Over the past month, Huawei has announced the details of the company’s portfolio of AI products and cloud services. The new “Huawei AI Cloud” (see Figure 1) seeks to remove the two main obstacles impeding the growth of AI development: (1) the significant time and effort to develop AI applications (such as data labeling, model training, deployment etc.), and (2) the increasingly expensive cost of limited computing resources. As articulated by Huawei’s rotating chairman, Eric Xu, at the recent Huawei Connect 2018 conference, the company’s vision is to globally make AI “more inclusive – making it affordable, effective and reliable for all.”

Figure 1. Architecture of Huawei AI Cloud

nTheta expects that Huawei’s “AI Cloud” will provide significant opportunities for AI developers – both domestic and foreign – to access the AI market in China.

“Full Stacks” and “All Scenarios”

Huawei describes its new AI offering as a “full-stack” portfolio of AI products and cloud services for “all scenarios.” As Xu explained: “’full stack’ means Huawei is able to provide AI application developers with unparalleled computing power and a strong application development platform.” Huawei’s full-stack AI portfolio includes its two new AI chipsets (the Ascend 910 and Ascend 310), chip enablement, a training and inference framework (Tensorflow, MindSpore), and application enablement (“ModelArts”). This means that developers can use Huawei’s AI portfolio to design a complete AI product including appropriate hardware and fully functional and verified software without requiring any additional resources.

By “all scenarios”, Huawei means various deployment scenarios for AI, including public clouds, private clouds, edge computing in all forms, industrial IoT devices, and consumer devices. Essentially, Huawei aims to deliver pervasive intelligence for a fully connected, intelligent and open AI ecosystem.

Abundant Computing Platforms

Figure 2. Abundant Hardware Platform Selection

Powerful computing resources are expensive, and it is both time-consuming and difficult to find a suitable hardware platform to match specific AI applications. The Huawei AI Cloud addresses these issues by allowing AI developers to conveniently sample, compare and ultimately choose the appropriate hardware platform to meet their computing requirements.

As Huawei explains it, the diversity of computing platforms not only saves time on AI hardware design, but also gives developers a “consistent development experience.” This allows AI companies to transition more smoothly from light applications to large-scale AI deployments. 

Benefits to AI Developers

With Huawei’s powerful AI Cloud, AI development will become easier, cheaper and faster. The value of Huawei AI Cloud for the China AI market is clear. There are 4,040 AI companies in China, but only 30% of them have received VC investment[1]. The reason for this investment deficit is the prolonged and expensive R&D process inherent in AI design. Without funding, AI start-up companies simply cannot afford the expensive computing resources necessary to develop AI. By making computing resources and development tools readily available and affordable, Huawei’s AI Cloud should provide a significant boost to AI startup efforts.

But the benefits of Huawei AI Cloud are not limited to Chinese developers. In fact, AI companies all over the world can make use of Huawei AI Cloud to accelerate their own AI development and penetrate the China market. While certain AI sectors will likely remain inaccessible to foreign companies (e.g., security-related applications), there should be numerous, non-controversial opportunities for foreign companies to participate in the China AI industry. In particular, nTheta expects that the following areas could be fruitful for foreign companies interested in leveraging the Huawei Cloud to launch their AI business in China:

  • Consumer Apps
    Huawei is in a strong position to establish its own AI ecosystem, but its success largely depends on attracting successful consumer apps.  Thus, AI apps, especially those with a proven user case, have an obvious opportunity to leverage Huawei Cloud to access the China consumer market.  Moreover, since Huawei’s smart phones must compete in the market based on user experience, any AI app which is optimized to run smoothly on Huawei’s phones is likely to receive Huawei’s marketing support or possibly be acquired for exclusive use on Huawei phones.
  • Large-Scale AI Deployments (SAAS)
    Large-scale AI deployments for enterprise use (e.g., SAAS) present another area of strategic opportunity. Companies providing AI solutions on the edge or in the cloud may be able to cooperate with Huawei or its internal chip division, Hisilicon, to bring their technology to the China market. Again, if the solution is optimized to perform on Huawei’s AI chips or other hardware, it may be possible to partner with Huawei to promote the technology in the China market and bring the solutions to Huawei’s core value partner network. Sell to, sell through or sell with models of cooperation are all possible options.
  • AI Companies in Strategic Geographic Markets
    Huawei’s global strategy is particularly focused on certain strategic markets, especially Europe and India.  AI companies that can help Huawei promote its brand and competitiveness in these geographic markets have an opportunity to receive reciprocal help from Huawei in the China market.
  • Solutions Based On Huawei AI Development Modules
    Developers who offer solutions based wholly or partly on Huawei’s AI development modules (e.g., its facial recognition module) also have a strategic opportunity to cooperate with Huawei. If these applications attract a substantial user base, Huawei can be expected to promote the app as a success story for its AI platform. It is also possible that Huawei may wish to buy the technology.
  • Safety Certification Service
    In recent years, Huawei’s global ambitions have been hindered by concerns raised by oversees media and governments about the company’s transparency and the security of its technology. This situation presents an opportunity for any service that can independently test and verify the security of Huawei’s AI portfolio of hardware and services.
  • Algorithms for New Functions
    Any algorithm that can support new functions on Huawei smart phones, especially AI related functions, should be of interest to Huawei.  The company sometimes prefers to purchase useful algorithms rather than spend the time and resources to develop the functions independently.

To summarize, AI companies with technology that utilizes Huawei Cloud or is optimized for Huawei smart phones, chips or other hardware, have significant new opportunities to access the China market. In addition to the ability to leverage the resources of Huawei’s “full-stack” AI portfolio, companies with unique AI technology may be able to cooperate with Huawei to promote their products and services in China or to sell to Huawei for integration into Huawei’s AI platform.

Leave a Reply

Your email address will not be published. Required fields are marked *