Data Center Accelerator Market Growing at the Fastest Rate in APAC Region

Posted by Steve Stark on September 15th, 2021

As multinational and domestic enterprises increasingly transition to cloud services providers (CSPs) and colocation solutions, the data center market in China continues to evolve. The demand for data centers in the country has now exceeded the available supply as organizations seek enhanced connectivity and scalable solutions for their growing businesses. Investments by the Chinese government for stimulating technological developments have led to an increase in the adoption of cloud-based services, such as Big Data Analytics and Internet of Things (IoT). Various government reforms, such as the establishment of free trade in Shanghai, are attracting international investors. The growing demand for high-density, redundant facilities is triggering a shift in the design and development of the country’s data centers.

The global data center accelerator market size is projected to grow from USD 13.7 billion in 2021 to USD 65.3 billion by 2026; it is expected to grow at a CAGR of 36.7% from 2021 to 2026. Factors such as growing demand for deep learning and surge in demand for cloud-based services are driving the growth of the market during the forecast period.

Driver: Growth of cloud-based services

Deep learning services being made available over the cloud are reducing the initial costs associated with executing business operations and curtailing server maintenance tasks. A growing number of tech giants and startups have begun offering machine learning as a cloud service due to the burgeoning demand for AI-based computation. Most companies and startups do not develop their own specialized hardware or software to apply deep learning to their specific business needs. Cloud-based solutions are ideal for small and midsized businesses that find on-premises solutions costlier. Thus, the increasing adoption of cloud-based technology is necessitating the need for deep learning.

Big data analytics has also played a pivotal role in the growth of cloud services. Big data analytics is the process of scrutinizing large datasets to uncover hidden patterns, unknown correlations, market trends, customer preferences, and other actionable insights. Big data has become important to many public and private organizations wherein massive amounts of domain-specific information is generated, which can contain useful information on national intelligence, cybersecurity, fraud detection, marketing, and medical informatics. The deep learning technique is used to extract high-level, complex abstractions from data through a hierarchical learning process. It is an important technique used for analyzing massive amounts of unsupervised data, making it a valuable tool for big data analytics wherein the raw data is largely unstructured. Deep learning is also used for extracting complex patterns from massive volumes of data, semantic indexing, data tagging, fast information retrieval, and simplifying discriminative tasks.

Download PDF Brochure:https://www.marketsandmarkets.com/pdfdownloadNew.asp?id=48984803

The evolution of technologies, namely machine learning and artificial intelligence (AI), has generated the demand for cognitive computing technology across various verticals such as automotive, industrial, and consumer. Rapid developments in the video analytics domain and increasing adoption of advanced technologies in the security and surveillance industry have resulted in the development of high-performance AI-capable processors such as GPU and TPU, which have higher memory bandwidth and computational capability as compared to traditional processors, i.e., central processing units (CPUs). Creative professionals, gamers, designers, and video enthusiasts require deep learning accelerators with parallel processing capabilities that can facilitate the provisioning of on-demand machine learning for augmented reality, virtual reality, and several other application areas.

Restraint: Limited AI hardware expertss

AI is a complex system, and for developing, managing, and implementing AI systems, companies require personnel with certain skill sets. For instance, people dealing with AI systems should be aware of technologies such as cognitive computing, ML and machine intelligence, deep learning, and image recognition. In addition, integrating AI solutions with existing systems is a difficult task that requires well-funded in-house R&D and patent filling. Even minor errors can translate into system failure or malfunctioning of a solution, which can drastically affect the outcome and desired result.

Professional services of data scientists and developers are needed to customize existing ML-enabled AI processors. AI is a technology that is still growing and emerging, and hence workforce possessing in-depth knowledge of this technology is limited. The impact of this restraining factor will likely remain high during the initial years of the forecast period.

Opportunity: Demand in the market for FPGA-based accelerators

An FPGA is an integrated circuit that can be configured by a customer or designer after it is manufactured (field programmable). FPGAs are programmed using hardware description languages such as VHSIC hardware description language (VHDL) or Verilog. FPGAs offer advantages such as rapid prototyping, short time to market, ability to be reprogramed in the field for debugging, and long product life cycle. They contain individual programmable logic blocks known as configurable logic blocks (CLBs). These logic blocks are interconnected in such a manner that a user can configure the computing system multiple times. FPGAs contain large resources of logic gates and RAM for complex digital computation.

In 2017, Intel (US) acquired field-programmable gate array (FPGA) chip designer Altera (US). With this, Intel is expected to further leverage FPGA accelerators into its primary data center server business. In May 2020, Aldec, Inc., a pioneer in mixed HDL language simulation and hardware-assisted verification for FPGA and ASIC designs, has launched a new FPGA accelerator board for high-performance computing (HPC), high-frequency trading (HFT) applications, and high-speed FPGA prototyping. The HES-XCKU11P-DDR4 is a 1U form factor board featuring a Xilinx Kintex® UltraScale+™ FPGA, a PCIe inference, and two QSFP-DD connectors (providing a total of up to 400 Gbit/s bandwidth), and which hits the ideal sweet spot between speed, logic cells, low power draw, and price.

Challenge: Unreliability of AI algorithms

AI is implemented through machine learning using a computer to run specific software that can be trained. Machine learning can help systems process data with the help of algorithms and identify certain features from that dataset. However, a concern associated with such systems is that it is unclear as to what is going on inside algorithms; the internal workings remain inaccessible, and unlike humans, the answers provided by these systems are uncontextualized. Researchers at the Facebook AI Research (FAIR) lab found that the chat bots they created had deviated from their predefined script and were communicating in a language created by themselves, which humans could not understand. While one of the important goals of current research is to improve AI-to-human communication, the possibility that an AI system can create its own unique language that humans cannot understand could be a setback. Moreover, several scientists and tech influencers, such as Stephen Hawking, Elon Musk, Bill Gates, and Steve Wozniak, have already warned that future AI technology could lead to unintended consequences.