This essay first appeared in The Diplomat.

For decades, South Korean strategy rested on a stable hierarchy within the international order. The alliance with the United States provided security; Seoul then managed relations with Japan, Europe, China, and regional institutions around that anchor. The alliance remains indispensable, but Seoul can no longer assume that this foundation will stay stable.

The evidence is visible in the way Washington now deals with allies. In 2025, the United States initially announced sharply differentiated “reciprocal” tariffs on South Korea, Japan, and the European Union. Negotiations later produced a 15 percent baseline framework for all three, but the process mattered as much as the final rate. Trade access, industrial investment, defense procurement, and alliance politics were increasingly handled as one bargaining package. Allies were not exempt from economic pressure because they were allies.

At the same time, Europe and Asia have been building connections that would have seemed highly ambitious a decade ago. The European Union signed security and defense partnerships with Japan and South Korea in November 2024. This year, the EU and Japan launched a defense-industry dialogue, while the EU and South Korea signed a digital trade agreement and began implementing cooperation on maritime security, cyber and hybrid threats, information manipulation, space, and the defense industry. The strategic map is becoming more networked. 

The network, however, remains uneven. Japan already has a mature economic partnership with the EU and rapidly expanding defense-industry ties. EU-South Korea relations are deepening, but they still lack the density and regular strategic consultation found in EU-Japan ties. A trilateral format would therefore do more than add another meeting: it would reduce asymmetry among three partners whose capabilities are increasingly complementary.

Sign up for APARC newsletters to receive our scholars' commentary and analysis >

South Korea, Japan, and the EU now face four converging challenges and pressures.

The first is demographic. South Korea is aging faster than any other OECD society; Japan’s working-age population has been shrinking for decades; and the EU’s fertility rate reached a new low in 2024. Demography is not a social-policy sidebar. It affects defense recruitment, industrial capacity, fiscal room, technological adoption, and the ability to sustain long-term commitments. Each actor is experimenting with immigration, automation, workforce policy, and welfare reform, but the security implications remain largely compartmentalized.

The second is concentrated economic dependence. None of the three can or should decouple from China. Yet South Korea’s experience after the THAAD deployment, Europe’s debates over economic coercion, and China’s export controls on gallium, germanium, and other strategic inputs have shown that interdependence can be converted into leverage. De-risking therefore requires more than national stockpiles. It requires shared risk maps, compatible certification, co-investment in alternative suppliers, and advance consultation before export controls or industrial subsidies create collateral damage among partners.

The third challenge is U.S. volatility. The United States is still the only actor capable of providing extended nuclear deterrence to South Korea and Japan, and NATO remains central to European defense. But dependence on U.S. power now coexists with uncertainty about U.S. policy. The classic alliance dilemmas of abandonment and entrapment are no longer opposite ends of a spectrum. Today, U.S. allies fear both being left out and being drawn into bargains or contingencies they did not shape.

The fourth pressure is the collapse of the old geographic separation between European and Asian security. North Korean munitions, missiles, and troops have supported Russia’s war against Ukraine. Moscow, in turn, has provided Pyongyang with political cover, economic support, and the prospect of military know-how. What happens on a European battlefield is changing the military balance on the Korean Peninsula. A regional response to a cross-regional threat is structurally inadequate.

This is the case for an EU-Japan-South Korea strategic dialogue. This framework should not be described as strategic autonomy, equidistance, or a hedge against the United States. A better term would be institutionalized hedging: risk diversification through standing, rule-based consultation among allies and partners that remain anchored in the broader U.S.-led system.

The proposal is more feasible than ever before. Japan-South Korea relations have improved significantly in recent years. The January 2026 summit in Nara and the reciprocal summit in South Korea in May demonstrated continuity across leadership changes in both countries since June 2025. The EU already has free-trade, digital, green, and security frameworks with both countries. The task is not to invent three new bilateral relationships; it is to connect existing ones.

The dialogue should begin modestly, through a Track 1.5 process involving officials, experts, and industry. Its early agenda must be concrete and specific: a joint critical minerals and supply chain risk assessment; consultation on export controls and investment screening; cooperation on AI, cyber resilience, and digital standards; exchanges on demographic and defense workforce adaptation; and shared monitoring of North Korea-Russia military cooperation. Successful projects could then be elevated to ministerial working groups.

