How U.S. Cloud Service Restrictions Could Hurt Chinese Companies

The Curious Case of Cloud Computing Restrictions

In early July this year, rumors circulated amongst the American media that the government sought to limit Chinese companies from harnessing U.S. cloud services. Furthermore, U.S. companies would require government approval before providing cloud services to Chinese clients.

This proposed constraint stems from the administration’s apprehension that Chinese corporations, especially those specializing in artificial intelligence, could obtain the formidable computing power of cutting-edge chips by utilizing the cloud services of American tech giants. Thus, they could legally skirt existing prohibitions and train various AI models with U.S. technology.

Compared to the contention over semiconductors and generative AI, cloud services may seem innocuous. However, as cloud services become more ubiquitous and commercialized, any restrictions on Chinese companies would prove just as impactful as a chip ban.

Cloud Computing Liberates Chips’ Potential

To grasp the logic behind U.S. policy, one must first comprehend concepts like “the cloud” and “cloud services.”

Of course, companies could invest heavily in purchasing expensive, massive supercomputers equipped with advanced chips. They could further devote substantial human and financial resources to maintaining physical servers or data centers on-site. However, this traditional local deployment model requires lofty initial outlays and operating costs, rendering it cost-prohibitive for most organizations.

With the popularization of cloud technology, enterprises no longer necessarily need to own physical supercomputers to harness computing power. Cloud computing enables companies without expansive physical servers to obtain computing resources through the network, mimicking local access to physical computers.

As Microsoft explains, under the cloud computing model, computers reside in the cloud service provider’s data center rather than around the user. Users do not own the computers yet can utilize them, with leading tech companies like Amazon, Microsoft, Alibaba, and Tencent managing and maintaining these “computers.” Their performance and security naturally eclipse what average enterprise servers can offer.

“The cloud has the power to transform daily life, business, and society,” said Microsoft CEO Satya Nadella in his memoir. Today, Microsoft positions itself as a “mobile first, cloud first” global platform and productivity company, with the cloud as its business cornerstone. Microsoft’s remarkable performance in recent years underscores the cloud’s potential.

Leaving the Cloud Curtails Innovation

In the past, personal computer users had to periodically clear redundant data and uninstall applications to free up limited computing power and storage. This maintenance preserved the computer’s speed. However, by uploading data to the cloud, individual users can theoretically leverage unlimited storage, assuming they purchase adequate cloud services. Downtime is also minimized.

For organizations, cloud services can impart tremendous flexibility, resilience, and scalability. Management can fluidly deploy resources in response to various situations. Enterprises can purchase services as needed and only pay for actual usage, without expending many resources on physical computers.

The prevalent video conferencing and online classrooms throughout the pandemic truly relied on the ubiquity of cloud computing. As Nadella also mentioned, “Cloud computing enables us to analyze vast amounts of data, garner specific insights and information, and transform guesswork and speculation into accurate predictions.”

A Global Ecosystem

Cloud services generally refer to “public cloud” services. The more familiar Microsoft Azure, Amazon Web Services (AWS), Google Cloud, Alibaba Cloud, and Huawei Cloud all fall into this category.

In truth, some enterprises cannot utilize general public cloud services due to regulatory requirements, data sovereignty, and security considerations. Rather than establishing multiple local servers independently, large multinational corporations or highly regulated industries like banking and healthcare are more likely to employ “private cloud” services. This establishes an enterprise-specific cloud environment for employees across locations. The cloud provides data access and sparks innovation.

Such enterprises may also opt to place more sensitive data or applications in private clouds, while some lower-risk data can be transferred to public clouds in a hybrid model. This allows them to harness the benefits of cloud technology without compromising regulatory and security considerations. Many public cloud companies also offer private cloud services, and some software companies like Hewlett Packard Enterprise (HPE) and VMware aid companies in constructing exclusive private cloud environments.

Within the cloud operations network, countless software service providers also exist, furnishing users with thousands of third-party cloud services and forming an enormous ecosystem. Taking Microsoft Azure as an example, multiple software companies like SAP and Oracle provide services like network security, data management, development systems, and architecture optimization to users of the Microsoft cloud.

Enterprises act as both suppliers and consumers of one another, forming a highly complex ecosystem that promotes mutual innovation. This means migrating away from U.S. cloud service providers equates to leaving a dynamic global innovation community, which is extremely costly.

More Intricate Than Chips

The intricacies of the cloud service ecosystem involving operators, developers, suppliers, and users likely exceed that of chip manufacturing. The various parties’ applications reflect the multitude of ways humans harness computing power. After years of viral growth, the interdependent cloud service ecosystem makes restricting use no simple task. It may be more complicated than chip fabrication.

Returning to the U.S. media’s recent rumors, determining whether a cloud service utilizes advanced chip computing power would give U.S. Commerce Department experts a mighty headache. Do Windows and Office necessitate advanced chip support? Does running programs on Windows or Azure to develop AI models involve advanced chips? How to judge the myriad third-party cloud services? If a service is deemed to use advanced chips, can cloud operators still provide PaaS and IaaS services or be part of a hybrid cloud?

This may necessitate extremely complex technical arguments, with many officials exercising discretion in judgments. Numerous controversies, disputes, and legal challenges would likely ensue.

Furthermore, defining “Chinese customers” is even more arduous in the cloud service arena. With chip manufacturing, the few dozen Chinese chipmakers like Huawei’s Hisilicon Semiconductor and SMIC purchasing ASML lithography machines are easily identifiable. However, there are potentially tens of thousands of Chinese companies using U.S. cloud services. Moreover, many foreign-invested enterprises establish multiple entities in mainland China.

For instance, it is unclear if restrictions would impact AI analysis by KPMG China’s Shanghai office or ChatGPT use by Lenovo Group’s headquarters in North Carolina. Considering China’s multiple jurisdictions, Hong Kong and Macau companies may also be affected. Forced constraints on cloud service use could prove far-reaching, with a heightened chance of unintended consequences.

Cloud services are more multifaceted than chip exports and would affect more sectors in both nations. The U.S. government may need to reflect further. In recent months, Western countries have gradually accepted that domestic companies with more elementary technologies will remain in China. The U.S. government will likely not prohibit Chinese enterprises from using ubiquitous public cloud services like Windows, Office, and Photoshop. More probable is a concentrated clampdown on AI and big data use cases, significantly impacting Chinese companies in those realms.

Apart from Google’s unique stance, the cloud giants Microsoft and Amazon have enormous cloud businesses in China and greater incentive to lobby the U.S. government for looser controls that do not damage their China operations.

Complacency is Ill-Advised

Although fully restricting Chinese customers’ cloud service use is challenging for the U.S. government, determination to push forward could profoundly impact mainland China, Hong Kong, and Macau companies. If the Commerce Department’s restrictions are too broad, Chinese firms would need to drastically recalibrate their operating models in the short-term. Therefore, prudent management should prepare for adversity in times of peace and brace for the potential storm ahead.

Presently, the U.S. government has not revealed plans to further limit cloud services. Chinese companies using U.S. cloud services need not hastily migrate data and applications yet. In the short run, they should express concerns over U.S. policies to Microsoft, Amazon, and other cloud operator management as customers – the lobbying teams of American tech giants wield tremendous influence in Washington political circles. Commercial interests could incentivize them to actively lobby for a less disruptive plan.

As Sun Tzu’s Art of War states, “More calculations win, less calculations sometimes win, no calculations always lose!” Chinese companies should also conduct situational planning and carefully devise several contingency measures for possible developments. Management could enlist an American consulting firm to leverage its expertise to evaluate various likelihoods and business operation impacts, enabling proactive risk mitigation.