In this context, progress in frontier industries — particularly general-purpose technologies like AI that promise broad, economy-wide spillovers — is not just a matter of national pride, but also a key strategy in service of CCP goals. And in an era where global competition is increasingly framed in zero-sum, militarised terms, the CCP’s push for technological supremacy takes on the urgency of existential necessity. It is no coincidence, then, that China has long treated AI as a national priority.
In 2017, the State Council issued the Artificial Intelligence Development Plan (AIDP), laying out a comprehensive road map for China to be the global leader in AI by 2030. This also explains China’s unlikely emergence as a global first mover in AI regulation: as a by-product of the CCP’s long-standing efforts to assert control over information flows and content dissemination.
Economist Chenggang Xu’s concept of “regionally decentralised authoritarianism” offers a useful lens for understanding the hybrid nature of China’s governance, where centralised political control coexists with decentralised economic management. In practice, China’s economy is shaped by the dynamic interplay of state priorities, market incentives and institutional dynamics. It is a system that defies the neat framing often imposed by Western observers. Within, innovation is not just tolerated but actively encouraged under politically acceptable limits. Where America has historically excelled at innovation through market-driven incentives and a robust talent pipeline, China paradoxically competes through state-led central direction, benefiting from an unparalleled ability to mobilise resources, harness economies of scale, and prioritise applied outcomes. Just as US export controls inadvertently catalysed the explosive growth of China’s open-source ecosystem, Beijing’s heavy hand often creates unintended ripple effects: amplifying certain innovations, suppressing others, and ultimately shaping the direction of technological progress. But for all its detractors, it is not inherently dysfunctional. China’s model has repeatedly delivered outcomes that confound Western assumptions. The logic may be different, but it is not inherently weaker. Different routes, same goal.
Central command
See also: Trump wins the trade war — and the advantage of future US relative productivity gains
China’s AIDP 2017 aligns with the CCP’s broader industrial vision under Made in China (MIC) 2025 — a strategic blueprint launched in 2015 to achieve tech sufficiency in high-priority sectors within the decade. Impressively, a 2024 Bloomberg review of China’s progress found that Beijing has achieved near parity or leadership across most sectors identified under MIC 2025, except for commercial aircraft manufacturing (see table below). If nothing else, these results demonstrate the efficacy of Beijing’s strategy of centralised state direction and sector-specific targets, an alternative counterpoint to Washington’s bottom-up model of innovation.
Speaking of the semiconductor industry alone, economist and former White House adviser Michael Boskin once quipped, “Potato chips, computer chips, what’s the difference? ... A hundred dollars of one or $100 of the other is still $100.”
It was not until the pandemic-induced chip shortages that Washington made a policy about-face, as seen through initiatives like the 2022 CHIPS and Science Act. In contrast, China has long treated semiconductors as a strategic priority.
See also: Trumpianomics will benefit the US greatly for years to come
As we covered last week, its progress in this arena has historically been constrained by gaps in research and development (R&D) capabilities, a dearth of quality talent and, more recently, US export restrictions limiting access to advanced tools. Nevertheless, Beijing’s push to uplift the domestic semiconductor sector has remained steady and deliberate: the US$48 billion National Semiconductor Fund (alternatively named Big Fund) conceived under MIC 2025 is a decade-long programme aimed at developing domestic chip capacity. Its third phase, announced last year, places specific focus on addressing chokepoints in lithography tools, electric design automation (EDA) software and advanced memory chips.
Since the 19th Party Congress in 2019, Beijing has increasingly framed AI as a strategic priority under the juguo (whole-of-nation) approach, a state doctrine that mobilises national resources in pursuit of strategic goals, at whatever cost. While US dominates in private capital and venture capital (VC)-backed funding, the picture is reversed when it comes to public-sector investments. China leads largely through subnational “investment guidance funds”.
One recent example is the RMB1 trillion national VC guidance fund announced in March this year that targets long-term funding for “frontier domains” — including AI, quantum technology and hydrogen energy storage — over a 20-year horizon.
Complementing this financial power are a range of centrally coordinated initiatives aimed at transforming China’s digital and industrial infrastructure. Prominent examples include the New-Type Infrastructure plan that launched in 2020, which committed US$1.4 trillion to a five-year national buildout of 5G, Internet of Things (IoT) and AI infrastructure; the Eastern Data Western Computing project from 2022, an industrial planning megaproject which allocates data centres to western provinces where energy and land are more abundant, to support data-hungry and more digitally-developed coastal cities in eastern China.
