AI chip startup Groq confirmed on Wednesday that it has raised $750 million in a new funding round, pushing its post-money valuation to $6.9 billion. The round, led by Disruptive, included participation from major investors such as BlackRock, Samsung, Cisco, and Deutsche Telekom Capital Partners. Existing backers, including D1 and Altimeter, also joined.

This funding round surpasses earlier rumors of a $600 million raise at a $6 billion valuation and comes roughly a year after Groq raised $640 million at a $2.8 billion valuation in August 2024, effectively doubling its valuation in twelve months. PitchBook estimates that Groq has now raised over $3 billion in total.

Founded in 2016 by Jonathan Ross, a former Google TPU core team member, Groq develops LPUs—Language Processing Units—designed specifically for AI inference workloads. Unlike traditional GPUs, LPUs focus on efficiently running large AI models at lower cost and higher energy efficiency, with Groq claiming up to ten times the efficiency of comparable systems. Groq's hardware integrates software and compiler optimizations, including a RealScale enhancement, to minimize overhead and maximize throughput for trillion-parameter models.

Groq offers its technology both as a cloud service and as on-premises hardware clusters, known as GroqRack, which consist of server racks stacked with integrated hardware/software nodes. These systems support popular open-source AI models, including those from Meta, DeepSeek, Qwen, Mistral, Google, and OpenAI. The company reports that its infrastructure now powers the AI applications of over 2 million developers, up from 356,000 last year.

CEO Jonathan Ross said the new funds will be used to expand Groq's global data center footprint, with new locations scheduled for 2025 and 2026, including the first Asia-Pacific facility. Groq is already a key AI infrastructure partner for clients such as Saudi Arabia's Humain and has supported OpenAI's GPT-OSS deployment in the region. The company operates 13 sites worldwide, with over 10% of recently added capacity already committed.

Despite investor enthusiasm, Groq has trimmed its 2025 revenue outlook from over $2 billion to $500 million due to limited data center capacity, with some revenue now expected to materialize in 2026. Ross declined to comment on specific sales targets, emphasizing instead the company's focus on scaling global AI inference infrastructure.

Groq's rapid growth and the strong investor appetite highlight ongoing demand for credible alternatives to Nvidia in AI compute infrastructure, particularly in inference workloads where the market remains open for disruption.