The gloves are off: Alibaba strikes back at Google Cloud and AWS with its own custom AI accelerator

The global off-premises cloud services market is set to more than double between 2018 and 2023, as enterprise adoption grows to improve application performance and increase scalability.

Global off-premises cloud services revenue will expand to $407 billion in 2023, up from $188 billion in 2018, according to the latest Cloud & Colocation Services for IT Infrastructure and Applications Market Tracker from IHS Markit | Technology, now a part of Informa Tech.

Within the off-premises cloud services market, the platform-as-a-service (PaaS) segment is where artificial intelligence (AI) and machine learning (ML) techniques are most heavily used. AI and ML equip developers with pre-built deep-learning application components that allow them to focus on creating the business-specific portions of their apps and reduce their time to market.

In the first half of 2019, PaaS generated the second-highest year-over-year (YoY) growth rate of any segment of the off-premises cloud services market, with a 41 percent increase. It also yielded the second-highest compound annual growth rate (CAGR) from 2018 through 2023, at 23 percent.

Cloud-as-a-service (CaaS) had the highest YoY growth rate in the first half of 2019, at 48 percent, and the highest 2018-2023 CAGR at 25 percent.

“Machine learning and artificial intelligence innovations are allowing cloud service providers (CSPs) to differentiate themselves by offering custom hardware or by fielding teams of highly skilled experts like data scientists that most enterprises cannot duplicate,” said Devan Adams, principal analyst at IHS Markit | Technology. Because of the rising adoption of these new techniques, PaaS usage continues to grow. This finding aligns with results from the IHS Markit | Technology Cloud Service Strategies & Leadership North American Enterprise Survey - 2018 , where respondents expect to increase the percent of applications they will develop using PaaS by 2020 to 39 percent on-average, up from 28 percent in 2018.

AI and ML remain in the cloud services market spotlight

CSPs are capitalizing on the rising demand for deep-learning services and are responding by introducing numerous AI and ML solutions, including:

  • Alibaba launched its Hanguang 800 AI inference chip, allowing it to compete with the custom accelerator offerings from two of its top rivals, AWS and Google Cloud. As previously announced, AWS’s Inferentia co-processors are designed for compute-intensive ML applications and will be offered in its Elastic Inference instances. Google Cloud’s tensor processing units (TPUs) are intended for creating ML models and can be selected individually or in clusters (TPU Pods).
  • Alibaba announced its X-Dragon Architecture, which is designed to improve the performance of AI and other highly parallel compute tasks. The company says the new architecture will increase queries-per-second (QPS) by 30 percent, decrease latency by 60 percent and reduce unit computing costs by 50 percent.
  • AWS announced Amazon Forecast, a fully managed service that uses ML to create more accurate forecasts; it’s currently available in select Availability Zones.
  • Microsoft partnered with OpenAI, an AI research company focused on discovering and ensuring safe artificial general intelligence, to increase innovations in AI and further OpenAI’s mission to create autonomous systems for general occupational tasks.
  • Google Cloud introduced several enhancements to its Vision AI portfolio, including:

  • AutoML Vision Edge added object detection and image classification on edge devices.

  • AutoML Video Intelligence added object detection to track the movement of multiple objects between frames.
  • The Video Intelligence API has been upgraded with Video Intelligence Logo Recognition, which detects, tracks and recognizes more than 100,000 logos of popular businesses and organizations.
This content extract was originally sourced from an external website (IHS Technology Press Releases) and is the copyright of the external website owner. TelecomTV is not responsible for the content of external websites. Legal Notices

Email Newsletters

Sign up to receive TelecomTV's top news and videos, plus exclusive subscriber-only content direct to your inbox.