It isn’t your imagination; Google Cloud is flooding the zone (techcrunch.com)

The headline: while Nvidia and OpenAI just cemented a headline-grabbing $100 billion-style agreement (non‑voting shares tied to massive chip purchases and capacity equal to training for millions of households), Google Cloud is doubling down on a different play — courting the next wave of AI startups rather than chasing headline incumbents. COO Francis deSouza says nine of the top 10 AI labs and roughly 60% of generative‑AI startups run on Google Cloud; the company has lined up $58 billion in new revenue commitments and offers startups $350,000 in cloud credits, go‑to‑market help, and access to a full “no‑compromise” AI stack from chips (TPUs) to models (Gemini, Vertex AI) with an “open ethos.” Why it matters: the industry is fragmenting into deep, bespoke hardware partnerships (Microsoft/OpenAI, Amazon/Anthropic, Oracle deals) while Google pursues breadth and future optionality. Technically significant moves include Google placing TPUs in third‑party data centers (e.g., a Fluidstack deal with up to $3.2B backing), hosting competitors’ models on Vertex AI, and publishing open protocols (Agent‑to‑Agent) — a strategy that both strengthens ecosystem lock‑in and exposes Google to antitrust scrutiny. Practically, Google’s startup-first approach could secure tomorrow’s unicorns and preserve interoperability at every layer, even as regulators watch whether its scale and data access give it an unfair edge.