Skip to content

Groq 3 LPX

Nvidia's rack-scale inference accelerator built on Groq's LPU (language processing unit) technology, aimed at low-latency decode alongside Nvidia GPUs.

Known aliases

  • Groq 3 LPX rack
  • Groq LPU
  • LP30
  • LPX
  • LPX rack
  • NVIDIA Groq 3 LPX

Current stories

investConfirmed6 publishers

Former Groq engineers ask Delaware to treat Nvidia's $17 billion license as a sale of the company

Two former Groq engineers sued the startup's board in Delaware, saying Nvidia's $20 billion license-and-hire deal was a sale that needed a stockholder vote. A ruling for them would extend that vote to other deals built the same way.

Perspective Coverage

7 publishers
Builder
Builder 19%
Operator
Operator 33%
Investor
Investor 48%

Reality

Evidence60
Adoption35
Hype gap+20
Incentives70
Confidence58
investOne report1 publisher

Nvidia's guide adds $11.8 billion after a quarter that added $14.7 billion

Nvidia reported $96.2 billion of revenue at 75% gross margins and guided the next quarter to $108.0 billion while assuming nothing from China Data Center compute, with gross margin slipping to 74%.

Publishers:investor.nvidia.com

Reality

Evidence62
Adoption72
Hype gap+25
Incentives80
Confidence74
investOne report1 publisher

DOJ weighs Nvidia's $17bn Groq licence against a $119.5m notification threshold

The remedy people familiar with the probe describe is a penalty rather than an unwind, and nobody has put a figure on it, which leaves the cost of routing a $17bn deal around merger review still blank.

Publishers:techtimes.com

Reality

Evidence45
Adoption68
Hype gap+22
Incentives62
Confidence52
productOne report1 publisher

Swapping out the GPU leaves four more rack lines on Nvidia's invoice

Nvidia's Vera Rubin lineup sells a CPU, an inference part and storage and networking racks around the Rubin GPU, so a custom accelerator bids on one line of five while the efficiency argument moves to the data path.

Publishers:techcrunch.com

Reality

Evidence38
Adoption22
Hype gap+32
Incentives76
Confidence42
buildOne report1 publisher

Nvidia's Groq-derived LPX rack posts 3,431 tokens/sec on 128GB of SRAM

The first third-party benchmark of the LP30 rack came in at roughly four times the next-fastest public endpoint, measured one request at a time on a model small enough to fit.

Publishers:tomshardware.com

Reality

Evidence58
Adoption20
Hype gap+32
Incentives74
Confidence55