Google's AI lead is growing in this key area. That's good news for Alphabet's stock.

MW Google's AI lead is growing in this key area. That's good news for Alphabet's stock.

By Britney Nguyen

Needham analyst Laura Martin notes that Google is extending its edge over rivals as it advances its custom-chip efforts

Google uses custom chips for various purposes, including to help power its search and YouTube businesses.

Coming off of Google's renewed chip deal with Broadcom, an analyst is highlighting just how much semiconductors serve as an advantage to Alphabet.

The artificial-intelligence race depends on having compute power to build and run increasingly complex models, and that's driving an intense competition for chips. And the fact that Google $(GOOGL)$ $(GOOG)$ has its own silicon could give it an edge over its AI rivals, according to Needham analyst Laura Martin.

For more than a decade, Google has co-developed its tensor processing units, known as TPUs, with Broadcom $(AVGO)$. These chips run Alphabet's search and YouTube algorithms, and they also power its Gemini AI models.

That all means Google "has an enormous lead in AI compute" relative to hyperscaler rivals Microsoft $(MSFT)$, Meta Platforms (META) and Amazon.com (AMZN), Martin wrote in a note to clients.

While the other hyperscalers do have their own custom-chip programs that are used alongside processors from Nvidia (NVDA) and Advanced Micro Devices $(AMD)$, Martin cited data from research institute Epoch AI that showed Google's TPUs make up a large share of the AI chips that it uses. Combined with chips that it buys from Nvidia, Google is "likely the largest owner of AI compute," according to Epoch.

See more: Broadcom's stock is rising. Here's why its new Google and Anthropic deals are so significant.

Designing its own chips also allows Google to control costs because it doesn't have to buy expensive compute from external sources, Martin said.

The Epoch data showed that Google's chip-ownership advantage has expanded since the first quarter of 2024.

Not only does Google have chips, but it also has a vertically integrated system of AI infrastructure that "puts it materially ahead of its hyperscaler peers in terms of total deployed AI compute capacity," Martin said.

In her view, Google's AI-chip lead is "a valuable hidden asset for [Google] shareholders that will drive upside to estimates over the next 24 months." Martin, who has a buy rating on Alphabet's stock, has a price target of $400, representing upside of 31% from the stock's closing price on Tuesday.

Google's full-stack approach with chips, infrastructure and AI models means that it can improve its models faster and spend less on running them, while also benefiting monetization, Martin explained.

"We see a flywheel whereby more compute drives better [large language model] outputs, which drives more usage, which justifies more investment," Martin said.

Therefore, she thinks Google will lead the pack when it comes to return on invested capital from generative AI over the next decade.

Don't miss: Anthropic's meteoric rise shocked the market - but the AI crown remains up for grabs

-Britney Nguyen

This content was created by MarketWatch, which is operated by Dow Jones & Co. MarketWatch is published independently from Dow Jones Newswires and The Wall Street Journal.

(END) Dow Jones Newswires

April 08, 2026 16:24 ET (20:24 GMT)

Copyright (c) 2026 Dow Jones & Company, Inc.