PRC Tech Firms Practice Proactive Alignment
Executive Summary:
- A new equilibrium is emerging in state–business relations following the cessation of regulatory campaigns against the technology sector in the People’s Republic of China’s (PRC). Private sector firms are increasingly pursuing “proactive alignment,” preemptively synchronizing their business models with state-directed strategic objectives before receiving explicit political directives to do so.
- State control over key resources constitutes the material foundation for proactive alignment. By enforcing commanding leverage over advanced computational infrastructure through mega-projects and directed subsidies, the Party-state renders comprehensive conformity a prerequisite for doing business in the PRC.
- Legacy platforms have pivoted their commercial core toward state-defined strategic priorities, but an ascendant artificial intelligence (AI) cohort operates “policy-native” architecture, prioritizing state alignment over profit maximization to avoid becoming political liabilities.
Beginning in late 2020, the People’s Republic of China (PRC) subjected its private technology sector to an unprecedented regulatory crackdown; but it has since engineered a visible recalibration. In February 2025, Chinese leader Xi Jinping hosted a Private Enterprise Symposium (民营企业座谈会) with six leading technology CEOs; and two months later the government promulgated a Private Economy Promotion Law (民营经济促进法), a political statement of intent that was widely interpreted as a regulatory retreat (Xinhua, February 17, 2025, April 30, 2025; China Brief, September 8, 2023; June 21, 2025).
Recent analysis has argued that Beijing is constructing a “hybrid model,” relying on institutionalized oversight mechanisms, statutory promises, and minority golden shares to govern the private sector (Foreign Affairs, February 20). Although this framework identifies visible top-down regulatory instruments, it fails to fully account for the private sector’s pivot. The economic history of PRC demonstrates that top-down directives targeting private enterprise frequently invite passive resistance, thereby neutralizing the intended impact. Substantive transformation consistently requires bottom-up drivers.
Instead, technology enterprises in the PRC have converged on an accommodation strategy to ensure survival. By pursuing “proactive alignment,” these firms preemptively synchronize their business models, capital allocation, and public postures with state-directed strategic objectives. Traumatized legacy platforms now pivot their commercial core toward state-defined strategic priorities, effectively eliminating the need for explicit regulatory coercion.
This dynamic is most pronounced among the newly ascendant artificial intelligence (AI) cohort. Emerging in the post-crackdown era, these entities are “policy-native” from inception. Founders deliberately configure corporate mechanisms to prioritize state alignment over profit maximization, recognizing that inadequate alignment precipitates catastrophic losses. The Party-state reinforces this posture concurrently by monopolizing critical computational resources. This systemic normalization of proactive alignment cultivates a highly durable integration of state and enterprise.
Legacy Platforms Retrofit Alignment For Survival
The crackdown against legacy platforms at the start of the decade delineated a political and regulatory baseline by signaling that operations contravening the will of the Party-state would be guaranteed punitive retaliation. The unilateral decision by rideshare firm Didi Global to proceed with its June 2021 listing on the New York Stock Exchange illustrated the cost of failing this alignment test. The Cyberspace Administration of China (CAC) severely penalized the firm, levying a Renminbi (RMB) 8 billion ($1.2 billion) fine (BBC, July 20, 2022). The CAC categorized Didi’s behavior as “feigning compliance while acting in opposition” (阳奉阴违、恶意逃避监管), which indicated blatant defiance of the regulatory authorities (CAC, July 21, 2022). Similarly, tech giant Alibaba’s aggressive push for Ant Group’s dual listing in late 2020 prompted a state-mandated suspension, subsequently inviting a record RMB 18.2 billion ($2.7 billion) penalty (Xinhua, April 10, 2021).
