Nvidia's $20 Billion Groq Bet Goes Live Ahead of Earnings

Nvidia announced Monday that its Groq 3 LPX rack has entered full production, marking the commercial debut of technology from the company's largest acquisition on record — the $20 billion purchase of chip startup Groq's assets last December. The Groq rack will be deployed alongside Nvidia's Vera central processors and Rubin graphics processors at cloud provider Nebius later this year, according to Nvidia senior director Dion Harris. The launch lands just two days ahead of Nvidia's quarterly earnings report on Wednesday, a release investors are watching closely for signs of continued AI infrastructure demand.

Each Groq 3 LPX rack packs 256 individual Groq chips, built with 500 megabytes of on-die SRAM to cut memory bottlenecks, and can process 3,400 tokens per second according to a benchmark from Artificial Analysis. Samsung manufactures the Groq chips, while Taiwan Semiconductor continues to make Nvidia's core GPUs. The bet is on low-latency inference — the speed at which AI models respond to a query — which is becoming critical for AI agents and coding tools that need to feel instantaneous rather than laggy. Nvidia says cloud providers can charge premium rates for this faster tier of service. The company isn't alone in chasing this niche: AMD has said it will integrate Cerebras systems for the same low-latency inference market, and OpenAI's new Ultrafast mode, powered by Cerebras, currently promises 750 tokens per second. CEO Jensen Huang has projected $1 trillion in combined sales from current Blackwell chips and the new Vera Rubin systems through 2027, and said he'd allocate roughly a quarter of data center capacity earmarked for coding workloads to Groq chips — with the rest running on Vera Rubin.

For investors, this is a reminder that the AI trade is fragmenting into specialized hardware niches beyond the general-purpose GPU, and Nvidia is moving fast to own more of that territory rather than cede it to challengers like AMD and Cerebras. The Groq commercialization also gives Wednesday's earnings report extra weight: it's tangible evidence Nvidia's biggest-ever acquisition is generating real deployments, not just sitting on the balance sheet. Investors holding Nvidia, or considering AI infrastructure plays more broadly, should watch Wednesday's numbers for commentary on Groq adoption and data center demand as a signal of whether this diversification strategy is paying off.