This could be Nvidia's next big move, with the stock in search of a positive catalyst

By Emily Bary

 A report says Nvidia is planning a new chip platform that seeks to address evolving customer needs as the AI era transitions more toward inference workloads 

 Nvidia, led by Jensen Huang, struck a major licensing deal with startup Groq late last year, and soon investors may see the early fruits of that arrangement. 

 A historic earnings beat wasn't enough to lift Nvidia's stock. What about a brand new chip platform? 

 The Wall Street Journal reported late on Friday that the company is soon planning to debut an inference-oriented chip platform that brings in technology designs from Groq, the startup that Nvidia (NVDA) struck a $20 billion licensing deal with at the end of last year. Through that deal, Nvidia also brought Groq employees, like founder Jonathan Ross, on board. 

 Nvidia didn't immediately respond to a MarketWatch request for comment. 

 The rumored new platform, which the WSJ said is due to be announced during Nvidia's annual GTC developer event in March, could mark an acknowledgment of the changing world of artificial-intelligence chips, the report suggested. Nvidia is by far the dominant maker of graphics processing units, which have largely powered the AI revolution so far. But Alphabet (GOOGL) and other Broadcom (AVGO) partners have been working on their own custom chips that are cheaper and potentially more targeted to AI's new era. 

 See also: Why Nvidia's stock fell despite a historic earnings beat 

 Specifically, customers are turning to AI chips more for inference, which is the process by which models reach conclusions based on information that's new to them. The WSJ report noted that customers increasingly want less expensive and more efficient processors for these sorts of workloads. 

 Groq has been known for making language-processing units, or LPUs. "Unlike the GPU, the LPU is a deterministic, low-power processor," wrote Cantor Fitzgerald analyst C.J. Muse after the companies struck their licensing and "acqui-hire" deal in December. 

 "LPU strengths? Extremely low-latency, energy-efficient single user token per second, one of the fastest in the market," Muse wrote. But the chips make use of a kind of superfast memory that has limited capacity, meaning that lots of racks would be needed. 

 Don't miss: Nvidia's earnings report is just a prelude to what should really matter for investors 

 At the time, Muse thought that Groq's newfound presence within Nvidia would "enable even greater share of the inference market, particularly for the next leg of the AI infrastructure buildout - real-time workloads like robotics and autonomy." 

 Nvidia's stock, meanwhile, has been in search of a positive catalyst. Shares are down 5% so far this year, with an upbeat earnings announcement from earlier in the week unable to sway investors. The PHLX Semiconductor index is up 14% over the course of 2026 to date. Chip-sector investors this year have been particularly enamored by makers of memory products and equipment for semiconductor manufacturing. 

 Analysts have said that GTC could end up helping Nvidia's stock, as the company has perhaps what could be its most market-moving updates for the big event. 

 See also: Trump blacklists Anthropic - and OpenAI swoops in 

 -Emily Bary 

 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

02-28-26 1024ET

Copyright © 2026 Dow Jones & Company, Inc.