Google launches faster, cheaper AI chips to challenge Nvidia's dominance in cloud computing

Google released two new TPU chips (Tensor Processing Units — custom-built processors designed specifically for AI tasks) that are faster and more affordable than the previous generation, directly competing with Nvidia's GPUs that currently dominate the cloud AI market.

The new TPUs deliver higher performance per dollar spent, which means companies running AI workloads on Google Cloud will pay less for the same computing power — a concrete advantage for startups and enterprises building AI applications at scale.

For business decision-makers choosing cloud providers, this creates real negotiating power: Google now offers a credible alternative to Nvidia-dependent vendors like AWS and Azure, potentially lowering your AI infrastructure costs by 20–30% depending on your workload.

Google is also continuing to support Nvidia chips on Google Cloud for now, so existing customers won't be forced to switch — but the company is signaling it wants to reduce dependence on Nvidia long-term by building its own silicon.