Groq Brand Guidelines: The LPU Inference Chip & a 2025 Licensing Deal With Nvidia

Former Google engineers with a single-purpose chip.

Groq was founded in 2016, according to English Wikipedia and a December 2025 report by Constellation Research. English Wikipedia says it was started by a group of former Google engineers led by Jonathan Ross, one of the designers of Google’s Tensor Processing Unit, an AI accelerator chip, together with Douglas Wightman, a former engineer at Google’s X research lab, who served as the company’s first chief executive. It gives the headquarters as Mountain View, California, with further offices in San Jose, Toronto and London among others; the company’s releases from 2026 are datelined San Francisco.

The company makes an application-specific integrated circuit for running AI models once they have been trained, the work known as inference. English Wikipedia describes the first chip as a 14-nanometre design measuring 25 by 29 millimetres. It was introduced as the Tensor Streaming Processor, codenamed “Alan.” After large language models spread with ChatGPT, English Wikipedia says, Groq’s vice-president of brand, Mark Heaps, and Ross renamed it the Language Processing Unit, or LPU, to make the nature of the processor more obvious.

The architecture itself is unusual. English Wikipedia describes the LPU as a single-core, deterministic design in which memory units are interleaved with vector and matrix units, and in which the compiler, rather than reactive hardware such as caches and branch predictors, controls every step of execution. Constellation Research quoted Groq in December 2025 as saying the LPU holds hundreds of megabytes of on-chip memory as its main store for model weights.

Chips, a cloud and a racing car.

The company was funded privately. English Wikipedia says it received seed funding in 2017 from Chamath Palihapitiya’s Social Capital, raised a Series C round led by Tiger Global Management and D1 Capital Partners in April 2021, after which it became one of the start-ups valued above a billion dollars, and raised a Series D round led by BlackRock Private Equity Partners in August 2024.

Groq grew through acquisitions as well as its own chips. On 1 March 2022 it acquired Maxeler Technologies, a company known for dataflow systems, which English Wikipedia says kept its own brand. In August 2023 Groq chose Samsung’s foundry in Taylor, Texas, to make its next-generation chips on a 4-nanometre process, which English Wikipedia describes as the first order for that new factory; it says the second-generation LPU will be built there.

In February 2024 the company soft-launched GroqCloud, a developer platform for renting access to its chips through an API, and on 1 March 2024 it acquired Definitive Intelligence, a business-software start-up, to help build that platform. In February 2025 Groq announced a commitment from Saudi Arabia tied to a new GroqCloud data centre in Dammam. By 2026 the company said it operated 13 data centres across North America, Europe, the Middle East and Asia-Pacific, and that more than six million developers had built on GroqCloud.

The brand has also appeared in motorsport. On 26 September 2025 McLaren Racing named Groq an Official Partner of the McLaren Formula 1 Team, and the joint announcement said the Groq logo would appear on the team’s race cars from the 2025 Singapore Grand Prix. English Wikipedia adds that Groq hosts open-source language models running on its LPUs for public access, through its website and a playground for developers.

How AI chip makers show themselves.

AI chip companies sell something most of their customers will never see. The processor sits inside a server, the server sits in a rack, and the rack sits in a data hall far from the people who use the models running on it. So the industry’s visual language has grown up around making that hidden hardware visible: a whole silicon wafer, as wide as a dinner plate, held up on a keynote stage; a single chip pinched between finger and thumb for the cameras; a die shot recoloured in neon false colour so that its blocks of memory and compute read like a city map at night.

The second set of images is the data centre. Press photographs tend to show long rows of racks in dark halls, lit by the glow of status lights and cold aisles, with cabling arranged as neatly as a diagram. These pictures stand in for scale, which is what buyers of AI compute are being asked to imagine, and they appear again and again in launch announcements, partner news and investor material.

Keynote slides do much of the remaining work. Model names, token counts and benchmark charts appear beside the chip’s own product name, and the companies’ identities are usually kept spare to let those numbers and images lead: black or white grounds, a single accent colour, a sans-serif wordmark and very little ornament. Several firms have adopted coined product categories, names for their processors that set them apart from the graphics processor.

The shop window is the technical conference. Hot Chips, NeurIPS and the AI Infra Summit are among the events where architectures are presented, demonstrations run and booths compete for engineers’ attention. This guide looks at how each company in the group presents itself, and keeps to what can be checked about who founded it, who owns it and what it makes.

A licensing deal, and a company that carried on.

On 24 December 2025 Groq announced a non-exclusive licensing agreement with Nvidia for its inference technology. Under it, the company said, Jonathan Ross, its president Sunny Madra and other members of the team would join Nvidia to help advance and scale the licensed technology, while Groq would continue as an independent company with Simon Edwards as chief executive and GroqCloud would keep running. Constellation Research reported the same day that CNBC had first described the deal as a purchase of Groq, and that the companies presented it as a licence. Groq said in June 2026 that Nvidia had announced, at its GTC conference that year, a next-generation LPX platform incorporating Groq’s inference technology.

The company then refocused on its cloud. In June 2026 Groq announced new growth capital led by the investors Disruptive and Infinitum; in August 2026 it said it had joined Nvidia’s Cloud Partner programme and would be among the first to deploy the NVIDIA Groq 3 LPX, working with Dell Technologies. On 17 August 2026 Groq LLC announced a Series A round led by Disruptive, with planned participation from Nvidia; the company said that round was subject to customary closing conditions, and its completion had not been announced at the time of writing.

For a design guide, Groq shows how renaming a product — from a streaming processor to a language processing unit — can tie a chip to the moment its market arrived. Compare Nvidia, which licensed Groq’s technology, and Cerebras, another inference company that sells its chips as a cloud service. For this guide we use black, a warm grey and an ember with a sans, an unofficial interpretation; no product or logo is reproduced.

Keep Groq’s identity on hand — save it to a board in the Made Good inspiration library.