Two design choices are essential. First, the initiative must remain function-driven. It is not NATO, not the Quad, and not an anti-China coalition. Its purpose is to reduce vulnerability without demanding economic separation. Second, it should be complementary to relations with Washington and transparent about that purpose. Horizontal networks strengthen alliances when they help allies absorb shocks, coordinate positions, and arrive at consultations with greater capacity.

For South Korea, the need is particularly acute. Japan already has the G7, a mature economic partnership with the EU, and expanding European defense ties. Seoul’s international weight has grown faster than its institutional depth. An EU-Japan-South Korea dialogue would help close that gap.

Horizontalizing alliances is not a vote of no confidence in the United States. It is insurance for the moments when the United States wavers, overreaches, or changes course. The postwar hierarchy cannot simply be preserved by demanding more reassurance from Washington. Its resilience will depend on whether allies can build connective tissue among themselves before the next crisis forces them to improvise.

Across the world, populations are aging rapidly as people live longer and fertility rates continue to decline. Asia is at the vanguard of this demographic shift. The number of older adults (aged 60 and above) in the region is projected to triple between 2010 and 2050, reaching nearly 1.3 billion people. As Asian economies face this “silver wave,” helping older adults live safely and independently at home – a concept known as aging in place – has become a policy imperative.

At a recent webinar held during Stanford Health AI Week, the Asia Health Policy Program (AHPP) at Shorenstein APARC brought together experts from China, Singapore, and South Korea to share insights into the potential of health AI to allow older adults to enjoy healthy aging and avoid or postpone institutionalization. 

Moderated by Stanford health economist Karen Eggleston, the director of AHPP, the webinar featured Hongsoo Kim, a professor of health policy and aging at Seoul National University’s Graduate School of Public Health and director of its Artificial Intelligence Institute’s Center for AI in Health and Care; Xiaochen Ma, an assistant professor of health economics at Peking University’s China Center for Health Development Studies; and Tien Yin Wong, a physician-scientist-innovator and the senior vice-chancellor of Tsinghua Medicine and vice-provost of Tsinghua University, who has also worked and held senior leadership roles in Singapore and Australia as a practicing retinal specialist with a research portfolio on retinal diseases, ocular imaging, AI, and digital technology.

Get APARC event invitations and guest speakers' insights delivered to your inbox >

Here are six lessons from the front lines of Asia’s efforts to integrate AI into elderly health care and advance aging in place:

1. Adopt a Whole Systems Approach

In South Korea, the world's fastest-ageing society, automated systems like "CLOVA CareCall" – an AI-powered well-being dialer – conduct natural-sounding check-ins with solo-dwelling seniors, boasting a 96% response rate. Yet, Professor Kim emphasizes that checking in with people in need of health care is only half the battle.

If an AI flags an isolated senior at risk of depression, cognitive decline, or a physical abnormality, but the local community lacks the social workers or clinical pathways to intervene, then the health care system has failed.

“The question is not only whether AI can detect something, but how a health and care system acts on it,” she says. “Detection by itself changes nothing. A warning that no one follows up on helps no one. So the gap I care about is not the model’s cleverness itself. It is whether the system delivers.”

2. Solve the Entire "Care Cascade"

In rural China, traditional diabetic screening rates hover below 33%, leaving millions at risk of Diabetic Retinopathy (DR), a leading cause of blindness. Professor Ma shared how deploying an AI screening model successfully pushed screening rates past 85%.

The research team, however, discovered a glaring bottleneck: only 21% of high-risk patients actually followed up to receive sight-saving treatments. To fill this gap, Ma’s team designed an "AI Plus” model (v2.0) that integrates immediate, local-language counseling at the point of screening. To keep seniors healthy at home, AI solutions must address the entire clinical journey, from initial scan to final treatment.

“Many of the AI tools have been focused on diagnosis accuracy or validation rather than going downstream to the entire cascade of whether improved screening will translate into improved referral and the ultimate health outcomes,” says Ma.

