Google is reportedly in talks with Marvell Technology to develop two new AI chips — a move that signals something more significant than a hardware procurement decision.
The billions in AI infrastructure spending projected for this year is not just a capital investment story. It is a control story. Google, Meta, and the rest of Big Tech are racing to reduce dependence on Nvidia not because Nvidia’s chips don’t work, but because dependence on a single supplier at this scale is a strategic liability. When the infrastructure of intelligence is concentrated in one place, the organizations that do not own a piece of it are permanently at a disadvantage.
This is the AI adoption story most enterprise leaders are not paying enough attention to. The conversation in most companies is still about applications — which AI tools to deploy, how to integrate them, how to train the workforce. Those are the right questions. But the organizations setting the terms of the next decade are asking a different question: who owns the underlying infrastructure, and what does that ownership mean for us?
The chip war is a proxy for something larger. It’s a race to determine who gets to define the conditions under which AI operates: at what speed, at what cost, under whose governance. Google building its own chips is not a technology story. It is a sovereignty story.
For enterprise leaders not in the semiconductor business, the lesson is not to build chips. It is to understand which of your AI dependencies are strategic liabilities, and to diversify before someone else makes that decision for you.
That’s the human judgment no algorithm will make on your behalf.
Source: Reuters, April 20, 2026
Get new articles from Robin Green delivered directly.
Insights on AI leadership, the future of work, and human collaboration. No noise. Unsubscribe any time.
Subscribe free