In conversation with Professors Chih-shian Liou and Chung-min Tsai
By Sarah Cao
On Monday 11 May 2026, the Oxford Taiwan Studies Programme hosted Professors Chih-shian Liou and Chung-min Tsai of National Chengchi University at the Kin-ku Cheng Lecture Theatre, China Centre, for a seminar on the political economy of Chinese high-tech development under intensifying US–China competition. Taken together, the two presentations offered contrasting verdicts on what state-led technological advancement can, and cannot, deliver in contemporary China.
Professor Liou’s talk, “Leapfrogging through Innovation Ecosystems: China’s Industrial Policy Regime and the Rise of Private High-Technology Companies,” examined how Beijing has moved beyond top-down subsidies toward a regime that integrates private firms into a state-led innovation ecosystem. Professor Tsai’s talk, “China’s Semiconductor Dilemma: Administrative Guidance or Market Orientation?,” asked why, despite vast funding and political attention, China’s semiconductor industry has so far failed to deliver the breakthroughs Beijing seeks. The juxtaposition was instructive in itself: where one panelist saw bifurcation and leapfrogging, the other saw structural collision and stagnation.
Innovation Ecosystems and the Rise of Private High-Tech Champions
Liou opened with an anecdote describing a BYD electric vehicle parked near the Eiffel Tower not long after pandemic-era travel restrictions were lifted. Both BYD and DJI, the drone manufacturer, are privately owned and have achieved global dominance in their respective sectors. Their rise, she argued, exposes a foundational puzzle for the political economy of Chinese state capitalism: how does a system widely understood to favor state-owned enterprises (SOEs) deliver leapfrog development through private firms that, by every conventional account, should sit at the periphery of industrial policy?
The puzzle also reveals a weakness in the existing literature. Conventional treatments emphasize SOE-led “national champions” backed by generous subsidies, with mixed results, and largely accord no role to private firms. Drawing on Dani Rodrik’s definition of industrial policy as “targeted and selective government policies that stimulate specific economic activities and promote structural change,” and on the policy-regime concept developed by Stephen Krasner and Barry Naughton, Liou proposed a reframing: in response to Sino–US strategic competition, China has incorporated private high-tech firms into a deliberately constructed innovation ecosystem, with a division of labor between SOEs and private firms that is now the engine of leapfrog development.
Two pieces of evidence anchored the argument. The 2020 Programme of New Infrastructure Construction (NIC) identified seven priority sectors, four of which are dominated by private companies; and among China’s major AI computing power centers, only Inspur is an SOE. Borrowing from Granstrand and Holgersson, Liou defined an innovation ecosystem as “the evolving set of actors, activities, and artifacts, and the institutions and relations, including complementary and substitute relations, that are important for the innovation performance of an actor or population of actors,” with institutions as “the rules of the game.”
Why test the argument on AI? Liou pointed to a dual crisis. The external pressure of US export controls intersected with the internal bottleneck of older, inefficient industrial policies, opening space for the Chinese state to select more capable actors by changing the rules governing resource allocation. The shift, she argued, has been from ownership-based to mission-based allocation, in ways “we’ve never seen before.”
Two flagship instruments illustrate the new approach. Government Guidance Funds (GGFs) operate under dual mandates: producing financial returns and achieving the state’s industrial policy goals via venture-capital-style decision-making. They function as patient capital for early-stage firms in emerging technologies and aim to crowd in 70 to 80 per cent of funding from “social capital,” signaling priority sectors to private investors. The National Venture Capital Guidance Fund (2025) and the National Integrated Circuit Industry Investment Fund (the “Big Fund”) are leading examples. Overlap between competing GGFs has produced inefficiency, but the model itself marks a real departure from older subsidy logic.
Challenge-based selection mechanisms, which Liou rendered as 揭榜挂帅 or “championing a task through open competition,” form the second pillar. A six-step procedure invites firms to bid for designated technical missions in order to overcome industry bottlenecks. The “Little Giants” programme, running across six batches from 2019 to 2025, has recruited more than ten thousand small and medium firms into a system of financial assistance, public-listing support, and R&D resources. Most are privately owned, including some Taiwanese companies. The sectoral distribution skews heavily toward manufacturing, with provincial variants such as the agricultural Little Giants of Fujian and Hebei.
Liou closed by arguing that private high-tech firms are now better equipped than SOEs for frontier innovation, and that an innovation ecosystem incorporating them is what now drives leapfrog development. The result, in her phrasing, is a “bifurcated industrial policy regime” in which the state has redefined the roles of public and private actors to suit different technological objectives.
The Semiconductor Dilemma: Why Money Has Not Bought Breakthroughs
Tsai opened with the question framing his current research: do US sanctions on China’s high-tech sector hinder semiconductor development, or accelerate self-reliance? His short answer was no. External pressure has not produced the breakthroughs Beijing intends; rather, US-led tech sanctions have exacerbated internal contradictions between the Chinese government and the industry.
“Made in China 2025” called for domestic semiconductor capacity of 40 per cent by 2020 and 70 per cent by 2025. By the end of 2023, the figure stood at 23 per cent. The deeper question, he argued, is why preventative measures are so difficult to implement when a socialist system is forced to exit a capitalist-constructed international division of labor. His analytical frame paired techno-nationalism against global interdependence.
After post-1980s expansion built on foreign technology and investment, China’s semiconductor industry remained concentrated in low- and mid-end segments and reliant on imports. The “Big Fund” era opened in 2014 with the National IC Industry Development Guidelines, mobilizing 138 billion RMB in national funding alongside a further 140 billion RMB in local funds, with the stated ambition of supporting the entire chain: design, manufacturing, packaging, and testing. Internal contradictions, he argued, have hindered effective upgrading.
