Join Transform 2021 for the most important themes in enterprise AI & Data. Learn more.
Intel announced late last week that it has formed a new AI group to consolidate a number of its programs and acquisitions. It’s headed by Naveen Rao, the former head of Intel acquisition Nervana. This means Intel is making sure is has a major seat at the table as artificial intelligence and machine learning branch out to touch virtually everything — from autonomous driving to IoT to enhancing corporate systems — over the next 5-7 years. (Full disclosure: Intel is a client of mine.)
In the short term, the group will focus on research related to its software and hardware (Nervana, Xeon/Lakecrest chips and subsequent families) to deliver AI for drones and autonomous vehicles, smart cities, health care, personal appliances, etc. But I expect a longer-term play. Intel will be putting together a complete set of products to bring to traditional manufacturers in its chip business across the full breadth of edge computing to big data center platforms for things like Xeon Phi based AI solutions that power financial models, biological research/modeling, and scientific research. For example, plans for Intel’s Mobileye and Nervana acquisitions will fall under this group’s charter. Mobileye is known as an enabler of vision for cars and drones, but its vision based models, together with AI and ML could prove highly valuable in fields like security, visual analysis of health-related issues, and real-time forecasting.
Additional technologies like Saffron (with its AI-as-a-Service model), which Intel acquired in 2015, have already started to give the Intel visibility into customer needs, enabling it to learn about future requirements and experiment on solutions. Nervana gives Intel the ability to experiment with optimized architectures in addition to Xeon Phy massively parallel processing systems that will power next generation AI systems. And Intel’s close alliance with the research and open source communities gives it the ability to learn as well as influence future directions in AI/ML. But the company will also deploy customized solutions through its consulting services and will eventually offer these as off-the-shelf enterprise cognitive computing toolsets for verticals like financial modeling, genetic research, transportation best routing, etc. The ultimate target: compete head to head with IBM Watson in particular and other up and coming solutions in general.
IBM is on a different path. In the current early phase of the emerging AI market, IBM has a lot of momentum. It has signed agreements with high profile companies (a partnership with Salesforce, for example) and acquired its own stable of AI related properties (such as Weather.com). IBM is investing heavily in Watson and related service offerings and has devoted substantial resources to make it successful. It’s highly likely IBM will be making additional acquisitions in this space over the coming 1-2 years. IBM considers Watson its primary general purpose platform for AI, much as it did with Websphere for web services, and is building on top of this platform with customized solutions through its service organization and through partnerships. And it will acquire complementary technology as required.
But Intel thinks IBM has a distinct challenge that it can exploit. Watson is taking a wide approach to cognitive computing by trying to do everything for everyone. It does have some vertically focused solutions, but IBM, as it has in the past with other products, is positioning Watson as a general purpose capability. This is a difficult road to take. Intel’s more focused approach could ultimately prove more successful. By looking for specific needs and “templates” to make neural networks do specific things that are then optimized for its hardware assets, Intel thinks it can beat the other contenders with targeted solutions fully optimized for specific markets and tasks.
Intel is leveraging standards and open source tools rather than developing proprietary solutions. It’s working with Tensorflow, the open source ML library developed by Google, and Neon, the open source deep learning language and library based on Python, which will be optimized for Intel hardware but remain open sourced. Indeed, Neon even runs on GPUs from Nvidia, a key rival. Intel is offering some of its own innovations to the open source community and is building relationships with key players like Google and Microsoft. This is much more aligned with Intel’s past experiences and business practices and should provide it with a model it can comfortably implement.
I expect Intel to make a number of additional acquisitions over the next couple of years in this rapidly evolving market. A piece-meal approach to specific problems is actually more in line with most business needs. Intel looks for targeted vertical solutions to solve defined business challenges rather than deploying broad-based systems. And it can productize these targeted offerings to be used by partners/solutions providers — its traditional channel.
The new Intel AI group has maximum focus as it reports directly to CEO Brian Krzanich. There is no doubt this group is critical to Intel’s future and will be a major revenue generator in the next 3-5 years. AI can touch nearly every product line, merging with existing capabilities to move to the next step incrementally. But it’s hard to fully comprehend what computing architectures are needed for AI. Taking standard CPU and GPU computer architecture and shoe horning them for AI may not be enough, which is why Intel has acquired technology like Nervana chip architectures due out this year. Intel has also made a Cloudera investment, looking to roll that into Xeon chips and other big data capabilities in the future. Where the architectures go over the next few years is still unknown but could be very compelling.
Bottom line: with this commitment, Intel is clearly signaling it does not intend to be left behind in AI. In fact, it expects that, with its extensive resources, it can be both a major implementer of new technologies/research, as well as a key contributor to better market understanding, which will give it a leg up in knowing what’s needed to power the next generation of its computing systems and reshape its business longer term.
Intel has laid down the challenge, and AI-focused solutions companies like IBM Watson and hardware companies like Nvidia should take heed.
Jack Gold is founder and principal analyst at J.Gold Associates, LLC., a technology analyst firm based in Northborough, MA., covering the many aspects of business and consumer computing and emerging technologies.
VentureBeatVentureBeat's mission is to be a digital town square for technical decision-makers to gain knowledge about transformative technology and transact. Our site delivers essential information on data technologies and strategies to guide you as you lead your organizations. We invite you to become a member of our community, to access:
- up-to-date information on the subjects of interest to you
- our newsletters
- gated thought-leader content and discounted access to our prized events, such as Transform
- networking features, and more