Groq, founded by a former Alphabet engineer, is developing specialized hardware called LPUs (language processing units) and an inference engine to accelerate AI models more efficiently and cost-effectively, challenging industry giants like Nvidia.

Major players like NVIDIA and AMD are also focusing on inference chips, reflecting a broader industry shift toward inference hardware as AI applications grow more sophisticated.

Groq's LPUs utilize advanced compiler technology called RealScale, which compresses neural networks effectively, enabling models with up to 1 trillion parameters and offering up to 10 times higher energy efficiency than traditional GPUs.

Recently, Groq raised $750 million in a funding round led by Disruptive, boosting its valuation to $6.9 billion and bringing total funding to over $3 billion since its 2017 seed round.

This funding more than doubled Groq's valuation within a year, with previous rounds valuing the company at around $2.8 billion in August 2024, and significant investments from firms like BlackRock, Samsung, and Cisco.

The new capital will be used to expand Groq's global operations, develop new products, and attract talent, positioning it to compete with established players like Nvidia, AMD, and Intel.

Groq plans to use the funds to expand its data center capacity, including opening its first Asia-Pacific facility within this year to meet rising customer demand and address supply shortages.

Since its initial funding, Groq has grown rapidly, now operating 13 data centers worldwide, with plans for new facilities to support the increasing adoption of AI hardware.

The company’s CEO, Jonathan Ross, emphasized that the expansion aims to meet the growing demand for AI chips and computing power, especially as AI workloads become more complex.

In February, Groq secured a $1.5 billion commitment from Saudi Arabia to deliver AI chips, with an expected revenue of about $500 million in 2025 from these contracts, supporting regional AI deployment.

Groq has increased its computing capacity by over 10% in the last month, with all new capacity already in use, indicating strong demand for its AI hardware solutions.

A significant portion of this capacity has been allocated to international markets like Saudi Arabia, supporting AI initiatives such as Humain and OpenAI’s GPT-OSS.

The AI inference market is projected to grow rapidly, from $97.24 billion in 2024 to over $253 billion by 2030, driven by investments in data centers and the rise of generative AI applications.