AI Chips Used to Be a One-Horse Race. Now Everyone's Winning. -- Barrons.com

Dow Jones04-17 22:41

By Adam Clark and Nate Wolf

Nvidia used to be the artificial-intelligence chip winner. Then it was Nvidia and Broadcom. Now it feels like everyone has an AI chip, and they're all being rewarded by the stock market.

Nvidia's graphics-processing units (GPUs) dominated the early stages of the AI revolution, as they proved ideal for training the models behind chatbots like ChatGPT. Gradually, custom chips from the likes of Google -- made in collaboration with Broadcom -- began to make inroads.

But this year has seen a significant broadening of the market. Shares of Marvell Technology, Intel, Arm Holdings, and Advanced Micro Devices have all posted strong gains in recent months. A bewildering mix of GPUs, CPUs, XPUs, TPUs, and NPUs are now powering AI for big technology companies. That's not even including memory chips from Micron Technology and its South Korean peers, or networking chips from the likes of Cisco Systems.

Investors appear to like the widening pool of AI players. The PHLX Semiconductor Index rose 1.4% Friday, putting it on track to extend its winning streak to 13 days -- its longest streak since 2014, according to Dow Jones Market Data.

All told, the PHLX has climbed almost 25% in April. If those gains hold, they would represent the chip index's best month since November 2002.

The most significant driver of the change in the AI chip market has been the shift from training to inference, the process of running models. As has long been predicted, the growth of AI agents, or programs that can act autonomously, has led to an explosive increase in demand for inference computing.

The inference pivot has been particularly good news for Intel, which has gone from being seen as an AI laggard to seeing its stock more than triple over 12 months, driven by projected demand for its central-processing units in AI servers. Intel is a recent Barron's stock pick.

"Commentary suggesting agentic AI could drive 4x higher CPU needs per gigawatts vs prior architectures helps frame the magnitude of the CPU demand increase we are seeing today," wrote Jefferies equity sales specialist William Beavington in a recent research note.

Even more players look set to join the market soon. Qualcomm has been a conspicuous laggard in the chip sector due to concerns about its core smartphone market, but has unveiled its AI200 and AI250 rack servers for inference.

Chip startup Cerebras Systems looks set to potentially list this year after delaying a planned initial public offering. ChatGPT developer OpenAI has already struck a multibillion-dollar agreement to buy computing capacity from Cerebras, according to The Wall Street Journal.

Write to Adam Clark at adam.clark@barrons.com and Nate Wolf at nate.wolf@barrons.com

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

 

(END) Dow Jones Newswires

April 17, 2026 10:41 ET (14:41 GMT)

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

At the request of the copyright holder, you need to log in to view this content

Disclaimer: Investing carries risk. This is not financial advice. The above content should not be regarded as an offer, recommendation, or solicitation on acquiring or disposing of any financial products, any associated discussions, comments, or posts by author or other users should not be considered as such either. It is solely for general information purpose only, which does not consider your own investment objectives, financial situations or needs. TTM assumes no responsibility or warranty for the accuracy and completeness of the information, investors should do their own research and may seek professional advice before investing.

Comments

We need your insight to fill this gap
Leave a comment