Experts from Morningstar from the pages of the resource Tech Trader Daily are trying to convince the public that Intel has a good potential in the development of artificial intelligence technologies.
The authors of the analytical note explain that the work of artificial intelligence systems implies not only the training stage , when a huge amount of data is processed for analyzing events and phenomena that have already taken place, but also the stage of logical conclusions when new "introductory data" appears. The corporation has not only central processing units and Xeon Phi accelerators, but also programmable Altera matrices, which can be configured for certain narrow tasks. It is important to consider that different types of processors can offer different levels of power consumption, and NVIDIA GPUs will not always be able to demonstrate higher efficiency.
Specialized processors like the proposed Google TPU will find their application as the development of artificial intelligence technologies, and making companies bet on them will be a winner. Special processors may appear in smartphones, as some of the computing load will still be delivered to them. True, we personally doubt that Intel will succeed in developing such solutions for smartphones. But in the automotive segment, its programmable matrices can manifest themselves from the best side. If you believe the analysts, Intel may well succeed in the segment of artificial intelligence technologies, provided that it intelligently disposes of its potential.