Google, the Sleeping Giant in Global AI Race, Now 'Fully Awake' (www.bloomberg.com)

After years of being cast as the laggard in the AI arms race that ignited with ChatGPT, Google has signaled a decisive comeback: reporters say the company is now “fully awake,” bolstered by a flurry of product, research and infrastructure moves capped by the 2025 debut of the Ironwood Tensor Processing Unit. That launch — and broader reinvigoration across Google’s teams — has changed industry perception that Google was behind, showing the company is aggressively scaling both hardware and software to compete on large-model training, multimodal systems and integrated AI services across its cloud and consumer products. For the AI/ML community the shift matters because Ironwood and Google’s renewed push expand the ecosystem of high-performance accelerators available for training and serving state‑of‑the‑art models, increase cloud compute options, and raise competitive pressure on rivals (OpenAI, Microsoft, Anthropic, etc.). Technically, a new TPU family implies optimized tensor math, throughput and system-level integration that reduce friction for large-scale training and inference workloads — accelerating research cycles and production deployments. The comeback also sharpens debates about compute concentration, industrial strategy and regulation as major cloud providers race to supply both the chips and tooling that underpin next‑generation AI.