3. Align with Local Workflows and Incentives

AI and other technology solutions for health often fail because they expect overworked care workers to adopt entirely new habits. Professor Ma noted that digital health interventions in rural China succeeded only when they integrated seamlessly into existing daily routines.

Instead of forcing clinicians to use complex new software, successful pilots utilized WeChat, the ubiquitous messaging app already open on every phone. Furthermore, the technology must align with the financial and professional incentives of frontline health workers. If an AI tool increases their administrative burden without simplifying their day or boosting their clinical efficiency, then it will remain unused.

4. Design Human-Centered AI for Health Equity

Professor Wong highlighted the ethical risk that AI tools will worsen, rather than reduce, health care disparities. This challenge is driven by the dynamics of “Inverse Care Law,” where AI disproportionately benefits the already advantaged, and the “Recursive Care Law,” where this inequality becomes a self-reinforcing cycle embedded in the system.

Because younger, more tech-savvy individuals generate more health data, AI models become better at serving them than the intended users of aging-in-place technologies. This creates a vicious cycle where the very tools designed to support aging populations end up marginalizing them. Governments must devise policies to mandate fair data coverage and usability, ensuring that AI serves society's most vulnerable members equitably, Wong stated.

Professor Kim noted that her team found that only about 38% of community care agencies in Korea have adopted AI and that the adoption rate varied sharply by region. In fact, districts with the greatest need may have the least access to these powerful tools. This challenge is a fundamental design gap rather than a technology gap, Professor Kim argues. To be genuinely equitable, a system must be built from the start to actively track who is missing and automatically route support back to them. This requires two  human-centered design key principles:

I. Universal by Default: The hardest-to-reach should not have to be the most persistent in navigating the technology.

II. Connected Across Sectors: Long-term care, social care, and health care must act as one integrated system rather than disconnected silos, each of which sees only part of the person’s needs.

5. Augment, Do Not Replace, the Human Touch

The panelists rejected the trope of robots replacing human caregivers. Instead, they view AI as an essential force multiplier for an overstretched workforce.

Whether it is South Korea’s deployment of 12,000 AI companion robots to combat senior isolation, or automated triage tools in clinics, the goal should be to offload administrative and routine tasks. This frees up human social workers and clinicians to do what they do best: deliver hands-on, empathetic care.

6. Value Real-World Outcomes Over Technical Novelty

Healthcare systems should prioritize rigorous, real-world case studies that prove actual clinical value, such as reduced mortality, lower rates of blindness, or fewer nursing home admissions, rather than celebrating high validation benchmarks in a laboratory.

To build robust future health AI systems, the experts concluded, the academic and tech sectors must also courageously publish and analyze their failed trials to understand what truly works in the chaotic reality of home-based care.

While AI holds immense promise for helping people grow old at home, “age tech” alone cannot solve the elder care crisis, the panelists agreed.

In the Media

Dr. Eggleston joined Jeffrey Snyder, host of the Broadcast Retirement Network, to discuss highlights and takeaways from the webinar discussion. Watch below:

Semiconductors: The Limits of Containment?
The conference returned repeatedly to America’s use of export controls—and whether they are working. The verdict was nuanced. Philip Wong, the Willard R. and Inez Kerr Bell Professor of Electrical Engineering at Stanford, argued that in the semiconductor space the strategy has plainly backfired. By cutting China’s firms off from American chip-making equipment and advanced logic chips, the controls created a large captive domestic market for China’s equipment suppliers who previously had no customers. “It basically enabled the indigenous supply chains to have a wonderful set of customers within China,” Wong said, “and so they were able to climb up the learning curve really quickly, much more quickly than before.” Other speakers suggested that may be a tolerable cost as long as export controls allow the US to reach certain frontier capabilities first—as has been the case with Anthropic’s Mythos model.

Wong pushed back on both alarmism and dismissiveness about China's broader technological rise. China, he argued, has genuine world-class talent and infrastructure across multiple sectors—a peer competitor, not a pretender. "If you are among the best athletes, sometimes you win, sometimes other people win. That happens all the time." To treat any given Chinese breakthrough as proof of American collapse, or to wave it away as a fluke, both miss the point: China is, in his words, "a bona fide good athlete."

