China Technology 2025: Fragile Tech Superpower

This chapter is part of the China Forecast 2025 report – read the Foreword and download the PDF here.

Our call:

  • By 2025, China’s technology ecosystem will have matured and be on par with Silicon Valley in terms of dynamism, innovation, and competitiveness.
  • That dynamism will increasingly take the form of industrial applications of information technology, as the locus of Chinese innovation shifts from the consumer internet to the industrial internet.
  • China will largely succeed in deploying highly capable “new infrastructure”—cloud computing, 5G networks, smart cities, and surveillance networks, among others—to facilitate this transition to the industrial internet.
  • US export controls on semiconductors will act as a modest brake on China’s new infrastructure rollout. But expanding restrictions on semiconductor manufacturing equipment (SME) will mean that China remains vulnerable to future interruptions to its supply chain for advanced chips.

Key assumptions:

  1. Development of new infrastructure will be heavily prioritized in the 14th Five-Year Plan (FYP), in an attempt to generate new sources of growth from applying emerging technology to traditional industries.
  2. The US government will largely maintain export controls on Huawei, but pressure from the US semiconductor industry and allies will prevent a major expansion of those controls that would block chip exports to tech giants such as Alibaba or Tencent.
  3. There will not be major exogenous shocks to the technology landscape (e.g. game-changing breakthroughs in quantum computing or synthetic biology) that would fundamentally alter the global balance of technological capabilities.

Leading indicators:

  1. Rollout of standalone 5G networks across the entire Jing-Jin-Ji Capital Economic Zone, the Yangtze River Delta region, and the Pearl River Delta Economic Zone.
  2. China’s cloud computing market surpasses $30 billion in 2022, doubling its 2020 total.
  3. Commercial production of advanced 7-nanometer (nm) chips by the leading Chinese chip fab, Semiconductor Manufacturing Industry Corporation (SMIC). 

Base Case (60%)

In our base case, China’s technology ecosystem over the next five years will have matured and will be largely successful in executing a transition from the consumer internet to the industrial internet. Specifically, the Chinese Communist Party (CCP) and private industry will try to use powerful emerging technologies like artificial intelligence (AI) and 5G to revamp and upgrade Chinese cities and traditional industries, such as manufacturing, agriculture, energy, and transportation.

This focus will mark a notable shift from the last decade of consumer technologies centered on apps like WeChat, Alipay, and TikTok. Instead, the Chinese tech landscape will likely enter a more capital-intensive phase as it targets industrial applications such as smart grids, dark factories, and autonomous vehicles (AVs).

By 2025, China will likely be a world leader in industrial technology applications, but that leadership will continue to depend on access to advanced semiconductors manufactured outside of mainland China. Our base case is predicated on three factors:

  1. The rollout of “new infrastructure” in key industrial and urban regions;
  2. The rise of a new generation of Chinese industrial tech startups;
  3. Progress by China in the design of semiconductors, but enduring weakness in the manufacturing of the most advanced semiconductors.


1. New Infrastructure Paves the Way

The term “new infrastructure” first surfaced in Chinese government language in 2018, but it picked up major steam in policy circles during the first half of 2020. Over the next five years, the push to build this digital infrastructure and use it to upgrade traditional industries will be central to China’s technological development.

The National Development and Reform Commission has outlined three categories of new infrastructure: “information infrastructure” (data centers, cloud computing, 5G networks), “integrated infrastructure” (smart cities, smart energy), and “innovation infrastructure” (STEM education, supercomputers) (see Figure 1).

In this triumvirate, “information infrastructure” is the physical foundation that can accommodate the industrial-scale flow and processing of data. The second pillar of “integrated infrastructure” consists of the productive applications, often large-scale, that are built atop that foundation. Finally, “innovation infrastructure” refers to the tools for cultivating and empowering human capital (Chinese researchers, engineers, and students) who can push the frontiers of indigenous technology.

Figure 1. The Structure of the New Infrastructure Initiative
Source: MacroPolo.

By stitching these segments together, Chinese policymakers aspire to create cities in which traffic flows and electricity grids are constantly optimized by AI algorithms. They aim to build airtight surveillance networks that allow automated monitoring of all public spaces nationwide, approximating the comprehensiveness of these networks in Xinjiang but without the mass internment of local citizens.

