Ant Group Part II
Blockchains, distributional databases and collaboration software. Yeah I bet you didn't expect that.
Welcome to all the new subscribers! I write a longform analysis about tech in China from the perspective of a 🐢 VC, sent out every Wednesday (today is not Wednesday, I know this). If you like the content, please share with friends!
It ended before it even began, a bland notice came from the Chinese regulators on November 3 halted the Ant Group IPO.
Barely a week had passed after Jack Ma’s speech at the Bund Summit. He was called into a meeting with regulators on November 2. The IPO was called off the next day. Comments from our readers speculated on the hows and the whys, but the news is - Ant is not going to float on the Shanghai STAR or the Hong Kong markets any time soon (the latest news is a delay of at least six months). And when it does, it may be a very different beast.
In this week’s vision of Ant’s future, we talk about one of Ant’s potential paths by looking at their product stack. Given Ant’s history of dogfooding, looking at their (and Alibaba’s included) pain points is a useful indicator for new offerings. Their most significant barriers to date have been gaining consumer trust, creating scalable architecture and internal coordination. We’ll review how they addressed these problems with their product offerings, and my bull and bear predictions for how each product may fare.
TLDR: Ant is pushing into SaaS with their blockchain, database and collaboration offerings.
A recap of Ant Group Part I’s takeaways:
Ant Group has 4 product lines: payment, credit, asset management and insurance
Digital payment (Alipay) made the physical wallet into a digital one, which opened up access to adjacent digital financial products
Data and meta-data from digital payments powers the selection AI algorithms for the rest of the Ant machine. Helping them to create highly targeted real-time risk profiles for consumers
Ant is their own first-and-best customer, for every product marketplace they build, they bootstrap it by using their own offering first before opening it to a broader base of other firms
In the Ant Group P&L, a line item of ‘innovation initiatives and others’ hints at the future of the company. These are the moonshots that Ant has in the works.
Blockchains Blockchains Blockchains
For Chinese consumers, trust in the products they buy remains the biggest concern. Scars from the numerous food scandals, including the infamous infant milk powder scandal in 2008 have left Chinese consumers with low trust for domestic goods.
So what does Ant do to solve this? What would any respectable financial institution do in 2020? Blockchains of course.
Simply put, a blockchain is a database. More technically, it’s a distributed database that allows every participant to share and synchronise information. There are specific rules about the way data is inputted into the database, which makes it consistent, immutable, ownable, canonical and decentralised. Data is maintained in chained records called “blocks” and not owned by any single authority. The decentralised design enables it to be transparent and tamper-resistant. Modifications by one party need to be verified by all others. We can delve into the weeds and talk about why that is good, but it doesn’t alter the basics. It’s a really safe and secure database.
Ant has been working on blockchain technology since 2015 (Jack Ma made it very public that he was bullish on blockchain technology) and currently holds the most patents for blockchain in China.
The prospectus does not mince words:
We believe that the development of blockchain technologies will revolutionise the way that the world interacts, transacts, and conducts business. - Ant Group prospectus
Ant has made substantial progress on creating usable blockchain products. Since 2018, they have launched cross-border remittance services, Blockchain-as-a-Service (BaaS) open platform, AntChain (which is a more off-shelf BaaS offering), a blockchain-backed smart contract platform called Trusple and many others.
What does this mean though? It seems like Ant is positioning the solution as a general-purpose immutable system of records and asset management service for retailers, manufacturers, governments and researchers. Here’s some practical case studies:
Glass crystal seller on Alibaba whose customers are predominantly abroad. Before blockchain, payments were paid out in parts upon receipt of wares. With Trusple, the software generates a smart contract once a buyer and a seller upload a trading order on the platform. As the retailer executes the order, the smart contract is automatically updated with crucial information, such as order placements, logistics, and tax refund options. The seller no longer has to confirm invoices, remittances or banks details. Everything is handled automatically on the blockchain.
Tea producers have digitised the tracking of their entire tea production process (from picking the tea to shipping it) through IoT sensors and blockchain platforms. It created end-to-end traceability from the farmer to the purchaser. The many steps in the tea’s preparation from picking, processing to shipping each represents a hash value and can not be altered on the blockchain. Consumers can get a full overview of the supply chain and what stage their tea is by scanning a QR code in the Alipay app. Tea on the blockchain.
