Friedberg’s core claim was that Google’s Gemini launch marked the beginning of Google being a major competitive player and a leading force in AI "for the foreseeable future (at least several years)."

What has happened so far (Dec 2023–Nov 2025):

• Google has aggressively iterated on Gemini: launching Gemini 1.5 with very large context windows and multimodal capabilities, and rolling it out broadly via Google AI Studio and Vertex AI, aimed squarely at high‑end frontier‑model use cases. (blog.google)
• It then introduced Gemini 2.0 (Flash and later Pro, Flash‑Lite), positioned as its cutting‑edge family of models, integrated into the Gemini app and exposed via APIs for developers and enterprises—exactly the behavior of a sustained top‑tier AI competitor. (blog.google)
• Market and ecosystem signals show Google as one of the central AI players:
  • A Reuters piece in November 2025 ties Alphabet’s near–$4T valuation surge explicitly to AI advances, including its Gemini 3 model and AI‑driven cloud growth, framing this as a comeback in AI leadership. (reuters.com)
  • Enterprise model‑usage data cited around Anthropic’s Claude Opus 4.5 launch reports Anthropic with ~32% enterprise AI share, OpenAI 25%, Google 20%, Meta 9%—placing Google clearly in the top tier of AI providers, not a marginal player. (businessinsider.com)
  • Reports that Meta is negotiating multi‑billion‑dollar deals to use Google’s TPUs and Google Cloud for AI workloads show Google is also a major infrastructure and chip player, competing against Nvidia on the hardware/compute side of the AI stack. (nypost.com)

Why the verdict is “inconclusive”:

• As of November 30, 2025, we are not yet two years past the December 8, 2023 prediction date—short of the "at least several years" horizon implied in the normalized prediction.
• The evidence strongly supports that so far Google has remained a leading, highly competitive AI player and that Gemini did indeed mark a durable strategic push rather than a one‑off launch.
• However, because the prediction explicitly concerns performance over several years into the future, the full time window has not elapsed. We can say the prediction is on track and directionally correct to date, but we cannot definitively confirm its multi‑year component.

Therefore, the appropriate classification is "inconclusive" (too early to fully judge), with the important nuance that all current evidence aligns with Friedberg’s forecast rather than contradicting it.