- New hardware and software product announcements confirm NVIDIA's push to HPC and AI accelerators.
- The new products put "more dollars into computing instead of vast infrastructure overhead," according to the company.
- High growth is expected for both HPC data centers and embedded AI accelerators for self-driving cars.
NVIDIA (NSDQ:NVDA) is leveraging its leadership in the Graphic Processing Unit (GPU) market to address new targets, including the high-growth High Performance Computing (HPC) and Artificial Intelligence (AI) markets. NVIDIA's GPUs are replacing conventional Central Processing Units (CPUs) in HPC data centers and embedded smart hardware.
Initially focused on scientific computing applications including material science, aerospace, and drug design, HPC is now entering the world of big business through emerging applications of AI to speech and image recognition, natural language processing, and smart search. Tech giants such as Facebook (NSDQ:FB), Alphabet Inc-A (NSDQ:GOOGL), and Apple (NSDQ:AAPL), deliver the power of AI to consumer PCs and smartphones through huge HPC data centers, which must be cost-optimized. These companies don't manufacture the production hardware used in their data centers, and can be expected to rely on the products of the market leader, which is NVIDIA.
In the automotive sectors, the manufacturers of self-driving cars, including Alphabet, Apple, and many traditional car makers, must integrate powerful and cost-effective HPC and AI on-board subsystems for context recognition and navigation, to analyze the surroundings of the car and drive themselves safely in the traffic. Therefore, autonomous driving is another promising high-growth market for NVIDIA.
At the annual ISC High Performance Conference in Frankfurt, Germany, NVIDIA unveiled its Tesla P100 GPU accelerator. Built on NVIDIA's Pascal GPU architecture, the device is able to reduce HPC costs by 70 percent, and a single Tesla P100-powered server delivers higher performance than 50 CPU-only server nodes on standard benchmark tests.
"Accelerated computing is the only path forward to keep up with researchers' insatiable demand for HPC and AI supercomputing," said Ian Buck, vice president of accelerated computing at NVIDIA. "Deploying CPU-only systems to meet this demand would require large numbers of commodity compute nodes, leading to substantially increased costs without proportional performance gains. Dramatically scaling performance with fewer, more powerful Tesla P100-powered nodes puts more dollars into computing instead of vast infrastructure overhead."
Forbes notes that the automotive and data center market segments have been especially strong in NVIDIA's first quarter earnings, driven in large part by demand for hardware accelerators for AI applications based on neural networks and Deep Learning (DL).
NVIDIA's traditional product line made the company a strong leader in the Graphic Processing Unit (GPU) market - essentially, graphic accelerators for video-intensive games. Now, the company is diversifying and targeting the HPC data center and automotive markets. These segments delivered revenues of $143M and $113M in the first quarter, with a 63 percent and 47 percent year-over-year growth respectively.
The similarities between GPUs and HPC/AI accelerators, both of which rely on intensive parallelization - running many elementary operations at the same time - indicate that the new targets are a natural evolution of the company's traditional business.
On the software side, NVIDIA launched at ISC High Performance three additions to its DL software platform.
The new version 5.1 of NVIDIA cuDNN, a set of building bloc ks used by the leading deep learning frameworks, delivers accelerated training of deep neural networks. With the new GPU Inference Engine (GIE) for production environments, cloud service providers can more efficiently process images, video and other data in their HPC data centers. Automotive manufacturers and embedded solutions providers can deploy powerful neural network models with high performance in their low-power platforms.
NVIDIA DIGITS 4 introduces a new object detection workflow, enabling data scientists to train deep neural networks to find faces, pedestrians, traffic signs, vehicles and other objects in a sea of images. This workflow enables advanced deep learning solutions - such as tracking objects from satellite imagery, security and surveillance, advanced driver assistance systems and medical diagnostic screening.
These new parts of the NVIDIA SDK are clearly aimed at the HPC data centers and embedded hardware for self-driving cars sectors, as explained above. Both sectors are exploding and have a large potential for further growth. The HPC market is expected to grow from $28 billion in 2015 to more than $36 billion by 2020, and the self-driving car market could reach $42 billion by 2025.
In both its traditional GPU market and its emerging HPC/AI accelerator market, NVIDIA has strong products, sensible strategies and road-maps, and few serious competitors. Of course, tech giants such as Intel (NSDQ:INTC) can be expected to step in, but it's worth noting that NVIDIA stock has done consistently well in the last five years.
NVIDIA stock is up more than 40 percent in 2016, making it the best-performing stock in the S&P 500 technology sector, and Goldman Sachs initiated buy coverage on NVIDIA earlier this month, citing potential for strong growth. The impact on the company's financials and stock value is likely to become evident in a few years, and therefore smart investors should consider buying now.