Grafa logo

Sign in

Nvidia absorbs another rival for $20B boosting decentralised AI debate

Written by Jon CuthbertPublished Dec. 25 2025Last Updated Dec. 25 2025

Share

Nvidia has agreed to pay about $20 billion to acquire assets from artificial intelligence chip startup Groq.

The transaction marks Nvidia’s largest deal to date and continues its strategy of neutralising emerging competitors.

Nvidia structured the agreement as a licensing deal rather than a full acquisition.

The approach mirrors a similar transaction completed just three months earlier.

In September, Nvidia paid more than $900 million to license technology and hire staff from Enfabrica.

Both deals use licensing frameworks that may limit antitrust scrutiny.

Nvidia’s $40 billion bid for Arm Holdings was blocked by regulators in 2022.

Groq raised $750 million only three months ago at a valuation of $6.9 billion.

That funding round included BlackRock, Samsung, Cisco and 1789 Capital.

Groq confirmed that Nvidia is acquiring substantially all of its assets.

The company’s cloud computing business was excluded from the transaction.

Groq described the deal publicly as a non-exclusive licensing agreement.

Groq founder Jonathan Ross will join Nvidia following the transaction

Ross previously worked at Google and helped develop its Tensor Processing Unit.

Groq president Sunny Madra and other senior executives are also joining Nvidia.

The startup will continue operating independently under a new chief executive.

Chief financial officer Simon Edwards has been appointed to lead Groq’s remaining business.

Market commentators say the transaction reflects a repeating Nvidia playbook.

Analysts note Nvidia is buying or licensing rivals before they can scale.

We will buy you before you can compete with us.

The Kobeissi Letter said.

Groq’s chip design focuses on real-time inference performance.

Its Language Processing Unit uses on-chip SRAM instead of external DRAM.

The architecture is designed for high energy efficiency and low latency.

The trade-off limits model size but excels at speed-sensitive workloads.

Nvidia is expected to integrate the technology into its broader ecosystem.

Competitive pressure across AI hardware has intensified in recent months.

Google recently unveiled its seventh-generation TPU, codenamed Ironwood.

Google also released Gemini 3, trained entirely on its own TPU infrastructure.

Nvidia responded publicly by highlighting its platform flexibility.

NVIDIA is a generation ahead of the industry.

The company said.

The Groq deal has renewed debate around decentralised AI infrastructure.

Projects offering decentralised compute argue Nvidia’s dominance strengthens their case.

Decentralised platforms position themselves as alternatives to centralised AI supply.

Advocates say distributed GPU networks can reduce costs and expand access.

People can put their own supply onto a network and get fairly compensated.

Jack Collier said.

Some decentralised platforms claim enterprise adoption and measurable cost savings.

Analysts caution that Nvidia’s expanding technical lead remains difficult to challenge.

The acquisition further widens the gap between incumbents and emerging competitors.

Independent AI chip developers face increasing pressure as consolidation accelerates.

Cerebras Systems, another Nvidia rival, is preparing for a public listing.

Observers question whether smaller competitors can remain independent.

Nvidia’s latest deal underscores its financial and strategic gravity.

The transaction highlights how consolidation is reshaping AI infrastructure markets.

Decentralised alternatives may grow in narrative appeal as concentration increases.

For now, Nvidia continues to set the pace across global AI hardware development.