In recent years, the government has also made great efforts to address existing shortfalls: removing patent subsidies to eliminate incentives for poor quality submissions, as well as reorganising the Ministry of Science and Technology to prioritise tech and innovation through establishing the smaller and more nimble Central Science and Technology Commission, which has authority to execute policy and direct funding. Separately, the Digital Silk Road (the digital dimension of the ongoing Belt and Road Initiative) seeks to extend China’s technological influence abroad via global investments in satellites, subsea cables, smart cities and so on.
Pragmatism at scale
Another point of divergence that stems from technology serving a dual role as a policy instrument for the CCP is the type of innovation that it prioritises. China has historically placed distinct emphasis on applied technologies that have immediate industrial utility. Technologies that, in other words, translate more easily into tangible socioeconomic returns — creating jobs, boosting productivity and generating substantial gross domestic product (GDP) growth. This emphasis on function is even embedded in language.
For more stories about where money flows, click here for Capital Section
Take “artificial general intelligence” (AGI), for instance, which is often seen as the holy grail of AI development. OpenAI’s founder Sam Altman speaks of it in hyperbolic terms of being “superintelligence in the true sense of the word” and the “glorious future”, whereas the Chinese term for AGI is far more utilitarian, translating as “general purpose AI”.
This pattern of prioritising the practical has played out to great effect before. In the early days of the internet, China lagged in foundational technologies but soon leapfrogged the West in e-commerce, e-payments and industrial robotics. These remain sectors where China dominates today. And a similar trajectory appears to now be unfolding in frontier technologies like embodied AI (humanoids and other robotics) and electric vertical take-off and landing (eVTOL) vehicles; or, more colloquially, “flying cars”. Case in point: According to Morgan Stanley, as of 2024, more than half of the companies globally confirmed to be working on humanoid robots are based in China.
In large part, China benefits from a structural advantage that is uniquely its own: scale. The sheer size of its domestic market provides fertile ground for rapid deployment and iteration, allowing companies to drive down unit costs and break even faster.
Take Xiaomi, the consumer electronics giant that entered the electric vehicle (EV) market only last year: its EV segment is already on track to turn a profit by the second half of 2025. The company’s parallel expansion into chips, AI models and embodied tech (with its CyberDog and CyberOne humanoid prototypes unveiled in 2021 and 2022, respectively) also reflects a broader pattern in China’s tech ecosystem. Commercial success is often used to help underwrite more ambitious and experimental R&D projects. Similarly, UBTech — China’s first publicly listed humanoid company — relies on high-margin education bots to partially offset R&D costs for its Walker S humanoid robot. The company is slated to be among the initial partners featured in the world’s first bricks-and-mortar humanoid robot dealership launching in Beijing this month (August).
It also helps that China is the world’s factory. Its mature industrial base and extensive supply chains carry a wealth of tacit knowledge and production expertise. Emerging sectors like humanoid robotics and eVTOL share significant technological overlap with China’s established strengths in EVs, drones and industrial automation. As such, they benefit greatly from cross-industry knowledge spillovers, which in turn speeds up the cycle of innovation. Hence, China’s lead in these sectors is perhaps not too surprising.
On a practical level, rapid commercialisation also unlocks another key advantage: access to proprietary, real-world data. As a recent Harvard Business Review report points out, AI differs from traditional technologies in that much of its foundational research is freely available. Google’s 2017 seminal paper “Attention is All You Need”, which introduced the concept of transformer architecture that underpins the generative AI boom, is a case in point. What matters more is access to large troves of high-quality data, in order to fuel a virtuous loop in which better data trains better models, which in turn generate better engagement.
Here, China is doubly advantaged. First, its massive digital footprint and the ubiquity of “super apps” provide an unusually rich stream of high-frequency behavioural data, far surpassing the fragmented digital ecosystems common in the West. Second, while long derided for its “good enough” innovation, Chinese AI models at present offer a compelling price-to-performance proposition with near-parity performance at a fraction of the cost (see chart on previous page). This in turn increases its broad appeal to everyday consumers who prioritise affordability over cutting-edge features.