The dilemma, in his account, is fundamentally political-economic. Administrative guidance, which Tsai glossed as “your boss is telling you what to do,” collides with corporate realities in which enterprises focus on profitability and market share that diverge from state strategic goals. The disconnect means incentives fail to drive innovation, producing technological stagnation and continued reliance on foreign technology. Projecting a chart of global wafer fabrication capacity by technology and region for 2022 to 2032, he noted that despite techno-globalism’s promise of a shared supply chain, “everyone feels insecure.” South Korea, the United States, Taiwan, and China each pursue self-sufficiency, even though, as he put it, “there is no way to build up a complete supply chain at home, as it is already shared by the entire world.”
Here Tsai diverged from Liou. Because so many SOEs sit at the center of the semiconductor sector, he argued, it is difficult for companies to innovate. The result is a self-fulfilling prophecy: the more state-led the sector becomes, the more it fails to ignite market-driven innovation. Subsidies expand market share rather than fund R&D, producing technological stagnation that in turn leaves the industry more vulnerable to external sanctions. Two cases illustrated the execution gap, the “Golden Card” episode and the 2006 Hanxin scandal, both involving state-owned firms, which exposed an incentive system rewarding visible expansion over genuine capability. His conclusion was a systemic collision: US sanctions will not produce Chinese self-sufficiency.
Q&A: A Bifurcated Regime, or a Two-Speed Economy?
The Q&A surfaced the productive tension between the two arguments. If Liou is right that ownership has been deprioritized in favor of mission-based allocation, how can Tsai’s account, in which money flowing to SOEs in semiconductors produces stagnation, also be correct? Tsai’s response emphasized the peculiarity of semiconductors as a hardware- and equipment-intensive sector: China lacks the necessary equipment, cannot easily import it, and finds it extremely hard to make on its own. Managers facing three-to-five-year promotion cycles cannot credibly run long-horizon R&D, and short-term incentives stifle the very innovation Beijing wants.
Audience members pressed on how far the bifurcation should be extended. DeepSeek and Huawei occupy ambiguous space in any private/state classification, and DeepSeek is reportedly now training models on Huawei chips, implying a build-up of technical capacity worth investigating. Liou pointed to Anhui province, where a fusion project promoted by the provincial government and anchored at the University of Science and Technology of China draws on small private companies as collaborators, as a site where the public-private boundary blurs in instructive ways.
Tsai observed that the United States remains the largest importer of mature chips from China, with active competition concentrated at the advanced end. China continues to acquire pre-owned manufacturing equipment, but that supply is finite. The strategic significance of the sector, particularly for military applications such as PLA fighter aircraft, explains why Beijing continues to invest disproportionately in SOEs even when the returns disappoint.
On the GGF mechanism, Liou clarified that the funds function differently from straightforward subsidies. They take equity stakes in private firms, are reviewed periodically by the fund’s performers, and reallocate capital from underperforming to winning investments. Even in provincial cases such as Anhui, where the fund is government-backed, a board still makes investment decisions, leaving some space for market discipline. On civilian-to-PLA technology transfer, she acknowledged the existence of mechanisms that are “not very transparent.”
On NVIDIA’s H200 chips, which Commerce Secretary Lutnick has indicated are not yet sold to China, and Jensen Huang’s reported push to expand sales there, Liou observed that China uses the power of its market in bargaining, creating a “shameful” position for Beijing as Huang prepares to join a US delegation to China this week. Western imaginings of AI innovation, she added, remain anchored in American and Western European models, in which a “mission-oriented economy” trickles down into a Silicon Valley ecosystem; in China, the role of leading universities such as Tsinghua reflects an architecture of its own.
Turning to the recent Cheng Li-wun visit to Beijing and the upcoming Trump–Xi summit, the panel suggested that the visit’s impact on Taiwan’s local elections would be limited, since those are dominated by domestic issues, and that the absence of a consistent voice from within the KMT had made Cheng’s trip salient in the media. China remains relatively stable on Taiwan, while Trump is “so unstable,” so continued volatility from Washington should not be ruled out. Liou added that Cheng’s trip seemed primarily aimed at consolidating power inside her party while drawing on outside authority for legitimacy. A telling aside from the audience captured the broader mood: although surveys have suggested that 70 per cent of young Taiwanese would fight for their country, in a recent classroom of two hundred students, none raised their hands.
Concluding Reflections
Taken together, Liou and Tsai met at a crucial analytic seam. Both accept that US–China competition is the catalyst reshaping Chinese industrial policy, and both treat the state, the market, and ownership as central variables. They part company, however, on whether Beijing has cracked the code. Liou sees a sophisticated bifurcated regime that has begun to deliver leapfrog gains in AI through private firms operating inside a state-built ecosystem. Tsai sees, in semiconductors, a sector so structurally bound to long horizons, equipment scarcity, and SOE-dominated incentives that even unprecedented resources have produced only modest technological progress.
The juxtaposition is the point. The question, the two talks suggest, is not whether the Chinese state can mobilize capital and political will at scale, which it manifestly can, but whether the design of its industrial policy regime suits the specific demands of a given technology. AI rewards agility, fast iteration, and a private-sector R&D culture; semiconductors reward decade-long investments, deep equipment ecosystems, and patient capital, none of which fit comfortably inside the incentive structures of Chinese state capitalism as currently configured. There will not be one Chinese high-tech story, but many, each shaped by the fit between the policy instrument and the technology it is meant to advance.
*
The seminar was hosted by Dr Bo-jiun Jing, Senior Research Fellow and Programme Manager in Taiwan Studies at the Oxford School of Global and Area Studies. It forms part of the Oxford Taiwan Studies Seminar Series, and this recap was produced in partnership between the Oxford Taiwan Studies Programme and St Antony’s International Review (STAIR).
