Unleashing the full potential of AI through AI hardware
"As Artificial Intelligence becomes more ubiquitous, hardware is playing a more significant role in the advancement of AI. This is a huge opportunity for the semiconductor industry, but also comes with pains that the industry needs to address."
Major trends driving AI evolution
While software drove the initial burst of advancements in AI, the parallelism of hardware and software development will achieve the next major AI breakthroughs.
Although the growth in AI creates tremendous demand for semiconductors and puts them in a position of power, the industry will continue to experience growing pains.
Demand for compute power far outstrips its supply and the industry’s current business model is not well suited to address this imbalance.
New devices and materials such as carbon nanotubes, graphene and spintronics.
New architectures and packaging such as 3D packaging, optical interconnects, and reconfigurable computing.
"The larger the neural network, the better the results. But, at the cost of energy. We relentlessly improved energy efficiency of ML platforms and in some cases yielded orders-of-magnitude, with no signs of the slowing down."