Limits of central planning
To be clear, significant gaps remain in China’s AI strategy. Chief among them is the state of its semiconductor industry, as we covered last week. Gauging the true progress of China’s advanced chip production will likely remain difficult, as non-transparency works to Beijing’s advantage. Still, history offers a valuable precedent for its ability to mobilise under pressure. Consider the original juguo campaign: the “Two Bombs, One Satellite” programme under Mao.
Despite severe resource constraints, China successfully developed both the atomic and hydrogen bombs, as well as launched its first artificial satellite, all in short order. Interestingly, this era also coincided with sweeping US trade embargoes that lasted until 1972.
In some ways, the more pertinent risk lies in the very state-led model that has brought China this far in the race. Past is not prologue, but again, history provides a cautionary tale with China’s digital economy. For years, the state tolerated and even championed the unchecked expansion of its internet giants. But that permissiveness had clear limits: in 2021, Beijing unleashed a sweeping regulatory campaign that wiped out over US$1 trillion in market value. Jack Ma’s public rebuke of regulators is widely viewed as the tipping point, but to frame these events purely as an example of the capricious hand of the state misses the deeper structural reality. In China’s political economy, every sector ultimately serves the CCP’s broader agenda. Success is never purely commercial, and failure is rarely just market driven. That’s the double-edged nature of centralised control. For now, though, AI remains the sharpest tool in the CCP’s arsenal, a strategic asset too critical to core Party objectives to be disrupted. But should it ever pose a threat by empowering individuals or institutions beyond the state’s reach, it too could be brought to heel abruptly.
Of course, America’s market-led model is not without its faults. Where Beijing prioritises political stability, Washington appears preoccupied with preserving competitive markets. This has given rise to an increasingly assertive antitrust environment that has, in recent years, targeted tech giants like Meta and Google. Specifically, preserving market dynamics is prioritised at the expense of letting individual firms achieve scale. Or as former Federal Trade Commission chair Lina Khan bluntly put it: “We need to choose competition over national champions.”
Neither system gives technology free rein. They just draw the line at different places.
Winning what, exactly?
The same logic extends to evaluating the AI race itself. Who’s “ahead” depends on where you’re looking. America clearly holds a commanding lead in foundational research and compute infrastructure. But just as convincingly, one could argue that this race won’t be decided in the labs of the Massachusetts Institute of Technology or Tsinghua University, but rather in factories, schools and city streets, wherever people and businesses adopt AI to make everyday choices faster, cheaper, smarter.
As the Information Technology and Innovation Foundation (ITIF) notes:
“... innovation is not invention. It is not science. It is not necessarily entrepreneurship. It is bringing to market new products or services at scale”.
And by that definition, China is undeniably in the lead.
The reality, however, is that even if China pulls ahead, it is unlikely that Western nations will embrace Chinese technology wholesale. Whatever foundational breakthroughs that emerge will likely meet the same fate as Huawei — scrutinised and restricted on national security grounds. The likelier scenario, then, is that we will face a future of two separately developing tech ecosystems, one forged along geopolitical fault lines.
For those of us in the Asean bloc, just as we were once courted from both sides by China’s Belt and Road Initiative and the now-defunct Trans-Pacific Partnership, we may yet find ourselves caught between competing digital visions. Or, perhaps, like today where platforms like ChatGPT, Gemini and DeepSeek are used interchangeably depending on specific needs, the better question for us may not be who is winning but what works best, where, for what purpose and at what cost? On that, the market has a way of finding its own answers.
The Absolute Returns Portfolio gained 0.2% for the week ended July 30. The gains lifted total portfolio returns to 30.9% since inception. The top three gainers were ChinaAMC Hang Seng Biotech ETF (+7.7%), Goldman Sachs (+1.8%) and JPMorgan (+1%). The biggest losers were SPDR Gold MiniShares Trust (-3.6%), Alibaba (-3.1%) and Trip.com (-2%). As articulated in our accompanying sidebar, “Selling into confirmation of good news ... cut US equities position by half”, we halved our holdings in US stocks and raised cash to 40.2% of total portfolio value.
The AI Portfolio outperformed both the Total Absolute Returns and Malaysian portfolios for the second straight week, gaining 1.5% and boosting total portfolio returns to 4.1% since inception. The biggest gainers last week were Cadence Design (+13.7%), Intuit (+4%)
and Datadog (+3.4%) while RoboSense (-4.8%), Alibaba (-3.1%) and Workday (-1.4%) were the top losers.
