SambaNova are creating a new computing platform to power machine learning and data analytics, with both software and hardware. It is believed that this will outclass the traditional GPU in terms of processing speed and the size of the chip. By rethinking the hardware, SambaNova’s first generation chip could make AI operations far more efficient and faster for operations such as speech and image recognition.
HOW THEY’RE DISRUPTING
It has been discussed that Moore’s Law, a theory where transistors or ‘chips’ continue to shrink in line with the increase in technological advancement, is gradually slowing and will plateau. By 2025, it’s thought that we will need a new generation of more efficient computer chips. This is a positive move for AI, allowing companies like SambaNova to design AI platforms by building from the hardware up, for more natural machine learning.
By 2020 it’s predicted that AI market will be worth $1.2 trillion with an annual growth rate of 3.5%. Investments in AI have been on the rise in past years, with investment in start-ups increasing 6x since 2000.
INVESTMENTS AND FUTURE
SambaNova have raised $56 million in a round led by Google’s parent company, Alphabet. It’s the first investment the venture arm has made in AI, and signifies that companies are starting to see the need for specialised infrastructure for AI and large-scale data analytics.