I’ll be honest; if Ant’s blockchain offerings does what it says it does (vaporware is far too common in any software firm, big or small), then I’m incredibly impressed. Since 2017 I have run a mile in the other direction every time someone mentions blockchain in a pitch. With the standard blockchain set-up, you trade-off speed and ease of maintenance for security and transparency. It seems like Ant has worked out these kinks, the most significant hurdle being the ability to process high volumes of transactions securely. AntChain reportedly allows 25k Transactions Per Second (TPS), ahead of the industry average of 1k TPS and way ahead of Bitcoin’s 5 TPS. It’s also compatible with other mainstream open-source blockchain technologies like Hyperledger Fabric and Enterprise Ethereum - Quorum.
There is a genuine market need for verified consumer products; fraudulent claims wrack both Alibaba, JD and Pinduoduo regularly. A decade onwards, I still get requests to bring in foreign milk formulas for my family’s distant acquaintances. This need could be the key factor that pulls the adoption of blockchain in China quicker than anywhere else. Especially as the advent of 5G opens up a new installation base for low-cost IoT sensors.
The bull case would be that Ant and Alibaba use their enormous market power to push this through their supply chains and SME merchants. Consumer expectations get heightened as they can finally trust the products they are buying. Government regulations steps in to help as they see this being a net benefit for consumers and a way to lower grey market activities. This creates a domino effect for every other consumer-facing enterprise to catch-up or be left behind.
As I’ve mentioned in the past, the rate of enterprise digitalisation in China is low. It’s asking a lot for industries to digitalise and also standardise their processes concurrently. The organisational change barrier is the biggest hindrance for any technology adoption (note: also true in the West, but as they start from a higher base of corporate standardisation). I have a hard time seeing blockchain technology overcoming this easily.
Even if Ant lowers the degree of organisational change required with customised solutions, deployment of blockchain technology is not straightforward. There is a shortage of internal talent and external system integrators in China to enable adoption and maintenance at scale. This could be another critical bottleneck for broad adoption.
Since blockchain asks for a reimagining of not just technology but organisational structures, it’s hard to produce credible ROI cases to benchmark against. Enterprises are not swayed, their attitudes having been biased from the crypto bubbles that came and gone. SMEs are too cash strapped to invest and maintain good upkeep. The blockchain dies a slow death.
That’s all assuming the blockchain works and that the public believes it to work. As the old saying goes “Trust takes years to build, seconds to break, and forever to repair”. A single scandal involving a previously ‘secure smart asset’ could wipe out the fragile public trust. You also need the public to believe it works in the first place. Have you tried to explain what a blockchain is to your grandmother lately?
Database scalability has been a pain for Alibaba since they had the bright idea to turn singlehood into a shopping festival. The infamous Singles’ Day festival posted a jaw-dropping $74.1bn in GMV this year (to be fair, this is now over 11 days rather than a single day). That’s transactions that are more than the annual GDP of Oman. That’s like ten US Black Friday sales in one day ($7.1bn for the US in 2019). Since the most heavily discounted goods are limited, the first few minutes of the festival are crucial. This year’s event had the world’s highest traffic peak with 583,000 transactions per second. Any glitches during this period translate to massive losses and bad will for Alibaba and their merchants. Hell has no fury like a consumer who’s just lost a bargain.
So like any self-respecting engineer-heavy group, Ant built their database and middleware components. The most prominent of which is OceanBase, a distributed relational database management system. It’s easy to get lost in the offering’s product specs, but OceanBase’s key attributes are rapid scaling capabilities and high concurrent transactions levels. Its maximum processing capacity is 61 million times per second, and it holds the world record for Transaction Processing Performance (as verified by the TPC committee). As expected from a financial database, they also note their business stability, consistency and availability. OceanBase’s pitch, in a nutshell, is that they are faster and cheaper than Oracle and just as stable.
The on-premise version was launched in 2018 and counts a small handful of first and second-tier banks as their customers. The SaaS version was released in March 2020, and targets insurance and mutual funds who are not prohibited from deploying on the cloud.
I think as long as OceanBase can perform tasks on parity with Oracle, there’s massive potential for them in the Chinese market. As digitalisation of Chinese firms is still underway; the eventual Total Addressable Market will be highly significant. Given the global rise of nationalist sentiment, China will look to prioritise local tech offerings and OceanBase could be its poster child (though hard to say whether recent regulatory shuffles would have negative repercussions).
I’m more bullish on OceanBase than the blockchain offerings since there’s an existing market for this solution. Unlike blockchain, distributional databases easily fit into a RFP shortlist for essential maintenance and don’t have to wait for the corporate innovation budget.