The Malaysian Portfolio, on the other hand, fell 0.6% last week amid continued weakness in the broader market. All stocks in the portfolio ended in the red, with Insas Bhd – Warrants C (-16.7%), Hong Leong Industries (-2.7%) and Malayan Banking (-1.5%) leading the losers. Last week’s loss pared total portfolio returns to 181.4% since inception. Nevertheless, this portfolio is outperforming the benchmark FBM KLCI, which is down 16.7% over the same period, by a long, long way.
Selling into confirmation of good news … cut US equities position by half
There is no longer any doubt as to who won the trade war that US President Donald Trump started with the rest of the world. And yes, contrary to the widely held delusion that there is no winner, the US won. We had previously articulated this in our series of articles on Trumpianomics, and the fact that an “optimal tariff” does exist. And no, there is no voodoo math involved. So, yes, the US will gain from more domestic production and exports, more investments and more jobs, increased productivity and higher wages, especially in the longer term (from reshoring, especially of manufacturing). We will write more on the economic impact on both the US and other nations in the near future, as a continuation of our Trumpianomics series. But for now, let’s turn our attention to the short-term repercussions.
After the dust settles on the trade war, markets will turn to the next pressing topic — can Trump also force the US bond market into submission with lower long-term yields? We think not.
The effective US tariff is likely to end up between 15% and 20%, which, while lower than initially feared (worst-case scenario), is still substantially higher than before Trump’s trade war. To be sure, this cost will be shared among the foreign producers-exporters, US importing companies and US consumers. But inevitably, it will result in higher domestic inflation in the US — and likely negatively affect US domestic consumption, which makes up 70% of gross domestic product (GDP). The most important upcoming data will be the US inflation adjusted consumption. And global growth and demand must surely fall too.
The higher cost of goods will negatively affect US companies in terms of demand and sales, margins and profits. The effect is not yet evident, due in part to front-loading (inventory build-up) ahead of tariffs, but it will materialise — and will dominate the storyboard over the coming months.
We think prevailing higher-for-longer inflation will keep longer-term yields, which are determined by market forces, high. Bringing down the short-term one-year Treasury yield — which more or less reflects market expectations of the US Federal Reserve’s policy rate — does not mean the longer-term yields will fall. As it is, the inflation adjusted yields for the 10-year and 30-year Treasuries are around 2%, slightly more than the historical averages. But bear in mind we have historical high fiscal deficits in most countries, an ageing population and an environment where inflation will likely rise, at least in the near term. Case in point: The recent fall in short-term yields did not bring about a corresponding decline in longer-term yields, which stayed high (see Chart 1). Higher long-term yields (borrowing costs) will dampen economic activities, especially the housing sector, and lower stock valuations (discounted future cash flow).
The tremendous bull run of the US market was justified, first by earnings, then by the momentum created by the promise of artificial intelligence (AI). Huge investments are now expected to translate into exceptional gains in productivity, efficiency and profits.
Consequently, US equity valuations are near all-time highs — see the Shiller cyclically adjusted price-to-earnings (CAPE) ratio in Chart 2. This also means the risk-reward proposition must warrant caution. We don’t know whether it will be a pause, consolidation or a major correction. But we know that the US stock market in 2025 has been driven by momentum — and storytelling — not fundamentals. Therefore, this means even more elevated risks.
This being the case, we cut our exposure to US equities by half across all the stocks in the Absolute Returns Portfolio. We guess you could say we are “selling into the confirmation of good news”. After the disposals, the portfolio now has 19.3% weightage in US equities, 32.1% in China-based stocks, 8.4% in gold exchange-traded fund (ETF) and just over 40% cash. This is an absolute returns portfolio with a 10% target return annually. We have way exceeded this target — currently sitting on 30.9% total returns since inception (March 2024), in less than 17 months — and can therefore afford to now sit out this risk.
Disclaimer: This is a personal portfolio for information purposes only and does not constitute a recommendation or solicitation or expression of views to influence readers to buy/sell stocks, including the particular stocks mentioned herein. It does not take into account an individual investor’s particular financial situation, investment objectives, investment horizon, risk profile and/or risk preference. Our shareholders, directors and employees may have positions in or may be materially interested in any of the stocks. We may also have or have had dealings with or may provide or have provided content services to the companies mentioned in the reports.
