Google is committed to both TPUs and GPUs, leveraging both for training and serving AI models, while maintaining flexibility in their infrastructure.

Google's Gemini 2.5 Pro is now the leading model across all categories, showcasing significant advancements in AI capabilities.

Google has the largest TPU cluster with a very advanced infrastructure for their Tensor Processing Units (TPUs), which are tightly connected across multiple data centers in regions like Nebraska and Iowa, allowing for efficient training of models like GPT-5, but their data centers are spread across multiple locations, complicating multi-datacenter training compared to tightly connected clusters.

Google primarily uses its own TPUs for AI models but also supports Nvidia chips, balancing performance and computational needs.

With the release of GPT-3, it seems the sky is the limit for this new paradigm of scaling large language models.

GPT-4.5 is OpenAI's largest and most humanlike model to date, representing a significant step in scaling up unsupervised learning and achieving a deeper understanding of the world and human experience.

The Google Translate project faced challenges scaling up from a small dataset to a much larger one, requiring significant engineering efforts to improve model quality.

Gemini 1.5 features a context window of one million tokens, significantly larger than GPT-4's 128,000 tokens, allowing for more extensive data processing and potentially transforming how consumer applications are built.

Our Gemini models are making Google Search more intelligent, agentic, and personalized, allowing users to ask longer and more complex queries.