The prediction’s time horizon has not fully arrived yet. Friedberg’s claim is that the 2023‑era, extremely elevated level of GPU/AI‑compute spending driving Nvidia’s results is not a long‑term steady state and that, as companies find the “efficient frontier,” aggregate Nvidia compute demand growth will slow within a few years (by ~2026).

As of late 2025, available data show:

• Nvidia’s data‑center and AI‑GPU revenue has continued to grow very rapidly through fiscal 2024 and into fiscal 2025, with multiple record quarters and strong forward demand guidance reported in earnings and widely covered in the financial press.
• Hyperscalers and large AI companies are still ramping GPU capex aggressively, and there is no clear consensus that total spend per application has already been materially “rationalized downward” in a way that has slowed overall Nvidia compute demand growth.

However, since the prediction explicitly allows until “~2026” for this rationalization and slowdown in growth to become evident, and the current date is 30 November 2025, it is too early to say definitively whether Nvidia’s aggregate compute demand growth will have slowed by the end of that window. The available evidence so far leans against a slowdown, but the prediction is framed about the state by 2026, which has not yet fully elapsed.

Because the forecast’s end date has not passed, the correct grading is: it cannot yet be decisively confirmed or falsified.