TECHnalysis Research Blog Extra

November 18, 2016
Intel Unveils Broad AI Vision

By Bob O'Donnell

At a special event in San Francisco, Intel debuted a sweeping new vision for the role it believes it can play in the rapidly evolving and highly topical field of Artificial Intelligence. The company put together an impressive set of messages that covered everything from definitions for the still little understood fields of AI, machine learning and deep neural networks, through silicon announcements, software unveilings, new customer partnerships and even a new sub-brand.

The company made clear that it believes the AI market is still in its infancy and that there are plenty of opportunities for it to make a very significant mark. The last point is important, because there’s been a great deal of press and attention to date on the role that GPUs can play in AI and deep learning, driven primarily by nVidia’s strong messaging work.

At Intel’s event, the company discussed a variety of different efforts they’re making to impact the AI market—an opportunity the company clears sees as being strategic to its long-term growth. On the silicon side, the company unveiled a new chip code-named Lake Crest, expected in the first of 2017, which uses the work done by Nervana Systems, the AI company that Intel purchased earlier this year. The new chip architecture is specifically optimized for deep learning algorithms and includes 32 GB of high-bandwidth memory (HBM2) and offers high-speed I/O and proprietary chip-to-chip protocols to handle very large deep neural network models.

Intel plans to use the Nervana sub-brand to help unify all its AI silicon and software efforts. Speaking of which, the company also described a complex set of software offerings that are designed to let data scientists pick from a variety of open source AI frameworks, including the company’s own Neon framework, which came as part of the Nervana acquisition. Essentially, Intel has created some core software that will optimize algorithms created in any of these frameworks to run quickly and effectively on a range of Intel hardware—from x86 CPUs, through Xeon Phi chips to Altera FPGAs and, eventually, to the Lake Crest family of AI chips.

In addition to the products, Intel announced several partnerships with companies such as Google and insurance company USAA to highlight their efforts. They also talked about a number of socially relevant efforts to use AI for good, such as working with cancer researchers and the Center for Missing and Exploited Children.

While the event had a bit of a “drinking from a firehose” burst of information, it’s clear Intel sees a strong opportunity for itself in AI moving forward.

Here's a link to the column: https://techpinions.com/unpacked-for-friday-november-18-2016/48044

Bob O’Donnell is the president and chief analyst of TECHnalysis Research, LLC a market research firm that provides strategic consulting and market research services to the technology industry and professional financial community. You can follow him on Twitter @bobodtech.