The 5 reasons why Google is suddenly on a tear and dominating the AI race (www.businessinsider.com)

Google’s comeback centers on Gemini 3, a new multimodal model that rolled out publicly in mid-November and has delivered strong benchmark gains across coding, design and analysis tasks—handling website design and simple game creation as well as specialized coding jobs. That performance has quieted concerns that Google had fallen irretrievably behind after 2022’s ChatGPT moment: investors have pushed the stock up roughly 70% this year (about +12% since Gemini 3’s rollout), and the company has reclaimed top market-cap positions. Importantly, reviewers say Gemini 3 shows scaling still matters—Google used its massive training stack to squeeze more capability out of models and integrate them into user-facing products like Search and “AI Mode.” Beyond the model itself, five strategic moves amplify Google’s position for the AI/ML community. First, a decade of in-house Tensor Processing Units (TPUs) trained Gemini and are now a cloud product that could eat into Nvidia’s dominance—reports of a multi‑billion dollar Meta deal underscore that. Second, a recent antitrust ruling left Google’s search business largely intact (with limits on exclusivity and some data sharing). Third, Warren Buffett’s Berkshire built a $4.3B Alphabet stake, signaling investor confidence. Finally, Search ad revenue rose ~15% in Q3 and Google is experimenting with ads inside AI Mode, showing monetization can survive an AI-driven UX shift. Together, these factors matter for researchers and companies: they change access to training infrastructure, competitive dynamics for models and chips, regulatory contours, and the commercial pathway for deploying large AI systems.