OceanBase’s bull case is that apart from replacing Oracle, it also opens up the wider market for distributed databases through its low pricing. It becomes the dominant player for the financial database in the China market and makes a sizable dent in emerging markets too.
Not everyone is Alibaba. Actually, no one is Alibaba. As mentioned previously, big tech companies face unique problems which they create unique solutions to. While rapid scaling is an issue for Ant and Alibaba during Singles’ Day, how many others have the same pain points? Does OceanBase reflect genuine market need or just their need? OceanBase’s solution could be a mismatch with the market’s requirements, which don’t place much emphasis on scalability and concurrency.
As much OceanBase’s record-breaking glee warms my tiny technologist heart, the cold, hard truth about enterprise sales is that sales and marketing trumps product any day. The best players have created tautological justifications about going with them. You go with Salesforce because it’s Salesforce, they invented SaaS and CRM (they didn’t invent either but you get my point around storytelling). You go with SAP because it’s SAP, just like all the other fortune 500 companies, have you no ambition? OceanBase, with Alibaba’s legacy, definitely stands a chance at cracking this. However, it’s a tough road ahead when you’re dealing with skittish enterprise buyers who care more about stability, deliverability and referenceability among their peers.
The deployment bottleneck also applies to OceanBase. One interesting indicator to watch would be whether Ant ramps up investment in the system integrator system. If they do, that goes some way to solve this. Ultimately, experienced technical consultants can’t be minted in a month.
Imagine if Confluence, Notion and Google suite had a baby, that’s roughly the shape of the collaboration tool called Yuque. Ant had created this offering internally for documentation and knowledge management. It was released as a SaaS offering towards the end of 2019. It’s free to use for SME, and there’s a paid enterprise edition. Apart from Ant’s recent acquisition of Teambition to augment this offering, not much else is known about Yuque. Users online raise the issue of a lack of a mobile app version, which probably drags its adoption in mobile-first China. I get the feeling that Yuque’s main purpose is to keep DingTalk users (Alibaba’s corporate chat software) to stay in their walled garden instead of venturing out.
Unsurprisingly 2020 has been the year where collaboration apps take prominence. There’s a lot of players vying for users’ attention: Yuque, WeChat Work, Lark (Bytedance), Onenote, Yuedao, Ink, Evernote are all in the mix. Going by the competition alone I’m fearful for Yuque, but the good thing is that since DingTalk is a market leader, that could buy them a decent initial user base.
I’ve yet to try out the product for myself (and also think it deserves a longer piece). So stay tuned here.
Ant has an AI-powered risk management system called AlphaRisk (I did think OmegaRisk could have been a wittier name reference but they don’t pay me the big bucks). Their models are capable of managing risk scenarios in real-time through self-learning and model auto-refit. Impressively, they’ve achieved a payment fraud loss rate of lower than 0.0006 bps in LTM ended June 30, 2020, which was the lowest payment fraud loss rate among digital payments providers whose total payment volume exceeded US$500 billion in 2019.
While Ant’s prospectus mentions this risk module, I’m not sure whether they will ever be sold. Or if it is sold, whether it would be a standalone or part of a modular core banking software.
The future of Ant
The trajectory of Ant becoming a SaaS company is clear to see. In the prospectus, they position themselves as ’a technology services company’ and that they ‘provide the platform and tools to allow partner financial institutions to serve underserved consumers and small businesses’. They’ve followed their rule book of being their own first-and-best customer before releasing to the public. It’s going to be a matter of well-placed sales and marketing execution now to see whether these SaaS products have legs.
The big upset has been how much Ant’s valuation will be cut once it gets valued as a financial firm rather than a technology company. But these products give indications that if things go well (and it’s a big if), we might wistfully look back to the days when we were only using tech multiples and not SaaS multiples for Ant.
If you’ve gotten this far, thanks! I wanted to ask you if there’s any interest in the following topics. Vote for them below in the comments or write to me to let me know which is of interest:
Comparison of Yuque, Lark and Wechat work collaboration suite
Ant’s international expansion plans
How new regulatory measures could affect Ant
Also thanks to everyone who voted on my Twitter poll, my update for y’all is this
So next week I’ll be curating the top 5 Chinese tech news this month. Looking forward to it!
Reply to this email to let me know what you think, or find me @lillianmli where I write about what didn’t make it into the post throughout the week