Graphcore
Bristol, United Kingdom
Industry: Semiconductors
Region: UK 🇬🇧
Founded: 2016
Expected Valuation: $2.8 billion
IPO Date: 2023 (expected)
Graphcore Limited
Key Company Facts
| Founded | 2016 |
| Headquarters | Bristol, UK |
| Sector | Technology |
| Industry | Semiconductors |
| Founders | Simon Knowles and Nigel Toon |
| Number of employees | 650 (2022) |
| IPO Date | 2023 (expected) |
| Number of investors | 41 |
| Total funding | $682 million USD |
| Valuation estimate | $2.8 billion USD |
Company Overview & History
Graphcore is a British semiconductor company that specializes in the development of accelerators for artificial intelligence (AI) and machine learning (ML).
The company was founded in 2016 by Simon Knowles and Nigel Toon, and it aspires to create a massively parallel Intelligence Processing Unit (IPU) that holds the complete machine learning model inside the processor.
Graphcore’s first funding round in 2016 was led by Robert Bosch Venture Capital, with other backers including Samsung, Amadeus Capital Partners, C4 Ventures, Draper Esprit, Foundation Capital, and Pitango.
The company announced the world’s first graph tool chain designed for machine intelligence in the same year, called Poplar Software Stack.
In 2017, Graphcore announced their first chip, the Colossus GC2, which was first available in 2018.
This chip is a massively parallel, mixed-precision floating point processor that performs the same role as a GPU in conjunction with standard machine learning frameworks such as TensorFlow1.
The company’s second generation processor, GC200, was presented in 2020. It is a 59 billion transistor, 823 square-millimeter integrated circuit with 1,472 computational cores and 900 Mbyte of local memories.
In 2019, Graphcore’s C2 IPUs became available for preview on Microsoft Azure, and in 2023, Meta Platforms acquired the AI networking technology team from Graphcore.
Financial Performance
Since its inception, Graphcore has secured several rounds of funding. The company raised a total of $682M over 6 rounds by the end of 2020. Their Series A funding round took place in 2016, raising $30M.
This was followed by Series B and C in 2017, securing $30M and $50M respectively. In 2018, Graphcore’s series D funding raised $200M, achieving a $1.7 billion valuation and making the company a unicorn.
The Series E funding round in 2020 raised $222M, with the Ontario Teachers’ Pension Plan as a notable investor.
Revenue
| Year | Revenue (USD) |
|---|---|
| 2022 | $29.0 million |
Business Model
Graphcore’s business model revolves around the development and sale of its unique IPUs and associated software for machine learning and AI applications. This includes the Poplar software stack, which is a graph tool chain designed specifically for machine learning, and their IPU hardware, such as the Colossus GC2 and GC200 processors.
The company’s processors are designed to hold the entire machine learning model within the processor, promoting efficient data processing and performance.
In addition to direct sales, Graphcore’s IPUs are also available for use on cloud platforms, such as Microsoft Azure, providing another revenue stream for the company.
Estimated Valuation
| Year | Valuation (USD) |
|---|---|
| 2018 | $1.7 billion |
| 2020 | $1.95 billion |
Risk Factors
As of 2021, some of the risk factors for Graphcore included its status as a private company, which means its financial information isn’t as accessible or transparent as a publicly-traded company’s would be. Furthermore, the company is heavily reliant on the success of its primary product, the IPU (Intelligence Processing Unit).
As such, any technical or performance issues with this product could have a significant impact on the company’s reputation and financial health. Finally, Graphcore operates in a highly competitive and rapidly evolving industry, and its success depends on its ability to continue to innovate and keep pace with these changes.
Market Opportunity
Graphcore has shown significant market opportunity in the field of AI and machine learning. Their unique product, the IPU, is a parallel processor designed specifically for machine intelligence workloads and used globally in machine learning applications such as powering autonomous vehicles and robots.
The company’s IPU technology is aimed at outperforming legacy processors like GPUs, offering a unique selling proposition in the market. Furthermore, the company’s successful fund-raising rounds, with the latest Series E round bringing its valuation to $2.77 billion and total funds raised to over $710 million, indicate strong investor confidence and market opportunity
Competitors
- Nvidia
- Intel
- AMD
- Google (via their Tensor Processing Units)
- IBM
- Qualcomm
- ARM
- Xilinx