State of Chinese Cloud - Part II

In case you missed it, I co-wrote the bear and bull case for Agora last week with Packy M and appeared on a new episode of European Straits this week.

Premium subscribers will be getting a deep dive into the Chinese Cloud this month with write-ups on the market positioning of different players. They’ll also be able to access a new set of product walkthroughs on the Chinese Characteristics Circle Community.

This is the second instalment of the state of the Chinese Cloud. Here’s Part I

I committed a personal sin in my last article: I was imprecise. As I thought more about the state of the Chinese Cloud since Part I, I realised that I used the words 'digitalisation' and 'cloud adoption' almost interchangeably, when in fact that they are very different. The difference explains why Chinese cloud adoption is slower relative to the US.

I want to break down a firm’s digital transformation journey into three phases: digitalisation, cloud adoption and automation1.

Digitalisation - the process of digitisation (which is the process of converting information into a digital format) to improve business processes. This can be done either in offline or online databases.

Cloud adoption / cloud migration - the process of moving digital assets — like data, workloads, IT resources, or applications — to cloud infrastructure. We often overlook private cloud in this discussion, but cloud adoption does not always need to be online.

Automation - introduction of technology processes that reduce human intervention in the day-to-day workings of a firm. All software that touches and improves workflow is part of this.

The progression of digital transformation in the West happened over many decades, starting with the ancient mainframe for digitalisation in the 1960s. This meant data and processes for most firms (especially the service sector, which accounts for ~77% of the US economy) were in a digital form by the early 1990s, through residing offline. When Salesforce and Software-as-a-Service arrived in the early 2000s, the migration process was not exactly painless but somewhat straightforward2. The data were already in a semi-structured digital form and could be ported over in a matter of months. The advent of automation has been happening slowly in the background3 but very much ramped up in the 2010s due to the prominence of AWS and the arrival of sexy AI branding. Automation also benefited from the vast quantities of structured and digitalised data residing in the cloud which enabled relevant automated decision matrixes.

The process I've described is not clear cut. In reality, there are significant variances between industries for their Digitalisation-Cloud-Automation journey (DCA journey going forward), and it’s not always in that order either. But we’re talking grand theory here; let’s not focus too much on the corner cases.

Each stage of the DCA journey enabled subsequent stages to occur and happened at a sequential pace for the West; the US is further along relative to Europe. The journey is sequential as each step unlocks new business models and technology needs that would be unfeasible or unwanted at an earlier stage. In a previous post, I postulated the F theory of startups, a supply-side theory of tech startups formation. The DCA journey is the demand-side of the equation.4

To give a few examples, Datadog (a monitoring service for cloud-scale applications) will not exist without a proliferation of cloud software in the enterprise. The recently IPO'd UiPath (a robotic process automation platform) will not be possible without existing digital processes to automate.

To iterate, each successive stage unlocks new potential as each new “layer” of tech is dependent on the foundations laid by the previous layer to succeed. Without a sound set of accessible, structured data through digitalisation, cloud adoption is difficult to achieve. Without a unified data source in the cloud, a single view of the customer through in-depth analysis is almost impossible.

At this point, you might say - this is all well and good, but so what?

To which I say, this development sequence is the assumed path for cloud adoption and growth of SaaS, but not so for China. China is undergoing all three stages of digitalisation, cloud adoption and automation concurrently rather than sequentially.

Since Chinese companies start with low levels of digitalisation, cloud adoption often has to start at the foundations. Hence, the pace of adoption for the Chinese cloud seems slow from the outside; they are trying to leapfrog several adoption cycles at once.

The fastest cloud adopting industries are in consumer internet, retail, healthcare, government and finance, typically the sectors facing the most customers. This is not a coincidence. In China, business' adoption of new technologies is often driven by consumer demands rather than their internal business needs. When you're used to super apps that bring you everything you desire with a few clicks, having to fill out paper forms to open a bank account is a rude awakening.

As indicated by BCG and Alibaba whitepaper (Chinese only, unfortunately), the areas that have digitised and adopted the cloud the fastest tend to industries with relatively standardised workflow and decentralised suppliers.

Relative to the US, where most firms have digitalised, and cloud adoption is the logical next step. In China, cloud adoption is predicated on whether the industry has digitised first and is highly heterogeneous across sectors.

A few predictions from my sequential versus concurrent DCA journey framework:

Let's unpack the reasoning behind each prediction.

Following Part I, we know Chinese firms have security concerns around the public cloud and want customised workflows. Driven by a scepticism that cloud adoption can generate real ROI for firms. Since Part I, I did more primary research and found that firms are still moving to the cloud despite the resistance. They aren't adopting cloud purely to adopt cloud, but for the promised AI workflow glory that cloud enables. Cloud adoption is seen as a necessary evil.

To assuage the security concerns, private clouds are preferred, especially by government and finance organisations5 who are larger buyers of cloud services. This has led to the rise in connectivity middleware and hybrid clouds since public cloud is still cheaper to maintain6. Relative to the standardised US cloud tech stacks, Chinese tech stacks in the immediate future will be varied. They will potentially benefit from being a late adopter by including newer technology such as edge computing and distributing cloud in their stacks.

Firms still have to see a clear ROI for digitalisation and cloud adoption, which only happens when the solution targets an organisation's specific pain points. In the context of China, with its large variance of workflow both within and between industries, solving an organisation's problem typically entails specialised domain offerings. This can either be through custom-built or vertical-specific solutions and preferably with built-in automation. The wholesale adoption of digitalisation to automated cloud solutions lends itself well to monolithic solutions that cater to various enterprise needs. Lastly, since China's true cloud potential lies in bringing on its vast swaths of factories and manufacturing capabilities online, digitising an entire supply chain and its many links are often the end goal in deployments. This factor heavily incentivise more verticalised solutions - so it looks to be a race between whether cloud platforms can verticalise before verticalised cloud startup can reach scale.

We can see these trends manifesting in the strategies of cloud players. IaaS remains the strongest growth area in all of the cloud segments with the key players being Alibaba, Tencent, China Mobile and Huawei. Though market share estimates differ between analysts, the consensus is that AliCloud is the current leader. AliCloud shows a significant proportion of revenue from value-added services, often custom projects deployed on top of their cloud platform. Every player shows a high degree of industry specialisation in their marketing materials in technology, finance, retail, manufacturing and more.

Source: Alibaba Investor Day materials 2020

Source: Alibaba Investor Day materials 2020

Given the cloud green space and need for domain-specific expertise, we're also seeing the rise of a plethora of vertical-specific cloud platforms often stemming from companies with expertise in that domain. These aren't always traditional consumer technology companies, Haier, Midea, and other white goods manufacturers have introduced their own industrial cloud platforms. Youzan (paywalled) has their e-commerce cloud; the list goes on.

It's hard to predict what will be the future of Chinese cloud. China has a different set of starting conditions and industry mix than the US, which can lead to a divergence on the general structure of cloud adoption and tech stack configuration. But on the other hand, the deployment base for cloud is still PC and data centres rather than mobile, and the trends happening in the cloud players right now (price wars, low-code and getting to economies of scale as quickly as possible) is aligned with US cloud players’ trajectories. Maybe all it takes is a few replacement cycles for Chinese firms to also arrive at the modularised stack that US companies are accustomed to, maybe not. Either way, you'll find out when I do.

The last instalments of the state of Chinese cloud will be focused on the relative positioning of the large cloud players and will be available only to premium subscribers.