A UK-based startup aiming to compete against Intel and Nvidia in designing chips for artificial intelligence (AI) applications has raised $30 million in funding in order to launch its new processors.
The company, which hopes to disrupt the current chip market, is called Graphcore and is planning to ship its chips to companies in a wide variety of fields including cloud computing and driverless cars sometime next year. The company is claiming that its processors will be up to 100 times faster than its competitors when it comes to teaching AI systems how to learn. Not only that, the chips will also be more energy efficient.
Today's machine learning applications utilise high-end graphic processing units (GPUs) in order to handle the vast number of computations they perform. Nigel Toon, Graphcore's CEO, believes that this will not be the case going forward and that his company's chips will power these applications in the future.
Toon explained how GPUs are not ideal for machine learning, saying: “GPUs have been built to run programs that completely describe the algorithm. Machine learning is different. You are trying to teach the system using data and that requires a different style of compute.”
This is why Graphcore has developed a new type of chip it calls an Intelligent Processing Unit (IPU) that will come to market in 2017. The company has spent two years working on this new technology with the goal of reducing data centre costs while at the same time making them more efficient. Graphcore's chips could be used to not only run a data centre, but also to train AI systems during a server's downtime.
Deep learning is a quickly growing market that will see continued growth going forward, with the marketing intelligence firm Tractica predicting that $41.5 billion will be spent on the hardware needed to power deep learning projects by 2024.
Toon offered more details regarding the companies current investors, saying: “We have Bosch as a strategic investor, we have Samsung as an investor and Bosch is interested in autonomous vehicles and the next generation of transportation, while Samsung is interested in missing word edge of network devices. We are working with partners on some these other applications.”
Image Credit: Wichy / Shutterstock