Today, the Party-state defines the survival path for private technology enterprises, as formalized by the February 2025 Private Enterprise Symposium. At the symposium, Xi Jinping declared that policies governing the private economy would remain in effect, directing private entities to “resolutely serve as builders of socialism with Chinese characteristics and promoters of Chinese modernization” (坚定做中国特色社会主义的建设者、中国式现代化的促进者). This effectively demanded the permanent tethering of private enterprise to the state apparatus. He further stated that the private sector “must unify thinking and action” (要统一思想和行动) with the Party’s perspectives on domestic and international issues, and with the Party’s economic policies, thereby mandating comprehensive alignment with central directives (Xinhua, February 17, 2025).
Alibaba immediately moved to proactively align with Xi’s directive. The industry giant committed an RMB 380 billion ($55.3 billion) three-year capital expenditure plan directed exclusively at cloud and AI infrastructure (Alizila, February 24, 2025). Financial media defined this as Alibaba’s “second growth curve” (二增长曲线), representing a pivot toward national technological priorities (Yicai, June 26, 2025). This proactive alignment yielded significant commercial dividends, driven in large part by state-affiliated procurement. In the year to the end of September 2025, cloud revenue growth accelerated to 34 percent, and the firm deployed approximately RMB 120 billion ($16.9 billion) in capital expenditure toward AI and cloud infrastructure. By that point, AI-related product lines had sustained triple-digit growth for nine consecutive quarters as enterprise clients expanded adoption of AI infrastructure (Alibaba Group, November 25, 2025). Management stated that the pace of AI server deployment could not keep up with the growth of client orders, indicating that demand, substantially from public-sector and state-affiliated enterprises, continued to outstrip supply (Changjiang Times, November 27, 2025).
Many leading Chinese tech firms today are global companies. As a result, proactive alignment also has a transnational dimension. ByteDance is an illustrative case: by internalizing the Party-state’s geopolitical red lines into its global corporate structure, Beijing mandates sovereign control over core algorithmic assets, restricting their transfer through technology export regulations. Facing escalating pressure from the Trump administration (and by the U.S. Congress) to divest TikTok’s U.S. operations, ByteDance executed a targeted joint venture reorganization. The firm ceded majority ownership of its U.S. entity to U.S. investors led by Oracle while structuring the transaction to preserve ByteDance’s control over the global platform and its underlying algorithmic architecture (The White House, September 25, 2025; TikTok Newsroom, January 22). This ownership framework accommodates Beijing’s technology export regime. By retaining algorithmic control in exchange for its global expansion, ByteDance proved that proactive alignment supersedes commercial imperatives.
AI Cohort Builds In Party Line
In the PRC today, DeepSeek is arguably the most prominent AI firm, making it a useful case study of an emerging form of proactive alignment. DeepSeek has consistently operated in policy-native ways since its inception. Incubated within the quantitative hedge fund High-Flyer Capital Management (幻方量化), DeepSeek operates under strict partition that separates the AI developer from quantitative finance, and distances the enterprise from speculative capital (21st Century Business Herald, January 14). Funded exclusively through internal trading revenues, the firm bypassed foreign venture capital, aligning with Beijing’s demand for indigenous, controllable AI. This financial insulation also neutralized external pressure for rapid commercialization. Anticipating sovereign computational bottlenecks, its founder executed a strategy of “patient capital” (耐心资本), defined as state-aligned, long-term funding immune to immediate market cycles. High-Flyer invested RMB 200 million ($29.1 million) in 2019 and RMB 1 billion ($145.6 million) in 2021 to stockpile roughly 10,000 advanced GPUs prior to sweeping export controls (36Kr, May 25, 2023).
This resource reallocation carried a less visible logic. By 2021, High-Flyer’s quantitative trading operations faced intensifying political exposure: the fund’s assets under management had surpassed RMB 100 billion ($14 billion), but mounting public hostility toward algorithmic trading and regulatory scrutiny of the quantitative sector were eroding the industry’s political standing (Securities Daily, April 8, 2023; The Paper, February 20, 2025). The firm publicly announced in April 2023 that it would “concentrate resources on … AGI” (集中资源 … 探索 AGI), with a senior executive pointedly declaring that “AGI is not for stock trading” (AGI不是用来炒股的) (Yicai, April 16, 2023). This identity decoupling preceded by ten months the February 2024 regulatory crackdown in which exchanges publicly censured and suspended the account of leading quantitative fund Lingjun Investment (灵均投资), which constituted a warning shot directed at the sector as a whole (Securities Times, February 27, 2024).