Wong’s recommended alternative to export controls was direct: rather than trying to slow a competitor, the United States should focus on “how do we make ourselves run faster.” That sentiment echoed throughout the day, particularly after he noted that the National Science Board had recently been dismissed and that American R&D funding continues to be primarily focused on defense-oriented research. 

Biotech: From Follower to Leading Force
Physician-scientist Chenjian Li, a research fellow at the Hoover Institution, offered striking data on China’s advancement in biotech. In the active pharmaceutical ingredients that form the basis of medicines taken by hundreds of millions of Americans daily, China has achieved near-total global dominance—some categories are 100% Chinese-sourced. “Medicine, be it advanced experimental drugs, high-end prescriptions or just daily over-the-counter pills, are actually much more impactful than weapons of mass destruction,” Li said, “because they affect 80% of the United States and global population.”

At the cutting edge of drug discovery, the picture is more nuanced but equally notable. Chinese biotech startups are increasingly producing competitive "me-too, me-better" drugs that improve on existing treatments, and pushing into "first-in-class" drugs—the crown jewels of pharmaceutical innovation. Major multinational corporations (MNCs) are paying billions to acquire them. The fact that Pfizer, Merck, and Eli Lilly are spending at this scale, Li argued, says something important: “The MNCs buy those new therapeutic assets because they are solid and unique, and because the MNCs don’t think that they can be as fast and as good in those lines.”

Economist Ruixue Jia of UC San Diego connected this pharmaceutical surge to China’s education system, which has spent decades steering enormous numbers of students toward engineering and STEM fields, including biology and life sciences. The founder of one of last year’s biggest biotech deals—a $5.6 billion transaction—fit a pattern Jia’s research keeps finding: educated in China, PhD in Canada, postdoc in the United States, returned to China in 2008. “It’s not just a success story of Chinese education,” she noted. “It’s also a success story of North American education.”

Fragmentation and the AI Race
A central tension runs through the broader debate: what do the world’s two largest economies actually gain or lose from their escalating technological confrontation?

Tsinghua economist Hong Ma argued that, measured by its own goals, the American trade war has largely failed. US import dependence on Chinese value-added has remained roughly constant despite years of tariffs, as goods simply reroute through third countries. Beyond tariffs, he warned of a longer-term cost: fragmentation into two separate innovation ecosystems, neither large enough to fully benefit from the other. The US would lose access to the Chinese market, Chinese engineering feedback, and the scale that sustains rapid innovation. “On both sides,” he said, “this is not the optimal equilibrium.”

Panelists pointed to China’s open-weight AI models as evidence of a different kind of competition playing out below the frontier. China’s models from Alibaba, Moonshot AI, and others are being used across the globe—often simply because they are cheaper and good enough for most applications. In this way, big US labs may be ahead on raw benchmarks, but that advantage does not automatically translate into leading global adoption.

Impressive Achievements, Costly Failures
Another useful synthesis came from Scott Kennedy of the Center for Strategic and International Studies. Kennedy’s “bumpy success” framework holds that China’s innovation trajectory is clearly positive—it now ranks tenth globally on the Innovation Index, ahead of Japan in the Asia-Pacific—but deeply uneven across sectors. He described China as a “slow tech dragon”: vast misallocation of resources produces genuine breakthroughs alongside enormous waste, and that waste is a real drag on the broader economy. Commercial aviation was one such example—despite being a signature priority for China’s leadership and the single largest recipient of state investment, the result is, in Kennedy’s words, “an American plane with Chinese paint.” The lesson is not that China cannot innovate, but that state-directed industrial policy produces highly variable results.

That framework—impressive achievements, structural waste, uneven outcomes—runs as a quiet undercurrent through the broader debate. Structural challenges remain: a domestic market that cannot yet absorb the premium prices of cutting-edge drugs; an education system optimized for solving known problems rather than identifying unknown ones; and an economy in which the benefits of technological investment are not yet reaching ordinary households.

The picture that emerges resists easy predictions, but carries a clear message: the old frameworks—China as technological follower, export controls as sufficient means to maintain America’s remaining technological edge, global supply chains as something susceptible to political redirection—often no longer fit the evidence. The task now is building better ones.
 

Discover more from the 2026 SCCEI China Conference.