They envision 5G networks enabling fully automated “dark factories” and the deployment of AVs. And they hope to create universities where researchers have access to abundant computing resources and cutting-edge instruments. This techno utopian and dystopian vision remains a ways off from messier realities, but it serves as a guiding light for policymakers.

Making progress toward that goal depends on the active involvement of private technology companies. In our base case, those companies will actively embrace the new infrastructure push due to political-commercial alignment domestically, as well as major constraints internationally.

Domestically, both private Chinese technology companies and the Chinese government have mutual interest in the success of the new infrastructure initiative. Leading firms such as Alibaba, Tencent, and Baidu see enormous commercial opportunities in “information infrastructure” (e.g. cloud computing) and “integrated infrastructure” (e.g. AVs and smart cities). From the CCP’s perspective, the social media apps of the past decade were often seen as a threat to its control over information, but industrial internet applications fit well with the Party’s view of technology as a tool for advancing its economic and political agenda.

Internationally, Chinese tech companies see their opportunities narrowing, particularly in Western markets. While these tech companies still have meaningful footholds overseas, especially in developing countries, international expansion no longer looks like a reliable source of growth in the medium term. This will force the companies to double down on domestic sources of growth, expanding beyond consumer-facing apps into enterprise services and products.

China’s new infrastructure rollout will face constraints on both local government access to financing and on the supply of semiconductors, particularly for Huawei (see below for more detail). In our base case, however, China will largely overcome these constraints in the medium term by initially focusing resources on pilot programs in key regions.

On financing, the central government will want to avoid another investment binge that could saddle regions with debilitating debt and litter the country with “ghost data centers.” Both local governments and tech juggernauts have pledged some eye-catching sums for new infrastructure development, but a large portion of these expenditures were likely already baked in and have simply been rebranded as “new infrastructure” investments. On semiconductors, Huawei’s dwindling stockpiles of advanced chips mean that China will initially have to pick its spots with the deployment of standalone 5G.

Given those constraints, the central government will choose to initially concentrate new infrastructure spending in key industrial and urban regions with the greatest immediate economic upside: the Jing-Jin-Ji Capital Economic Zone around Beijing, the Yangtze River Delta region surrounding Shanghai, and Guangdong’s Pearl River Delta. 

These regions have the necessary fiscal resources and human capital to make new infrastructure investments work, with major projects already underway in smart manufacturing and AVs. The addition of standalone 5G networks will enable more experimentation in areas such as smart cities, energy, and agriculture. Other regions with a specific industry cluster (Guizhou province’s cloud computing cluster or the massive surveillance architecture in Xinjiang) will also see new infrastructure investment, but the focus will initially be on key coastal hubs.

Forty years ago, these coastal regions acted as laboratories for China’s experiment with global trade and foreign investment. In the next decade, they will again become test beds for a major experiment in applying emerging technology to industrial and urban problems. 

China will be largely successful at building new infrastructure in these regions, leading to real economic dividends around 2022. Those experiments will then be expanded to other regions, and the gains will become more nationally distributed toward the end of the 2022-2025 period.

2. New Startups Take the Lead for the Industrial Internet

The benefits from the growth of China’s industrial internet will not be distributed equally across its technology companies. In our base case, China’s incumbent technology giants will take the lead in building out much of the physical “information infrastructure,” but a new batch of emerging Chinese startups will reap the greatest rewards from the diverse products built atop that infrastructure. Of the incumbent juggernauts, Alibaba is likely best positioned to capitalize on the new infrastructure initiative.

The reason for this lies in the incentives and business models of large platform companies like Tencent, ByteDance, and Baidu. These companies all succeeded by building software platforms used by hundreds of millions of Chinese consumers, with most revenue derived from advertising or fees attached to standardized services. 

That platform model is relatively transferrable to information infrastructure like cloud computing. China’s tech giants saw what the growth of the cloud market did for their US peers, and they have poured resources into recreating that success at home. China’s cloud market clocked in at 67% annual growth in the last quarter before Covid-19 hit, with Alibaba, Baidu, and Tencent all pledging major investments in cloud over the next five years (see Figure 2).

Figure 2. China’s Cloud Market Sees Rapid Growth
Source: Canalys.

