Alphabet Inc. (NASDAQ:GOOGL) is deepening its push into AI infrastructure with new custom chips, as competition with Nvidia (NVDA) intensifies. The move highlights a broader shift among cloud giants to build in-house hardware, even as reliance on Nvidia's technology remains firmly in place. 

Google Unveils AI Chips

Google announced two new AI-focused custom chips on April 22, Wednesday, at its Cloud Next 2026 event, as reported by TechCrunch. The launch splits its eighth-generation Tensor Processing Units (TPU) into separate roles for model training and inference. The TPU 8t is designed for training large AI models, while the TPU 8i is optimized for inference, which refers to running models after prompts are submitted.

The company said the new chips deliver up to 3x faster training performance, alongside a 2.8x improvement in price-to-performance and up to 80% better efficiency per dollar compared to previous generations. Google also highlighted the ability to scale deployments to over one million TPUs in a single cluster, signaling a push toward large-scale AI compute infrastructure.

The TPU line continues Google's strategy of developing custom silicon tailored for AI workloads rather than relying solely on traditional GPU architecture. Both TPU 8t and TPU 8i are expected to become available later this year.

Big Tech Expands Chip Ambitions As Demand Surges

Competition in AI hardware is rising as demand for compute power accelerates. This puts pressure on Nvidia and Advanced Micro Devices, Inc. (AMD), as cloud providers expand in-house chip development.

Major hyperscalers include Amazon.com, Inc. (AMZN), Microsoft Corp. (MSFT), and Meta Platforms, Inc. (META). They are all increasing internal chip efforts to reduce infrastructure costs and gain greater control over AI workloads. Meta is also advancing its MTIA chip program for both training and inference workloads.

Google Cloud is scaling its TPU capacity through expanded multi-gigawatt arrangements with Anthropic. The company is also extending TPU availability to additional AI partners, including OpenAI. Amazon has expanded a more than $100 billion long-term AI infrastructure agreement tied to Anthropic, according to a report by Yahoo Finance.

Nvidia Remains Key Partner As Google Maintains Cloud Integration

Google Cloud continues to rely on Nvidia systems within its infrastructure, signaling that its TPU expansion is additive rather than a replacement strategy. The company has also confirmed it will offer Nvidia's next-generation Vera Rubin chips later this year. 

Founder and Chief Analyst at MoonInsStrat, Patrick Moorhead, revisited a 2016 post that questioned what Google's TPU development could mean for Nvidia and Intel following its early unveiling at Google I/O. While concerns were raised at the time, Nvidia has since grown into a near $5 trillion company, reflecting how earlier replacement fears did not materialize.

Tech Stocks Trade Higher 

According to Benzinga stock data, Google, Nvidia, Advanced Micro Devices, Amazon, Microsoft, and Meta all carry a "Buy" consensus rating. Google is trading around $337, up 2.2%, while Nvidia is near $202, up 1.3%.

Advanced Micro Devices is trading around $303.46, up 6.67%. Amazon is priced near $255, up 2.18%, while Microsoft trades around $432, up 2.7%. Meta is valued at $674, up 0.88%.

Benzinga Disclaimer: This article is from an unpaid external contributor. It does not represent Benzinga’s reporting and has not been edited for content or accuracy.