TL;DR
CNBC reports that Nvidia has agreed to acquire AI chip startup Groq for $20 billion, a purchase described as the company's largest to date. Groq develops a language processing unit (LPU) it says can run large language models faster and with far lower energy use, a capability that could reshape AI hardware competition.
What happened
According to a CNBC report, Nvidia is acquiring Groq for $20 billion. The deal is described in the report as what would be Nvidia’s largest-ever purchase. Groq is a startup focused on a chip architecture it calls a language processing unit (LPU); the company has asserted that its LPU can run large language models about ten times faster while using roughly one-tenth the energy compared with other approaches. Groq’s CEO Jonathan Ross, who previously worked at Google and contributed to the development of the tensor processing unit (TPU), leads the company. In September, Groq raised $750 million at a reported $6.9 billion valuation. The company says its technology supports AI applications used by more than 2 million developers, up from approximately 356,000 a year earlier. TechCrunch reached out to Nvidia for comment, per the report.
Why it matters
- If completed, the acquisition could consolidate Nvidia’s position in the AI chip market by adding a different accelerator architecture to its portfolio.
- Groq’s claimed performance and energy efficiency for LLM workloads could influence how companies choose hardware for large-scale AI deployments.
- The deal underscores the high strategic value investors and incumbents place on specialized AI accelerators amid rising demand for AI compute.
Key facts
- Report source: CNBC (as cited by TechCrunch).
- Reported purchase price: $20 billion.
- Described as Nvidia’s largest acquisition to date.
- Groq develops a language processing unit (LPU) intended for LLM workloads.
- Groq claims its LPU can run LLMs 10x faster while using one-tenth the energy.
- Groq CEO Jonathan Ross previously worked at Google and helped develop the TPU (tensor processing unit).
- In September, Groq raised $750 million at a reported $6.9 billion valuation.
- Groq says it powers AI applications used by more than 2 million developers, up from about 356,000 the prior year.
- TechCrunch contacted Nvidia for comment, according to the article.
What to watch next
- Whether the reported deal will be finalized — not confirmed in the source.
- How Nvidia would integrate Groq’s LPU technology with its existing GPU ecosystem — not confirmed in the source.
- Any regulatory or antitrust review prompted by the acquisition, given its size and market implications — not confirmed in the source.
Quick glossary
- GPU: Graphics processing unit; a processor well-suited to parallel computations and widely used to accelerate AI training and inference.
- LLM: Large language model; a type of machine learning model trained on large text corpora to perform language-related tasks.
- LPU: Language processing unit; a specialized accelerator architecture referenced by Groq for running language models (definition kept generic).
- TPU: Tensor processing unit; a custom AI accelerator originally developed by Google for machine learning workloads.
Reader FAQ
Has Nvidia confirmed the acquisition?
Not confirmed in the source; TechCrunch reported it reached out to Nvidia for comment.
How much is Nvidia paying for Groq?
The report cites a $20 billion purchase price.
What does Groq make?
Groq develops a chip architecture it calls a language processing unit (LPU), aimed at accelerating LLM workloads.
Will this make Nvidia the dominant chip supplier?
The source says Nvidia would be poised to become more dominant with Groq, but detailed market impacts are not provided in the report.
IN BRIEF Posted: 2:03 PM PST · December 24, 2025 IMAGE CREDITS: GROQ Amanda Silberling Nvidia acquires AI chip challenger Groq for $20B, report says Nvidia is buying the AI…
Sources
- Nvidia acquires AI chip challenger Groq for $20B, report says
- Groq is set to raise ~$600M in a funding round led by …
- List of unicorn startup companies
- Source: Meta approached over a dozen people at Thinking …
Related posts
- Nvidia to acquire AI chip maker Groq for roughly $20 billion in cash
- How AMD Broke Into CPUs 50 Years Ago With an Intel 8080 Clone
- Roomba maker iRobot files Chapter 11 as Chinese manufacturer moves to take control