Technology

Chip Giants Defy AI Pause: Nvidia and AMD Post Record Earnings Amidst Frontier Model Debates

Nvidia and AMD report record earnings, proving that chip demand remains robust despite pauses in frontier AI model training, driven by surging inference workloads.

S
SciTech Editorial Team
Newsroom
··2 min read
Chip Giants Defy AI Pause: Nvidia and AMD Post Record Earnings Amidst Frontier Model Debates
The recent pause in frontier AI model training by leading labs has failed to dampen the booming demand for semiconductor hardware, as evidenced by record-breaking earnings from industry giants Nvidia and AMD. Despite earlier concerns that a slowdown in developing the most powerful AI models could derail chip sales, financial results released in late 2026 reveal a market that is accelerating rather than contracting. This resilience highlights a structural shift in the AI economy, where the sheer volume of inference workloads is driving demand independent of new model training cycles. Nvidia posted a blockbuster quarter, with data center revenue surging 110 percent year-over-year to reach $107.6 billion in the three months ended October 25. The company reported that its latest AI computing system, the Vera Rubin NVL72, experienced the "fastest product ramp in the company’s history," generating nearly $20 billion in revenue in its first quarter of production shipments alone. Major tech firms, including CoreWeave, Google, Microsoft, and Oracle, are already deploying these racks, while Meta plans to expand Vera Rubin deployments for both training and running AI agents. Nvidia’s interim CEO, Colette Kress, noted that major frontier labs account for approximately 15 percent of its data center revenue, prompting the company to raise its revenue growth forecast for the next fiscal year from 70 percent to 75 percent. AMD also reported its biggest quarter ever, with total revenue up 45 percent to $13.4 billion, driven largely by data center sales that nearly doubled to $8.5 billion. OpenAI and Anthropic were among the first customers to run workloads on AMD’s new high-end Helios systems, which utilize the latest MI450 GPUs. This competitive landscape ensures that chipmakers benefit from the intense rivalry among AI providers, regardless of temporary pauses in specific training runs. The market is clearly prioritizing the deployment and utilization of existing models over the exclusive development of new frontier capabilities. Analysts suggest that a frontier slowdown would not significantly hurt chip sales due to long-term procurement agreements and the rising demand for inference. Neil Shah, vice president of research at Counterpoint Research, explained that it would take about six months to see the financial effects of a slowdown due to existing computing and memory contracts. Furthermore, data from corporate spending platform Ramp indicates that frontier model families accounted for about 60 percent of token usage among businesses in early November, a record high and three times their share at the start of the year. This surge in usage, coupled with token prices falling nearly 60 percent since May, is spurring more widespread adoption and sustained demand for high-performance accelerators. In conclusion, the AI chip market is decoupling from the pace of frontier model development, driven instead by the massive scale of inference workloads and enterprise adoption. As competition drives down token prices, the volume of AI usage increases, creating a net increase in demand for all types of accelerators. Nvidia and AMD are positioning their most powerful chips to serve both training and inference needs, ensuring that the industry remains robust even if the pace of new model creation fluctuates. The financial data confirms that the AI boom is here to stay, underpinned by a growing infrastructure of high-performance computing resources.

Share

More from this section: Technology