At Lenovo’s Transform event in June, Lenovo DCG leader Kirk Skaugen made the proclamation that the company’s goal was to be #1 in the HPC (high-performance computing) market. While details on the “when” and “what measured” weren’t clear, it was gutsy, got everyone’s attention, and I liked it. This week, the largest HPC show and likely now the largest AI and machine learning show, SuperComputing 2017 (SC17), is happening in Denver, CO. SC17 is where Lenovo would need to emphatically make their point they are the vendor to beat in HPC.
As part of Lenovo’s broader $1.2B AI investment, Lenovo’s Data Center Group announced today at Supercomputing 2017 a number of initiatives geared towards making AI more accessible for customers looking to transform “business and society”. It’s good to see Lenovo making headway in HPC and AI, but more interesting to me the leverage the company seeks to bring from their HPC AI efforts to the enterprise, specifically manufacturing, healthcare and research. HPC is a great attention-getting market, but I believe the new Lenovo DCG profit opportunity for enterprise AI is much larger.
Here’s my rundown of the announcements made.
A “multi-pronged” AI strategy
Adopting AI solutions can be intimidating for many organizations, requiring large amounts of technical know-how. By outlining these three steps, Lenovo is trying to show customers that with the right partner, it doesn’t have to be all that scary. I like Lenovo’s holistic strategy here as enterprises are new to AI, they need to be spoon-fed, and no vendor has ownership of the space, so Lenovo has some space to take customers. Enterprises rarely change vendors, but when they do, it’s in market shifts like “Web 1.0”, “big data”, “cloud” and now “AI”.
New hardware and software AI offerings
Lenovo announced a couple new AI offerings to its HPC portfolio, designed for future AI workloads and to attract customers looking to transform through AI. First, Lenovo announced new hardware, new GPUs for ThinkSystem SD530, which includes support for both the Intel Xeon Scalable Systems family of processors (Purley) and NVIDIA’s Tesla V100 (Volta) GPU accelerators. Lenovo says that the ThinkSystem SD530’s power and scalability make it an ideal system for HPC, and that the Tesla V100 will make it well-suited for machine learning training workloads at scale and inference. Given the performance and form factor (up to two GPUs for two CPU/two GPU nodes per 2U enclosure) and I/O, I agree, this is optimized for AI.
Also announced was the Lenovo Intelligent Computing Orchestrator (or LiCO), a management suite that Lenovo says will accelerate development of AI applications. To this end, LiCO boasts several crucial features: it includes popular open-source AI frameworks and is capable of scheduling AI workloads in multi-project environments, monitoring neural network training, and working across various solution providers. I cannot tell you how important software is when it comes to machine learning. I believe Lenovo will shrink “time to discovery and insight” by lessening the time required to experiment, optimize, deploy and manage AI. Most enterprises are spending their AI time in the weeds getting things setup versus actually experimenting and solving real problems with AI, so this help a lot.
Announcing new innovation centers
Another part of Lenovo’s multi-pronged AI approach is that they have opened three new AI innovation centers in Morrisville, North Carolina, Stuttgart, Germany, and Beijing, China. At these centers, Lenovo says that over 100 scientists and AI developers are working together to develop new AI solutions for customers. Conveniently, Lenovo customers don’t actually have to be in those three cities to take advantage of the benefits—they can access the centers’ services remotely. Customers also have the ability to test and refine their AI workloads and apps across a myriad of high-performance systems. Lastly, Lenovo says these centers will serve as an intellectual hub for AI—a community of partners, data scientists, and customers, all coming together to share valuable insights. Similar to LiCO, these innovation centers can reduce the time it takes customers to go from experimentation to deployment plus removing the need to buy the hardware in those early phases.
At the SC17, Lenovo, like many other vendors, went into detail about its partnership with their premier customers. North Carolina State University talked about they are doing together in agriculture and water management. They have jointly developed an AI-enabled geospatial image analysis process to help recognize farmland, identify the farm crops, monitor soil conditions, and calculate water requirements to design maps of drought areas and better manage resources. Lenovo says that these techniques have the potential to improve profitability for farmers across the globe, while simultaneously conserving the world’s single most precious natural resource.
Another case study Lenovo showcased is the company’s collaboration with University College London (UCL). You may have heard of the ATLAS particle collision experiments being conducted at CERN’s Large Hadron Collider—researchers at UCL are trying to recreate these events, and are using Lenovo’s AI capabilities to reconstruct particle trajectories using imaging data. Lenovo says that using AI in this manner streamlines computation resources, and allows researchers to model much more complex events than they would be able to use traditional methods.
All of Lenovo’s SC17 announcements indicate to me that the company is taking its goal of #1 HPC market and accelerating AI very seriously starting with its $1.2B investment. With its three-step customer-centric approach, it is positioning itself as a premier company that can make AI easy for organizations looking to level up and transform their business. Additionally, I believe that with the new GPUs for ThinkSystem SD530, and the LiCO management suite, Lenovo is making good strides towards having all the hardware/software needed to back up all of this big talk. Hardware is important, but software is the biggest issue right now. I love the concept of the new AI innovation centers being a hub for sharing and disseminating industry insights, and I love that customers will have remote access to the resources as well. I think Lenovo is on a good track, and I’ll continue to watch with interest to see how well this AI strategy and offerings pan out.