Groq challenges Nvidia's AI chip dominance with $6 billion valuation bid

California-based startup Groq eyes global expansion as it raises fresh funding, strikes a major West Asia deal, and accelerates revenue growth

Chip startup Groq is raising up to $500 million in new funding, which would bring its valuation to around $6 billion, according to The Information. The California-based company is positioning itself as a serious challenger to Nvidia’s dominance in AI chips, riding on growing global demand and a $1.5 billion deal in Saudi Arabia.

 

This comes at a time when artificial intelligence (AI) is reshaping industries worldwide—from healthcare and finance to energy and defence. The hardware behind AI models, used to power software like ChatGPT, is often hidden from public view. However, Groq’s rise could signal growing momentum behind alternatives to Nvidia’s near-monopoly-like position.

 

 

What is Groq?

Founded in 2016 by Jonathan Ross, a former lead engineer behind Google’s Tensor Processing Unit (TPU), Groq builds chips specifically for AI inference — the part where AI models generate outputs (like text or images).

 

Unlike Nvidia, which makes chips for both training and inference, Groq focuses only on inference and aims to do it faster and more efficiently.

 

Groq’s main product is the Language Processing Unit (LPU). These chips can generate up to 750 tokens per second, far outpacing systems like GPT-3.5, which outputs around 40 tokens per second.   

How is Groq different?

Groq’s edge comes from using SRAM (a much faster on-chip memory) instead of the more common HBM. SRAM is expensive and holds less data, but it’s extremely fast, making Groq’s chips great for real-time AI, not model training.

 

According to a report by Forbes, each LPU card costs around $20,000, and Groq builds systems that connect them using a switchless high-speed network, cutting down on data bottlenecks and energy use.

 

Most of Groq’s revenue comes from cloud-based services, similar to how companies use Amazon Web Services or OpenAI. It also sells physical hardware and operates data centres for enterprise clients.

 

Challenging Nvidia

Nvidia currently controls over 90 per cent of the AI chip market, largely thanks to its powerful GPUs that serve as the backbone of many machine learning models. The company made history earlier this year when its market capitalisation reached $4 trillion.

 

However, as demand for AI services grows exponentially, supply chain constraints, cost, and energy consumption have become growing concerns.

 

Startups like Groq aim to address these issues by offering more specialised hardware solutions.

 

Why Saudi Arabia is interested

In regions like West Asia, where Nvidia chips are less accessible, companies like Groq and other startups, including SambaNova Systems, are filling the gap.

 

The Saudi government has been investing heavily in AI infrastructure and partnerships. Groq recently announced its $1.5 billion commitment to support Saudi Arabia’s AI ambitions by supplying its chips. Though details remain unclear on whether the figure represents guaranteed purchases. 

Groq is also setting up a data centre in Finland to support global demand.

 

Number game so far

Revenue: $90 million in 2024 → Projected $500 million in 2025  Chips in use: Around 70,000 (less than originally expected by the company)  Funding so far: Over $1 billion from major investors like BlackRock, Samsung, and Tiger Global  New raise: Up to $500 million  Current valuation: Potential to reach $6 billion (up from $2.8 billion in 2024)  

What’s next in AI hardware?

If Groq locks in the full funding round, it will be one of the biggest private raises in the AI hardware space this year. The money will go toward scaling production, expanding cloud services, and building partnerships in markets where Nvidia is hard to access.