In the days before President Trump's inauguration in January, Alibaba's Jack Ma was among the parade of billionaires and other dignitaries appearing at Trump Tower in New York City. At an impromptu press conference with Trump following the meeting, Jack Ma promised to create 1 million jobs in the U.S. within five years. This week, Jack Ma is back on U.S. soil to host the Alibaba Gateway 17 conference in Detroit, Michigan, where the central proposition is that Alibaba's e-commerce platform is the gateway for small and medium-sized businesses in the U.S. to access the booming Chinese economy. The Alibaba Cloud platform is a key enabler of this gateway.
Ma argues that the figure of one million new U.S. jobs is not implausible. After all, Alibaba claims to be responsible for the creation of 30 million new jobs in China already and current growth rates for the main Alibaba are impressive:
• Core commerce business units (Taobao, Alibaba.com TMALL) - 45% YoY, 62% EBITA margin.
• Alibaba Cloud - 121% YoY, -7% EBITA margin.
• Digital Media and Entertainment (Youku) - 271% YoY, -44% EBITA margin.
For this year, Alibaba expects to generate $10.0 billion in free cash flow, growing at 37% CAGR. The guidance for FY 2018 calls for revenue growth in the 45-49% range, among the best in the world for companies of its size. As previously mentioned in this series, Alibaba's gross merchandise value (GMV) is expected to hit $547 billion this year and the target is $1 trillion for FY 2020. Since Alibaba already accounts for 11% of all retail sales in China, the company is seeking much of its future growth to come from overseas. Hence, this week's Alibaba Gateway 17 conference in Detroit hopes American SMBs will jump on board, lured by the 500 million active Chinese consumers on the platform.
Given the size of its operations in China, it is no exaggeration to say that Alibaba is built on data. During its November 11th selfie shopping festival last autumn, online ordering reached a peak of 175,000 transactions per second for a total of 657 million delivery orders that data. With data centres capable of handling such a load, it is clear that Aliyun's cloud capacity is highly scalable. Its data centre architecture, while not as publicly transparent as that of Facebook, Microsoft or Google, evidently have succeeded in overcoming the technical challenges involved in moving so much data so quickly across data centres and national network.
Among the technical skills required to operate at such scale is the search engine and personalisation algorithms to ensure that each user can find the product, services and entertainment that best matches their unique profile. Alibaba says it has 507 million mobile active users and over 1 billion active listings in its database of products and content. As AWS and Google, Alibaba is moving swiftly to implement speech recognition as a new customer input for generating system requests. Ultimately this requires a knowledge graph matched to an evolving user profile. Millions of customer inputs or even customer support inquiries must be processed for the 100,000+ merchants on the Alibaba platform already.
As noted previously, Aliyun's major domestic data centres are in Beijing, Hangzhou, Qingdao, Zhangbei and Shenzhen. Other active data centres are located in Hong Kong, Singapore, Tokyo and Silicon Valley. The company has announced plans to expand this fleet to 17 data centres, including many in neighbouring Asian countries.
Intel, ARM, Nvidia and Barefoot
One of Alibaba's high-profile partnerships to solve the scalability challenge is with Intel. Earlier this year, Aliyun kicked off a pilot program with Intel for a cloud-based FPGA acceleration service. Specifically, Aliyun will use Intel Arria 10 FPGAs, Intel Xeon processor-based servers and software development tools for application acceleration as a ready-to-go pre-configured infrastructure. The Aliyun service offers systems designers cloud-based workload acceleration as an alternative to investing in on-premises FPGA infrastructure. It might not be all Intel inside for future Aliyun data centres. In April 2016, a China-based Green Computing Consortium (GCC) was formed with the goal of establishing a deep ecosystem in China for big data, enterprise and cloud computing platforms based on the ARM architecture. Under this consortium, ARM is working with Alibaba, Baidu, China National Software and Service, Dell. Guizhou Huaxintong, the joint venture company of Guizhou and Qualcomm Hewlett Packard Enterprise/H3C (HPE), Lenovo and Phytium. Alibaba has also been working with Nvidia to develop its AI processing performance. In 2016, Aliyun stated that it would use Nvidia Tesla K40 GPUs to power a range of HPC, AI and deep learning applications.
Meanwhile, earlier this month Barefoot Networks, which is developing high-speed switching silicon, announced that Aliyun had adopted its 6.5 Tbit/s Tofino switch, claimed to be the fastest and P4-programmable switch chip. Although the size of this order was not disclosed, the deal suggests that Aliyun might be developing its own networking gear, much like Google and Facebook, rather than buying brand-name equipment from the traditional vendors.
Clean power for Aliyun?
In its mid-2017 investor presentations, Alibaba noted green data centres as an area of focus for Aliyun infrastructure. However, the Greenpeace Clicking Clean report from January 2017 criticised Alibaba for not committing to a timeline for 100% renewable energy for its data centre operations and for a lack of basic transparency about the energy efficiency of its existing facilities.
Will Alibaba and Aliyun catch on in the U.S.?
For Chinese multi-nationals already in the U.S. market, the opening of local Aliyun data centres is good news. A seamless service on one hyperscale public cloud is easier to manage than having to host some in Aliyun China and some in AWS U.S. There is a good chance that Aliyun U.S. can capture many of the Chinese companies abroad. But are those customers enough to challenge AWS or Microsoft? Probably not.
Jack Ma's vision is decisively in favour of the little guy, the SMBs. His outreach to the U.S. gives entrepreneurs and independent farmers easy access to Chinese buyers. One criticism of Alibaba's global expansion plan, as noted in a recent Bloomberg perspective is that the company lacks the distribution logistics and warehouses in China that Amazon possesses in the U.S. Moving goods from the US to the Chinese market will never be as easy as domestic operations. There are other bi-lateral trade issues between the nations that stand in the way, such as tariffs and quotas on many types of products.
Then there is the question of capex investment needed to keep pace with Amazon, Microsoft and Google in the U.S. cloud market. Some of these players are investing multiple billions per quarter in the cloud infrastructure. While Alibaba did not provide capex guidance during its Investor presentations, the figures do not appear to be close to the levels of the U.S. players. Each of the U.S. public cloud giants has also made significant investments in network infrastructure. For instance, Facebook and Google are teaming up to build the highest capacity, trans-Pacific cable system to date. The Pacific Light Cable Network (PLCN), which will stretch 12,800 km between Los Angeles and Hong Kong, will have an estimated cable capacity of 120 Tbit/s when it enters service in the summer of 2018. Both players are betting that their platforms will be super-heavy of trans-Pacific bandwidth. This level of commitment from Alibaba has not been evident, at least not yet.
