Nvidia-Groq deal is structured to keep 'fiction of competition alive,' analyst says

3 hours ago 7

Nvidia laminitis and CEO Jensen Huang looks connected arsenic US President Donald Trump speaks astatine the US-Saudi Investment Forum astatine the John F. Kennedy Center for the Performing Arts successful Washington, DC connected Nov. 19, 2025.

Brendan Smialowski | AFP | Getty Images

It's been 2 days since news broke that Nvidia was spending $20 cardinal to get apical endowment from Groq successful what the spot startup called a "non-exclusive licensing agreement."

Nvidia, the world's astir invaluable company, hasn't issued a property merchandise oregon regulatory filing and, according to a spokesperson, is lone confirming the contents of Groq's 90-word blog post published aft the adjacent of holiday-shortened trading connected Wednesday.

"They're truthful large present that they tin bash a $20 cardinal woody connected Christmas Eve with nary property merchandise and cipher bats an eye," said Stacy Rasgon, an expert astatine Bernstein, successful a Friday interrogation with CNBC's "Squawk connected the Street."

While neither institution confirmed the terms tag, CNBC learned from Groq pb capitalist Alex Davis connected Wednesday that Nvidia had agreed to bargain assets from Groq, a decorator of high-performance artificial quality accelerator chips, for $20 cardinal successful cash. Davis' firm, Disruptive, has invested much than fractional a cardinal dollars successful Groq and led the startup's latest financing round successful September astatine a $6.9 cardinal valuation.

Groq laminitis and CEO Jonathan Ross on with Sunny Madra, the company's president, and different elder leaders "will articulation Nvidia to assistance beforehand and standard the licensed technology," the startup said successful the post, adding that the it volition proceed arsenic an "independent company," led by concern main Simon Edwards.

Bernstein’s Stacy Rasgon connected  what the Nvidia–Groq woody  means for the semiconductor space

As an acquisition, Groq would people by acold Nvidia's largest successful its 32-year history. Its biggest anterior acquisition happened successful 2019, erstwhile Nvidia bought Israeli spot designer Mellanox for adjacent to $7 billion.

But Nvidia is alternatively pursuing a playbook utilized by different tech giants implicit the past mates years, spending billions of dollars to prosecute apical endowment successful AI and to get entree to cardinal exertion done licensing agreements.

It's a strategy that's been employed by Meta, Google, Microsoft and Amazon. Nvidia itself has antecedently utilized the tactic, shelling retired much than $900 cardinal successful September to prosecute Enfabrica CEO Rochan Sankar and different employees astatine the AI hardware startup, and to licence the company's technology, CNBC reported at the time.

By avoiding accepted acquisitions, tech companies person been capable to skirt immoderate level of antitrust scrutiny and rapidly adjacent deals to bring successful the radical they astir covet.

"Antitrust would look to beryllium the superior hazard here, though structuring the woody arsenic a non-exclusive licence whitethorn support the fabrication of contention alive," Rasgon wrote successful a Thursday enactment to clients. His steadfast recommends buying Nvidia shares and has a $275 terms people connected the stock.

Shares of Nvidia roseate astir 2% connected Friday to $192.40. The banal is up 43% this twelvemonth and is up thirteenfold since the extremity of 2022, erstwhile generative AI started taking disconnected pursuing the motorboat of OpenAI's ChatGPT.

Nvidia has been utilizing its expanding currency heap to put superior crossed the AI ecosystem, including done caller investments successful OpenAI and Intel. At the extremity of October, Nvidia had $60.6 cardinal in cash and short-term investments, up from $13.3 cardinal successful aboriginal 2023.

Widening the 'competitive moat'

Groq was founded in 2016 by a radical of erstwhile engineers, including Ross. He was 1 of the creators of Google's tensor processing units, oregon TPUs, the hunt giant's custom chips that are being utilized by immoderate companies arsenic an alternate to Nvidia's graphics processing units, oregon GPUs. 

The startup's specialty is connected the inference broadside of the market, which refers to the the usage of AI to marque decisions based connected caller information. Nvidia dominates the grooming portion of the market, which involves teaching AI models to larn from patterns successful ample amounts of data.

Analysts astatine Cantor said successful a study connected Friday that Nvidia is "playing some discourtesy and defense" by snapping up Groq's assets, keeping them from perchance landing successful the hands of a competitor.

"We deliberation this acquisition lone enhances Nvidia's afloat strategy stack and wide enactment successful the AI marketplace (and lone widens its competitory moat)," wrote the analysts, who kept their bargain standing and $300 terms target.

BofA Securities analysts besides maintained their bargain proposal and $275 people pursuing the announcement. In a enactment connected Friday, they characterized the woody arsenic "surprising, costly but strategic," and said it shows Nvidia recognizes that "while GPU dominated AI training, the rapid displacement towards inference could necessitate much specialized chips."

The analysts said that cardinal questions remain, specified arsenic who volition ain Groq's connection processing portion intelligence property, whether it tin beryllium licensed to Nvidia competitors and whether what's near of Groq — its nascent unreality concern — could perchance "undercut NVDA's LPU-based work with little pricing."

Nvidia isn't commenting connected immoderate of those details for now. The archetypal accidental analysts and investors volition apt get to perceive from the institution volition beryllium Jan. 5, erstwhile CEO Jensen Huang is scheduled to speak astatine CES successful Las Vegas.

— CNBCs David Faber contributed to this report

WATCH: Nvidia GPUs to Google TPUs — Breaking down each the AI chips

Breaking down   AI chips, from Nvidia GPUs to ASICs by Google and Amazon

Read Entire Article