But building the more targeted industrial applications made possible by new infrastructure requires a different business model, one centered on highly customizable enterprise technology products. For example, a robotic inventory sorting system needs to be tailored to each customer’s needs. Each new contract will likely require new engineering work and after-sales service. Tech juggernauts often lack the incentive structures to compete effectively in these verticals, except in the case of extremely lucrative contracts for large products such as smart city systems.

Instead, a new generation of Chinese technology startups will grow to fill these markets. Their niche products will leverage new infrastructure for industrial upgrading: smart agriculture systems, autonomous delivery drones, and enterprise software managing all of the above. These startups may not grow into the sprawling behemoths that their predecessors did, but they don’t necessarily have to. They can build a useful and profitable business around a few concrete use cases and a few large customers.

Large companies will continue to lead in some applications of new infrastructure, particularly high-profile projects that involve working closely with municipal or provincial officials. Alibaba has already built out a number of “City Brain” systems, and Baidu is piloting its fully autonomous “robotaxi” fleet in Changsha, Hunan. Among the incumbent juggernauts, Alibaba will likely see the largest upside from new infrastructure because it already controls 40% of the Chinese cloud market and has had some early success with smart cities.  

But for the hundreds of more niche products that will make new infrastructure valuable across so many industries, it will be this new batch of startups leading the way.

3. A Mixed Bag of Chips 

Access to cutting-edge semiconductors remains a potentially serious obstacle to China’s medium-term technology trajectory. Here China faces two external challenges: 1) restrictions on the export of advanced chips to certain Chinese companies such as Huawei; and 2) restrictions on the export of semiconductor manufacturing equipment (SME)—the crucial tools needed to manufacture chips domestically.

Our base case holds that through 2025, export controls on chips will have a modest impact on China’s ability to execute on its digital infrastructure initiative. During this period, export controls targeting Huawei will substantially slow the nationwide rollout of standalone 5G networks, but it will not have a debilitating impact on most other new infrastructure applications. However, during this period China will struggle to gain ground in the fabrication of advanced semiconductors due to controls on SME exports. That lack of access to SME means China will continue to be dependent on foreign suppliers for advanced chips, and thus remain vulnerable to supply disruptions (see Figure 3).

Figure 3. Global Market Share of Semiconductor Sales, 2019
Source: Semiconductor Industry Association.

In the short term (2020-2022), the greatest impact of US export controls on chips will be hampering Huawei’s deployment of 5G networks in many parts of China. The company may have stockpiled enough inventory to install standalone 5G in a few key regions, but it has already slowed down the construction of new 5G base stations.

Our base case holds that, in response, China will seek to reconstitute its 5G industry around companies that are not subject to US export controls. Huawei—in consultation with the Chinese government—will likely license its large library of standards-essential 5G patents to domestic and foreign competitors, such as ZTE, China Information and Communication Technologies, and Sweden’s Ericsson. In addition, Huawei’s large workforce of radio frequency engineers experienced in 5G deployment will likely migrate to these companies, transferring their experience across a broad range of companies that attempt to continue building 5G in China.

This disruption and the messy process of re-bidding existing contracts will delay the national rollout of standalone 5G networks until 2022 or 2023. But these delays are unlikely to fundamentally affect other economically productive parts of China’s new infrastructure. Many early experiments will first be conducted in coastal regions where standalone 5G networks can be prioritized for early deployment. In the meantime, the rest of the country can begin harvesting other low-hanging fruit from the many non-5G applications of new infrastructure technology.

While export controls on advanced semiconductors will likely have a modest effect over the next five years, our base case holds that restrictions on the export of SME will put a ceiling on China’s ability to domestically fabricate advanced chips in the medium term.

Successive generations of semiconductors are classified by their “node” (also called “process”), with lower nodes representing more advanced chips. For much of the past 20 years, Chinese semiconductor fabrication plants (“fabs”) have lagged 2-3 nodes behind the world’s leading fabs. As of 2020, the world’s most advanced semiconductor fab (TSMC in Taiwan) has begun producing chips at the 5 nm node. China’s leading semiconductor manufacturer, SMIC, is currently three nodes behind, fabricating at the 14 nm node, though at low volume and with high manufacturing error rates. SMIC plans to begin limited production at the 7nm node in 2021.

