SMIC (Semiconductor Manufacturing International Corporation) became another Chinese company restricted by the US. In the global supply chain, collateral damages caused by these restrictions may go far beyond what people can think of.
A quick look at these restrictions would lead to limited production capacity and detriment to the company's business. Furthermore, SMIC's role as a part of the supply chain may involve more companies, who source chips from SMIC, in supply shortages.
If we delve into the situation, it is really hard to evaluate or even substantiate the national security threats claimed by the US government who then took unjustifiable actions against China.
Most smartphones today can efficiently recognize faces and fingerprints. These biological data are so regularly and securely used simply because of an area inside the central processor called Trusted Execution Environment (TEE).
TEE is particularly built for processing sensitive data separately. This process results in an simple output of yes or no so that the risk of data leakage is eliminated. For instance, when a smartphone user unlocks a device or makes in-app payments, TEE will verify his or her biological data. The requestor APP will only receive the message from TEE of whether or not the identity belongs to the real owner, with no access to the raw data.
Inherently interrelated with chips, TEE’s technical standards and market share align with that of chip manufacturers. Put simply, only chip giants talk.
In the computer and server realm, TEE solutions based on X86 architecture are represented by Intel's SGX and AMD's SEV. On mobile devices, ARM’s Trustzone is used by chip manufacturers like Apple, SAMSUNG, and Qualcomm.
A cruel reality faced by Chinese companies is that most of these giants are American companies. They have to comply with the government’s actions toward China, no matter it’s export restriction or patent ban. And their clients, namely, Chinese companies, will be left in a “to be or not to be” question.
Be that as it may, the damage is still a scratch on the skin. The agony comes from an invisible cut that penetrates the deep. Technically, chip manufacturers can leave a backdoor to extract data from TEE. In a world where chips are ubiquitous, the synergy between sanction and backdoor may become an imminent threat not only to Chinese companies, but also to China’s national security.
In 2017, Alibaba Cloud and Intel jointly released the encryption computing technology based on Intel SGX, the data on the cloud are protected by Intel chips with intrinsic TEE capacity.
Unfortunately, the foreseeable risks are also "intrinsic". Risks like supply cuts or malicious actions will be lethal to Alibaba Cloud or even the entire Chinese IT industry, as one of Alibaba’s senior executives mentioned: “80% of China’s tech companies are using Alibaba Cloud.”
Hopefully Alibaba is powerful enough to fix the problem. But not everyone is that rich and resourceful, particularly those startup companies specializing in TEE solutions. An anonymous insider revealed that recent restrictions on chip manufacturers have put them into a difficult situation. Their TEE solutions are just too much reliant on major chip manufacturers like Intel. Their clients’ subsequent worries will eventually become their own – a gloomy perspective for future business.
The collateral damage is indeed substantial.
Robin Li, CEO and co-founder of Baidu, mentioned in an earlier interview that in the past two years the world has generated as much information as in the entire human history.
Just like how the world works, big data create value only when they flow. But the circulation of sensitive data has long been a labyrinth. That’s the reason why confidential computing emerges as a technological solution, apart from policies or legislations, to harness the value of data by enabling flow, sharing, and computing of data from different sources.
TEE shares part of the mission by processing data in an isolated environment. Instead of solving the problem by hardware, there are some other ways using software algorithms only to achieve the same goal, such as federated learning and multi-party computation (MPC).
But there is a shared weakness of the encryption algorithms – they are usually task-oriented and unable to scale when it comes to generic use cases. Federated learning, for example, is designed and built exclusively for data sharing in machine learning.
Originated from cryptography, the computational logic of MPC is usually tailored for a very specific scenario – when a use case evolves, the algorithm will take extra time and effort to catch up. MPC is technically premature to consume an excessive amount of computational resources for low efficiency.
The weaknesses of TEE are also apparent: reliance on chip manufacturers, vulnerability to hardware attack, and malicious backdoor. But the advantage of TEE over other algorithm-based solutions is its generic application – TEE supports 90% of all common use cases, according to industry experts.
TEE is just too versatile to be replaced by another single technology. But changes and updates can still be made for TEE itself to be immune from commercial and political restrictions.
Nevertheless, the fast-growing blockchain technology would be a promising option.
It’s been decades since the idea of a distributed network. Decades ago, CPUs around the world were linked together to produce massive computational power through distributed networks. 12 years ago, one of the distributed networks became a distributed ledger, or the very first blockchain network called Bitcoin.
The way how CPUs were connected can also be applied to TEE – anyone can register his or her TEE capacity in the CPU onto a blockchain network, and the network will assign tasks to different CPUs.
The strength of blockchain network is innate. It is technically unlikely to ban a decentralized network, just like the Bitcoin network from its inception 12 years ago to now. Similarly, TEE run by individual chip manufacturers can be reclaimed or manipulated, but a worldwide cluster of TEEs on a decentralized network can never be controlled or shut down.
Token is another benefit to incentivize users. Tokens enable TEE contributors to be incentivized from their computing powers while requiring users to pay for the service, TEE resources over the entire network can be managed and utilized in a highly-efficient way.
Note that only a public chain (fully decentralized blockchain network) can work on what we are seeking. A public chain is not solely controlled by any entity and will hardly be shut down. Yet, a private chain or permission chain run by cooperated third parties is not able to achieve such independence. might not be independent enough to support the long-run maintenance of such a decentralized TEE community.
TEE and blockchain are complementary on functions to each other because blockchain endows TEE with enhanced security, accessibility, and scalability, while TEE offers the privacy protection that has long been missing and anticipated on blockchain, as on-chain data are usually open for public scrutiny which causes problems for private and confidential information.
The business world values innovations. Many startups have been working on the synergy between TEE and blockchain, among which there are rising stars such as Oasis Labs founded by Dr. Dawn Song from UC Berkley, and Phala Network project led by a group of local elite scientists.
Phala Network is in nature an on-chain protocol that serves as a unified privacy data trading infrastructure based on blockchain and TEE technology. With confidential smart contracts, it makes data collection, management, authorization, and analysis possible in a trustless way.
To eliminate any possible backdoor, Phala uses a “random miner” mechanism to randomly designate tasks to TEEs. Therefore adversaries can’t target the specific TEE chips to create backdoors.
Phala is built as a parallel chain to Polkadot, a world-class blockchain project started by blockchain pioneers around the world and renowned for interoperability so that smart contracts on different blockchains can interact with each other. But what matters most here is that Polkadot is a genuine public chain able to bypass bans.
In a world heading towards Web 3.0, big data, AI, and blockchain technologies will revolutionize the way we communicate and interact with machines, people, and the world. It’s all happening on data flow secured by the latest innovative solutions, among which on-chain TEE is expected to be the most efficient, cost-effective, and enduring.