When DeepSeek subsequently released the open-source V3 and R1 models, state media praised the firm for reducing systemic hardware dependency and catalyzing ecosystem-wide AI adoption (Xinhua, January 29, 2025). The founder’s January 2025 appearance at Premier Li Qiang’s (李强) consultative forum confirmed a political status transformation that quantitative trading success could never have delivered (China Brief, February 12, 2025). Beyond software, DeepSeek has prioritized domestic chipmakers Huawei and Cambricon (中科寒武纪科技) with early V4 access, sidestepping U.S. alternatives amid supply chain localization efforts (Financial Times, February 28). Ultimately, DeepSeek’s choice of the permissive MIT software license went substantially beyond what any policy document required. The March 2026 Government Work Report, which announced direct support for “AI open-source community development” (支持人工智能开源社区建设), merely formalized a direction of travel that DeepSeek had already pioneered (Xinhua, March 5).
Zhipu AI (智谱华章科技) provides another example of anticipatory alignment with Party-state preferences. Following its January 2025 inclusion on the U.S. Department of Commerce’s Entity List, the company bypassed temporary workarounds to embed itself completely within the domestic semiconductor ecosystem (eCFR, accessed March 18). Partnering with Huawei, Zhipu trained its multimodal model GLM-Image entirely on the firm’s Ascend processors and MindSpore framework (Securities Times, January 14). It also engineered custom optimizations exclusively for Ascend’s architecture, deliberately abandoning the Nvidia-dependent toolchains favored by domestic peers. By February, Zhipu scaled this approach to GLM-5, an open-source model trained entirely on 100,000 Huawei Ascend 910B chips (NxCode, February 13). This trajectory illustrates the critical distinction between passive compliance and proactive alignment. While U.S. sanctions imposed initial constraints, Zhipu responded by building a full-stack domestic training capability and ensuring compatibility with other state-backed chipmakers like Cambricon and Moore Threads (摩尔线程). This preemptive integration synchronized the firm’s technical architecture with Beijing’s computational self-reliance mandates well before any explicit directive required it.
Analysis characterizing these “new chives” (新韭菜) as the next cohort of private-sector wealth waiting to be harvested by the state misreads the institutional logic (Foreign Affairs, February 20). The preceding generation of technology founders built commercial empires whose scale and autonomy eventually provoked political friction. Firms like DeepSeek and Zhipu, meanwhile, have configured their enterprises to operate within state priorities from inception, structuring funding, hardware dependencies, and product strategies so that alignment is the default condition rather than a costly retrofit. Where the previous generation’s survival required reactive compliance after a political crisis, the new generation’s survival depends on preemptive integration, rendering firms incapable of becoming a political liability in the first place.
Figure 1: The ‘Policy-Native’ AI Ecosystem
(Source: compiled by the authors based on corporate filings, National Development and Reform Commission policy documents, and industry reporting.)
Resource Dependence Drives Alignment
State control over key economic and technological resources constitutes the material foundation for proactive alignment across the PRC’s technology sector. Private technology enterprises secure superior resource allocations by demonstrating conformity to central directives. Using the AI sector as an example, the Party-state exercises commanding leverage over advanced computational infrastructure, operationalizing this control through mega-projects and directed subsidy frameworks. Amid U.S. semiconductor export controls, state mediation supersedes free-market capital in governing access to these critical assets.