Progressing to the next node is an immense challenge that often requires the purchase of new SME. Supply chains for SME are global, and the large majority of manufacturers use some US components in their production process. Limiting Chinese access to crucial SME gives the US government leverage, which it can exercise either by narrowly restricting direct US exports to Chinese fabs (using the Entity List) or through more expansive international controls that bar international SME companies that use US components from selling to China (using the Foreign Direct Product and de minimis rules). 

If these expansive international controls were taken to the extreme, they could temporarily cripple Chinese fabs like SMIC by cutting them off from the supply or servicing of crucial SME components.

In our base case, the United States will likely selectively limit Chinese access to SME, effectively preventing Chinese fabs from progressing past the 7 nm node. It will, however, avoid the “nuclear option” of entirely cutting off Chinese fabs from foreign SME exports (see more in the secondary scenario below). The goal of capping Chinese progress at 7 nm could be achieved by working with allies such as the Netherlands to formalize narrow controls on the export of the most advanced EUV photolithography machines, which are needed to make 5 nm or below chips.

These more targeted restrictions—blocking chip sales to Huawei and restricting but not entirely blocking SME exports to China—won’t pose a grave threat to Chinese technology over the next five years. Most Chinese companies will still be able to source leading node chips from abroad, and China’s leading fab will continue making incremental progress toward 7 nm fabrication. 

But as leading global fabs like TSMC and Samsung progress from 5 nm to 3 nm and beyond, China’s inability to advance past 7 nm will present a major vulnerability for its technology ecosystem. That vulnerability will give the United States, Europe, Japan, and South Korea enduring leverage in dealing with China’s technological advancements.

Secondary Scenario (40%)

In the secondary scenario, China will struggle to deploy and leverage new infrastructure due to a combination of three factors: 1) debilitating frictions in applying new infrastructure applications to traditional industries; 2) long delays in restructuring China’s 5G industry; and 3) major expansions of US export controls on semiconductors and SME. In this case, China’s new infrastructure push turns into a boondoggle, and the country proves unable to leverage emerging technologies for upgrading traditional industries. 

Frictions, missteps, and waste are usually part and parcel of a Chinese industrial plan on the scale of new infrastructure. But whether those frictions end up undermining the plan as a whole depends on their degree. In our secondary scenario, these difficulties in implementation prove costly enough that they undermine the real-world impact of new infrastructure.

Rather than private companies and startups, many of the most promising industries for new infrastructure upgrading are dominated by state-owned enterprises, companies not known for their nimbleness or innovation. These entities prove incapable of leveraging new infrastructure, and large expenditures turn into debt accumulation on the public ledger.

In addition, the rollout of standalone 5G networks proves far messier, setting back China’s progress by up to four years. This may result from misguided government attempts to keep Huawei on corporate life support or the successive US blacklisting of any Chinese or foreign company working on 5G in China.

This leads into the third aspect of this scenario: a major expansion of US export controls on chips and SME to China. In this secondary scenario, restrictions on chip sales are expanded to include some of China’s leading cloud providers such as Alibaba. In addition, US restrictions on SME exports are unilaterally expanded to cover most global sales of SME to China that use significant US components, as well as the servicing of existing equipment, effectively crippling production lines at Chinese fabs for a time. In the short term, these actions would constitute an enormous blow to China’s technology ambitions, one that would send the country’s ecosystem reeling. 

But over the medium and long term, they could also backfire on the United States if done unilaterally. Such a move would alienate key allies in Europe and Japan that produce SME, and incentivize all global SME companies to strip American components out of their own products to avoid US export controls. Replacing those US parts would be a slow process (experts estimate 5-10 years), but once completed it would effectively eliminate US influence over foreign SME suppliers and dramatically undercut the domestic US SME industry. 

In essence, this is a card that can only be played once, and in this case Washington chooses to do so during 2020-2023, giving the United States a one-off bump in technological competitiveness by temporarily cutting off Chinese access to advanced semiconductors. In doing so, the United States sacrifices its long-term leverage over China and its credibility with allies, while also hampering the ability of US semiconductor companies to innovate and compete.

Matt Sheehan is a Fellow at MacroPolo. You can find his work on tech policy, AI, and Silicon Valley here.

Get Our Stuff