Nvidia Groq 3 LPX Rack Enters Production After $20 Billion Acquisition
Key Points
- Nvidia announced Monday that its Groq 3 LPX rack is in full production following the company's $20 billion acquisition of Groq assets in December.
- The Groq 3 LPX rack delivers 3,400 tokens per second according to Artificial Analysis benchmarks, outpacing OpenAI's Ultrafast mode at 750 tokens per second.
- Nvidia senior director Dion Harris said the racks will deploy at neocloud Nebius alongside Vera central processors and Rubin graphics processors later this year.
The Groq architecture includes 500 megabytes of speedy SRAM on the chip's die itself to reduce memory-related bottlenecks. Groq chips are manufactured by Samsung, while Taiwan Semiconductor Manufacturing Co. makes Nvidia's GPUs. Harris emphasized that Groq chips are not intended to replace GPUs, which remain the workhorse of AI chips for both training and inference. "This isn't about replacing GPUs," Harris said. "It's about using the right price, right processor for the right part of the workload."
Low-latency chips like Groq mainly focus on the "decode" phase of serving models, a specialized function that enables AI agents to feel responsive without long lags for users, especially for coding applications. Nvidia CEO Jensen Huang in March projected $1 trillion in cumulative sales between current-generation Blackwell chips and the new Vera Rubin systems through 2027. Huang stated he would allocate one-quarter of data center space intended for coding applications to Groq chips, with the remainder dedicated entirely to Vera Rubin systems.
Read Next
Shein Hong Kong IPO Values Fast-Fashion Giant at $27 Billion
15 hours ago
Alibaba Stock Falls 10% After $10.2 Billion AI Share Placement
23 hours ago
Nvidia is currently ramping up shipments of its Vera Rubin systems, which started production earlier this year. The company is scheduled to report earnings on Wednesday.
Why this matters: Nvidia's deployment timeline and competitive token-per-second advantage position the Groq acquisition as a strategic response to emerging competitors targeting the specialized low-latency inference market. At 3,400 tokens per second versus Cerebras-powered alternatives at 750 tokens, Nvidia's Groq technology offers cloud providers material differentiation for premium service tiers, directly influencing margin potential for customers serving latency-sensitive AI applications.
Related coverage: Shein Hong Kong IPO Values Fast-Fashion Giant at $27 Billion
Related Guide: Read our complete guide →
What This Means
Groq's entry into production accelerates Nvidia's dominance in specialized inference hardware, potentially capturing premium segments of cloud AI spending. However, AMD's Cerebras integration and OpenAI's vertical stack strategy suggest sustained competition will pressure token-per-second benchmarks and pricing over the next 18 months, particularly for coding and real-time agent workloads.
Sources: CNBC and other international news outlets.
Disclaimer: This article was produced with AI assistance based on publicly available news sources. While we strive for accuracy, NewsOracle makes no warranty as to the completeness or accuracy of the information. Errors and omissions may occur. Readers should independently verify all information before acting on it. NewsOracle does not intend to defame any individual or organisation and accepts no liability for any loss or damage arising from reliance on this content. Content is for informational purposes only and does not constitute legal, financial, medical, or professional advice. All rights reserved. Unauthorised reproduction prohibited.
NewsOracle Editorial
The NewsOracle Tech Desk covers breaking technology news including AI, Apple, Google, Tesla, Meta, OpenAI and product launches.
Latest coverage: Nvidia