At the infrastructure level, megaprojects such as Eastern Data, Western Compute (东数西算) centralize processing power by routing computational workloads through state-directed facilities (NDRC, February 16, 2022; China Brief, March 2025). The Ministry of Industry and Information Technology (MIIT) integrates this hardware by deploying “national computing interconnection nodes” (国家算力互联互通节点) (MIIT, February 6). Functioning alongside this grid, authorities steer development through directed subsidies, allocating “computing vouchers” (算力券) and “training-power vouchers” (训力券) via government-invested centers. These mechanisms provide discounted or free processing capacity to startups routing research toward state-priority domains. In March 2025, Shenzhen distributed nearly RMB 200 million ($27 million) in training-power vouchers to over 40 AI and embodied intelligence firms, with individual grants reaching RMB 10 million ($1.45 million) (Southern Metropolis Daily, April 3, 2025). Shanghai followed with RMB 6 billion ($870 million) in compute vouchers offering up to 100 percent rental subsidies for qualifying firms, alongside RMB 3 billion ($435 million) in model vouchers and RMB 1 billion ($145 million) in data vouchers (Shanghai Municipal Commission of Economy and Informatization, July 29, 2025). The eligible domains, including intelligent chips, embodied intelligence, brain-computer interfaces, and AI for science, mirror the central leadership’s strategic priority list, channeling subsidized computing power toward state-favored sectors.
Figure 2: The State Computation Monopoly Network
(Source: Compiled by the authors based on NDRC policy documents, provincial computation subsidy programs and industry reporting.)
This state-engineered resource dependency shapes the political signaling behavior of the technological elite. The 2026 Two Sessions (两会) formalized a macroeconomic shift toward the “intelligent economy” (智能经济), elevating computing power to a foundational national resource. As one delegate to the National People’s Congress and university vice president put it, “the intelligent technology is taking a new step in reshaping the underlying logic of China’s economy” (智能技术在重塑中国经济的底层逻辑中迈出新步伐) (Xinhua, March 8; Economic Daily, March 11). In response, technology executives demonstrated rapid synchronization. Founder of 360 Group Zhou Hongyi (周鸿祎) and Xiaomi CEO Lei Jun (雷军) submitted legislative proposals mirroring the state’s strategic bottleneck list, addressing domains from inference compute layouts to humanoid robotics standards (Beijing Daily, March 4).
Such synchronized corporate accommodation functions as “biaotai” (表态), or public declaration of position, serving as a visible demonstration of political alignment. The signal conveyed to the central leadership is that state priorities have been received and adopted. Given the Party-state’s control over key resources, a biaotai functions not merely as public relations but as an operational prerequisite for continued access to subsidized computing power, procurement channels and regulatory approval.
Conclusion
The convergence of lessons learned by legacy tech platform companies, the emergence of “policy-native” AI firms, and calibrated political signaling from central and local governments marks a substantive transition toward a new equilibrium in the PRC’s state–business relations. Proactive alignment represents a manifestation of Party-state influence: private technology enterprises deliberately reconfigure their business models because state control over key resources renders alignment a primary survival strategy. Overt punitive campaigns have consequently become less necessary as control is internalized. State influence now operates through embedded corporate incentives, leading technology firms to anticipate and conform to central priorities before formal directives are issued. This marks a departure from both earlier forms of political accommodation, in which firms treated compliance as a constraint on profit-maximization and adjusted reactively when they overstepped, and from standard commercial adaptation, in which alignment would be merely instrumental. Today, alignment is the primary objective; profit follows.
Under current conditions, proactive alignment is driven by structural factors. This rationale ultimately bolsters the PRC technology sector’s capacity to pursue computational self-reliance and reduce vulnerability to external decoupling pressure. Proactive alignment nevertheless introduces its own vulnerabilities. State-affiliated theorists acknowledge these risks, explicitly warning that “new quality productive forces” (新质生产力) must not become a catch-all category for undisciplined capital misallocation driven by the blind pursuit of policy favor (Beijing News, November 3, 2025). These internal calibrations, however, have yet to gain policy traction, allowing the mechanisms driving proactive alignment to deepen.
