AI Just Got Cheaper. Here’s Why That’s a Problem for NVIDIA.

AI Just Got Cheaper. Here’s Why That’s a Problem for NVIDIA.

More

Summary

Nate B Jones examines a growing tension in the AI economy: customers are learning to move work to cheaper models, which could slow the spending growth needed to justify the huge data center and chip build-out. He points to reporting that Nvidia CEO Jensen Huang warned his finance team about committed chip wafers and memory arriving like an avalanche.

The analysis leans on Ramp’s business spending data, which showed almost no growth in token spending across August and September, even as the most expensive frontier models gained share in October and account for roughly 70% of spend. Jones notes the data is volatile but argues it is still far from the tenfold growth that labs’ capacity commitments imply.

He also covers Google’s new Gemini models, Carbon for hard reasoning and Hydrogen for fast execution, and the Orchestrator feature in Antigravity, which switches between them. In Google’s testing it came close to Carbon’s success rate while using 24% less quota and finishing 19% faster. Open-source models can cut costs by 80 to 90% or more, he says.

The video closes by looking ahead to Nvidia’s upcoming earnings call and a bipartisan push, backed by Anthropic and OpenAI, to tighten chip exports to China.


📺 Source: AI News & Strategy Daily | Nate B Jones · Published November 16, 2026
🏷️ Format: News Analysis

1 Item

Channels

3 Items

Companies