23 Mar Intel Forms New AI Group Reporting Directly To CEO Brian Krzanich
As we have been writing for a while now, artificial intelligence will transform pretty much everything we do in our lives in the next five years. This isn’t some new technology early on the hype curve. AI is actually in use today helping us to match faces, identify photos, videos, the spoken word, doing our taxes, improving collaboration, and even help in healthcare diagnosis, and soon, will help drive our cars and trucks for us. While AI has been around for a while, the big breakthrough was machine learning using deep neural networks that actually got smarter with more information you threw at it.
GPUs, with NVIDIA being the biggest recent beneficiary, have become the most recent standard for cutting edge deep neural network training, and inference today is spread across CPUs, GPUs, FPGAs, ASICs and even DSPs. AI is a quick moving target and I think it’s unwise to think the engines today will be static in the future. The engines that drive AI are a very competitive space and not only are we seeing startups engaging, but the largest semiconductor companies like Intel.
Intel has been on an absolute tear in their goal to be the leader in AI. It acquired Altera for $16B and FPGAs are key to DNN inference. The company just announced the intention to purchase Mobileye, an ADAS supplier using primarily LiDAR with an AI roadmap, for roughly $15B. Intel bought Nervana Systems, too, a leading AI startup we had been covering for a while. And of course, we cannot forget their massive Intel investments in Xeon and Xeon Phi.
What Intel needs to do now is stitch these investments together into a cohesive and leading roadmap that gets executed flawlessly, on-time, with an investment in the future. Intel has already laid out their AI roadmap at their AI day both Karl Freund (MI&S AI analyst) and I attended in November 2016. Here we saw how Xeon, Xeon Phi, Nervana and Altera would be stitched together. Ostensibly, these came from four different product groups and companies who were recently acquired. If you have had to ever deliver silicon with enabling software, you can imagine how challenging a pan-group roadmap could become at such a large company. And then add on top of that with the rapidly changing AI landscape, the execution blinders cannot be put on, so you need to be scanning, anticipating and engaging in the AI future. That brings us to today’s announcement from Intel.
Intel announced it is aligning all its AI efforts under Naveen Rao, former Nervana Systems CEO, in a pan-Intel organization called the Artificial Intelligence Products Group (AIPG). Rao and the AIPG group will report directly into Intel CEO Brian Krzanich.
AIPG will “own” the execution for certain deliverables and the overall AI strategy and roadmap and will also “align” resources needed from other groups to have an AI roadmap spanning Intel Xeon, Intel Nervana, and Intel Xeon Phi and software stacks. AIPG will also include an applied AI research lab that will be looking out three to five years for additional building blocks, like new algorithms, architectures.
I like this move Intel is making……and Intel needs to move quickly. Its AI roadmap spans many different groups with many different P&Ls and quick decision-making, improved execution and resource allocation is paramount. While Rao of course will work collaboratively with his peers, he still has to “borrow” resources to deliver an AI roadmap, and has a direct link to Krzanich to escalate issues and get quicker resolution. It’s also natural in “blinders execution mode” to focus on what’s directly in front of your face. I’ve been there- it’s easy to do. Therefore, it would be easy to miss the rapidly changing AI space. Two weeks ago, I attended Google’s GCN 17 event where they said they use 4,000 different AI algorithms… and the change is increasing. This is where the applied research group comes into play, to keep tabs and in some cases influence and drive industry change for the next five years. It’s also a nice pipeline to attract top talent.
In the end, Intel’s AI success will come down to market share and the financials, but before that, the company needs developers and customers to get really excited about what they are doing in AI. And to do that, roadmaps matter, but executing on those roadmaps and delivering compelling price performance platforms is what really matters, which is just what Rao and the new AIPG intend to do.