Groq

Groq is an AI infrastructure company whose business centers on inference: running trained AI models to produce outputs for products, customers, and applications. The company describes inference as a growing bottleneck as demand for AI increases, and positions its services around making inference fast, reliable, and affordable at scale. Groq says it pioneered the LPU, its purpose-built inference technology. Its offering extends from silicon to cloud: the company provides inference infrastructure and services, and says its integrated stack is anchored by bare-metal infrastructure for production-ready workloads, alongside enterprise governance and control. Groq also presents LPX as a way for its technology to work alongside NVIDIA’s next-generation GPUs. The company reports that more than five million developers and thousands of AI-native companies use Groq, running trillions of tokens each week. It says it operates data centers across North America, Europe, the Middle East, and Asia-Pacific, and is building additional capacity. Groq’s current public positioning is as a specialized neocloud for inference rather than a general-purpose consumer brand. The supplied company page identifies Adam Winter as CEO and says he joined Groq in 2024 to lead its international business before becoming CEO in 2026.

Groq’s history, as described in the supplied company materials, is tied to the development and deployment of its LPU inference technology. The company says it took the LPU from silicon into a global production cloud, combining expertise in inference operations, hyperscale infrastructure, and enterprise software. The materials do not give a founding date, identify a founder, or set out earlier corporate milestones. Groq’s public narrative emphasizes the growing importance of inference as AI use expands. Its homepage contrasts training, which creates the possibility of AI applications, with inference, which creates value when models are used. It presents the increasing amount of inference required by AI products and tasks as a capacity challenge, and says Groq was built to address it. The company describes its goal as making inference work at scale, with reliability and affordability as well as speed. The company’s offering is described as extending from silicon to cloud. Groq says its integrated stack is anchored by bare-metal infrastructure that powers production-ready inference, with enterprise governance and control. It also identifies LPX as a development that works alongside NVIDIA’s next-generation GPUs. The supplied materials do not specify LPX release timing or technical specifications, so these details are not established here. Groq reports that more than five million developers and thousands of AI-native companies use its services, running trillions of tokens each week. It says it operates data centers across North America, Europe, the Middle East, and Asia-Pacific, and is building hundreds of megawatts of capacity, with further capacity planned. These are company-reported figures and statements; the supplied material does not provide an independent measurement or dated breakdown. The company page identifies Adam Winter as CEO. It says he joined Groq in 2024 to lead its international business and became CEO in 2026. The same page identifies Matt Eng as CFO and Alan Rice as COO. It describes Winter’s background as spanning technology, cloud, cybersecurity, and AI, and notes his earlier tenure at Cisco. The available evidence does not document other leadership transitions or a fuller chronology of Groq’s corporate development.

• 2026

  Adam Winter becomes CEO

• 2024

  Adam Winter joins Groq

A specialized neocloud for fast, reliable, affordable inference at scale, built around an integrated stack spanning silicon, bare-metal infrastructure, and cloud services.

• LPU

  Groq's inference technology, which the company says it pioneered and took from silicon to a production cloud.

• Groq inference cloud

  Cloud-based inference services presented as part of Groq's integrated infrastructure stack.

• LPX

  A Groq offering that the homepage says works alongside NVIDIA's next-generation GPUs to deliver inference capability.

• Focus the business on inference at scaleOther

• Unconfirmed

  Reported2026.08.25

  Nvidia ramps production of Groq 3 LPX rack, launching later this year[4]英伟达Groq 3 LPX机架量产，今年晚些时候上线163.com

  Nvidia is accelerating production of its Groq 3 LPX rack, which packages 256 Groq 3 chips and is claimed to reach 3,400 tokens per second in an Artificial Analysis benchmark. Nvidia senior director Dion Harris said the Groq rack will be deployed alongside Vera CPUs and Rubin GPUs on Nebius's platform and go live later this year.

• 2021.12.09

  Reported2021.12.09

  Groq appoints Ford Tamer to its board[5]Groq 任命 Ford Tamer 为董事会成员prnewswire.com

  Groq announced the appointment of Dr. Ford Tamer to its board of directors on December 9, 2021.

• 2021.02.25

  Reported2021.02.25

  Groq appoints Andy Rappaport to its board[6]Groq任命Andy Rappaport加入董事会prnewswire.com

  Groq announced the appointment of Andy Rappaport to its board of directors.

• 2019.11.12

  Reported2019.11.12

  Groq appoints Stuart Pann and adds senior leaders[7]Groq任命Stuart Pann并增聘高级管理人员prnewswire.com

  Groq announced that Stuart Pann would join its board of directors and that Adam Tachner and Bill Leszinske were appointed to vice president roles, while Sander Arts became a strategic marketing advisor.

Read more

1. 1Groq is the premier neocloud for fast inferencegroq.com
2. 2Company | Groq is the premier neocloud for fast inferencegroq.com
3. 3Groq Incwikidata.org
4. 4英伟达Groq 3 LPX机架量产，今年晚些时候上线163.com
5. 5Groq 任命 Ford Tamer 为董事会成员prnewswire.com
6. 6Groq任命Andy Rappaport加入董事会prnewswire.com
7. 7Groq任命Stuart Pann并增聘高级管理人员prnewswire.com
8. 8200亿美元收购成果落地：英伟达Groq机架已全面量产，响应速度快4倍163.com
9. 9Groq Adds Former Amazon Leader Raju Gulabani to Board of Directorsprnewswire.com