Google to Release New AI Chips, Challenging Nvidia | Bloomberg Tech 4/20/2026

Editor's take

Google plans to introduce new custom-designed AI processors specifically engineered for inference tasks, aiming to compete with established players like Nvidia. This move signals Google's continued commitment to vertical integration in AI hardware, leveraging its deep expertise in both software and silicon to optimize performance and cost for its vast AI workloads, from search to Bard.

The significance lies in Google's potential to disrupt the AI chip market, which is currently heavily dominated by Nvidia's GPUs. By offering specialized inference chips, Google could reduce its reliance on external suppliers and potentially offer more cost-effective solutions to other enterprises. This also intensifies the hardware arms race, pushing other cloud providers and AI developers to consider similar in-house silicon strategies.

Future developments to monitor include the actual performance benchmarks of these new Google chips compared to Nvidia's H100 or successor models, and whether Google will make these inference accelerators available to external cloud customers. Success here could accelerate the trend of custom AI silicon, while broader adoption hinges on competitive pricing and robust software ecosystem support beyond Google's internal needs.