Urgent.News

What's breaking now, across thousands of outlets.

AI

Nvidia's Groq acquihire is on the DOJ's radar, but it's already too late

Even if regulators did somehow unwind the $20B deal, there's a growing list of alternatives ready to take Groq's place, no merger required

Nvidia's Groq acquihire is on the DOJ's radar, but it's already too late

Nvidia's $20 billion acquisition of Groq, an AI accelerator startup, has drawn the attention of the US Department of Justice (DOJ). While the deal technically left Groq's core business intact, it was designed to evade regulators. The DOJ's probe questions whether Nvidia's acquisition, which stripped Groq of its engineering team, harmed competition.

Nvidia acquired Groq's mature silicon and talent necessary for continued development. Groq's SRAM-heavy dataflow accelerators enabled LLM inference up to thousands of tokens per second, outperforming GPU-based systems. Together, Nvidia's LPX racks and Groq's accelerators deliver optimal performance across various inference workloads. However, other companies like Cerebras, SambaNova, and d-Matrix are developing similar disaggregated compute architectures.

Critics argue that Nvidia's acquisition of Groq, though not a traditional merger, effectively bought the startup's technology and talent. While the DOJ may be able to undo the deal, it is likely too late to prevent the potential negative impact on competition. Even if the DOJ succeeds in blocking the acquisition, Nvidia's ecosystem, centered around its networking business, could still accommodate alternatives.

Written by urgent.news from The Register Science's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

This story

This is one outlet's version. Read the fullest account.

Read the original at theregister.com →

More in AI

NASA Uses AI to Read the Moon Before Humans Return to Its Surface

NASA and IBM Research have launched an open-source artificial intelligence model to extract new insights from nearly two decades of lunar data, aiding future crewed missions to the Moon. Announced Sept. 10, 2026, the NASA-IBM Lunar Foundation Model helps scientists map craters, locate potential ice deposits, and identify geological…

One Python Agent Core, Four Ways to Run It: Nova on Server, Web, TUI, and Desktop

Nova is an open-source personal AI agent runtime for developers. One Python core is available through terminal, web, desktop, and API, so you get a single local workspace for model providers, tools…

  • Nova is an open-source AI agent runtime with a single workspace.
  • It provides four interfaces: terminal, web, TUI, and desktop.
  • The agent core drives all four interfaces with consistent loop and storage.

More from Saturday 12 September →