Ant Financial intends to launch its proprietary blockchain technology as a “blockchain-as-a-service” this summer, says Hu Danqing, head of the company’s blockchain product development.
The company’s first customer will be Kweichow Moutai, the high-end spirits manufacturer, but it has also been tested domestically with an insurance company and will be applied to financial services.
Hu didn’t elucidate financial use cases but told DigFin that trade and supply-chain finance are “interesting” because they involve the need to instill trust among a large number of entities.
“Trust can not go beyond two or three hops,” he said of the status quo in trade finance, referring to degrees of separation among counterparties. “Can we use blockchain to get more institutions to collaborate?”
Ant Financial’s strategy outside of China is to work with companies in parent Alibaba’s ecosystem, letting them bring it potential use cases for Ant technology. Other than Hong Kong, where Alipay is revving up a business, Ant is not attempting to compete directly. Rather it will use its technology to support local partners in which it has a stake, such as Lazada in Indonesia or Paytm in India – leaving regulatory and licensing issues to them.
Ant has developed its blockchain to fit somewhere between R3’s Corda and Hyperledger’s Fabric, while trying to outperform them all. Hu says Corda is just for financial institutions, while Fabric tries to be all things to all people. Ant’s blockchain is targeted at a middle level of applicability, with its partners in mind. “It’s a driving force for technology challenges that haven’t been solved,” Hu said.
Its closest equivalent is Ethereum, with a similar smart contract capability, but Ant’s blockchain is permissioned, not public.
And, more dramatically, Ant claims it has made much greater progress on performance. Hu says the company’s blockchain now processes 25,000 transactions per second, versus 1,000 TPS for Fabric, the IBM-backed project under the Hyperledger governance model. Ethereum is still around a mere 20 TPS.
Ethereum developers are focused on this problem and are using techniques such as sharding to boost performance. (Sharding is a way of splitting the processing of transactions among nodes, rather than having each node be responsible for validating each trade.)
But Hu says that although Ant has incorporated some of these techniques, its secret sauce lies in its algorithms. “Our approach is to maximize the utilization of I/O [inputs and outputs, i.e., the throughput of the network] and computing resources.”
Consensus, the process by which nodes validate blocks of information, is a process of rounds of computations. Ant’s engineers have found computational rules (algos) that, when unleashed on Aliyun, the group’s proprietary cloud computer, reach consensus much faster than others.
Work in progress
Hu says several challenges remain when it comes to blockchain technology. Performance is good but has a ways to go.
The second challenge is privacy. The nature of pure blockchain is that nodes validate information, which means everyone on the network sees the underlying data. This is not good for encouraging trust. Ant’s approach is not to segregate which nodes can validate a given transaction, but to have a broader, more universal notion of privacy. Hu says Ant prefers to operate on the basis of use-case by use-case, and add privacy functions such as zero-knowledge proofs if that’s appropriate to what users want. (A zero-knowledge proof is a protocol allowing one party to confirm she can validate a block of data without actually seeing the underlying, sensitive information.)
The third, and biggest, challenge is what Hu calls “anchoring”. This is using sensors, chips and other devices to map any physical item, or asset, that is not native to the blockchain (which means almost everything).
For example, if Ant wants to use its blockchain to transact exports of milk from Australia to China, it needs to confirm the origin of that milk, so that Chinese consumers trust it’s the real McCoy. Putting a chip on a cow in Australia is the starting process, but the hard part comes in tracking the milk once it leaves the udder. The same principle applies to purely financial transactions.
The simplest solution would be encrypt/decrypt protocols, but regulators don’t always allow this. China’s government may not allow the storage of encrypted information regarding Chinese citizens outside the country; the European Union also has strict rules such as the right to be forgotten, which complicate the immutable nature of distributed ledgers.
Hu says he’s looking at outside examples to help figure out solutions. One is Singapore’s Project Ubin, a competition among Corda, Fabric and Ethereum Enterprise Alliance to develop a payments method via blockchain. “They are solving the anchor problem by incorporating MAS as the anchor,” Hu said, referencing inclusion of MAS on the ledgers as an all-seeing supervisor. But solutions are a work in progress: “There is no universal scheme for privacy on a blockchain.”