Global Tech Outlook has listed the top AI chip startups to look out for in 2021.
Just when AI is thriving in the global market with cutting-edge technology, startups are coming up with AI chips for multiple industries such as automotive, healthcare, manufacturing, aerospace and many more. These silicon chips are a combination of AI, machine learning, deep learning and IoT to solve computational as well as mathematical problems efficiently and help in the rapid decision-making process. These AI chips are used in data centers to enhance server optimisation and better security management. It is expected that the semiconductor industry will face a shortage due to the increasing demand for smart devices, smart cities, smart robots and so on. Thus, there is a tremendous growth of AI chip startups to work on on-chip memory and supply these silicon chips to industries. Millions of dollars are being invested in the semiconductor industry for the new generation of AI chips. Let’s explore some of the top AI chip startups to look out for in 2021.
Cerebras Systems has a mission to revolutionise compute for deep learning with a world-class team. Andrew Feldman and Gary Lauterbach are the Co-founders of this AI chip startup. The California-based startup debuted the new 2.6 trillion transistor Wafer Scale Engine-2 (WSE-2) AI chip that is a successor chip to the world’s largest chip, the WSE. It comes with 40GB of on-chip memory, 20 petabytes of memory bandwidth as well as 220 petabits of aggregate fabric bandwidth. Cerebras has successfully revolutionised the industry by launching WSE. WSE-2 can solve complex problems in much lesser time as it is the fastest AI processor in the world.
Groq is a California-based AI chip startup that has developed its first AI chip with customers from data centers, autonomous vehicle and US$67 million in funding over the last five years. It is planning on offering a second-generation AI chip that are 10 times faster than the other AI chips in the global market. The AI chips of Groq perform one quadrillion operations per second and the chip design involves a single enormous processor with hundreds of functional units. It caters for clients from financial services and autonomous vehicles.
Lightmatter is a Boston-based AI chip startup that is focused on developing photonic AI chips. It is one of the startups that use light instead of electrons where the photonic processors are not bound by Moore’s Law. It has also developed its first AI chip known as Envise to encode data with light. The encoded data is processed by the 256 photonic tensor cores on each Envise chip that consists of a photonic controller, a graph processor, 500 MB RAM as well as a network interface. The AI chips can perform usual functionalities like ReLU and GELU while supporting INT8, INT16 and other numeric operations. The CEO expects that the US$80 million will enable the customers to embrace AI and machine learning more efficiently.
SambaNova Systems is a Silicon Valley-based AI chip startup that develops semiconductors for AI computations. Recently, it has raised US$676 million from the third biggest venture capital deals in AI chips. The startup has developed its own AI chip architecture like ARM and x86 in smartphones or laptops. Kunle Olukotun and Chris Re are the Co-founders of SambaNova Systems to produce AI training and inference chips in 2017. It has become one of the most capitalised AI startups with over us$1 billion funding.