.

Powered by Apisod.com