Exuberance and discipline: Finding durable returns in Asia’s crowded market

“History never repeats itself. Man always does.”
Voltaire

In the summer of 2026, you could see young people strolling the streets of Hong Kong wearing black vests emblazoned with a distinctive logo. The product in question? An exchange-traded fund (ETF) tracking SK Hynix, the Korean memory-chip manufacturer.

In June, this ETF, run by asset manager CSOP, overtook the iconic Tracker Fund of Hong Kong as the market’s biggest.¹ The Tracker Fund has a history dating back a quarter of a century and offers exposure to a range of blue-chip stocks listed on the Hang Seng Index. By contrast, the CSOP ETF follows only Hynix and uses leverage to amplify returns – a strategy that can also magnify losses on the way down.

From a diversified portfolio to a high-stakes bet on a single technology name: this story mirrors the broader trajectory of Asian markets over the last 12 months. The enthusiasm for artificial intelligence (AI) has meant returns are clustered around a narrow group of companies deemed to be the beneficiaries of higher spending on data centres and AI supply-chain equipment, with chipmakers like SK Hynix leading the way. 

We and our readers no doubt agree that AI is one of the biggest – if not the biggest – technology shifts of our time. We find increasing AI applications (with clear guardrails) in parts of our investment process. From consumers to enterprises, use cases continue to arise and experts agree that we still are in the early innings. 

Genuine value is likely being created amid the surge in Big Tech capital expenditure (capex). However, as share prices rise and more capital chases the theme, significant risks are emerging. AI is a revolutionary technological leap, but it is not immune to the capital cycle – or herd behaviour characterised by greed and fear. We remain mindful of Voltaire’s wisdom quoted above, and this is reflected in the sizing of positions in the strategy. 

Consider how the AI boom has led to extreme market concentration. South Korea and Taiwan, home to semiconductor champions and other AI-connected manufacturers, now account for 23.4% and 26.9%, respectively, of the benchmark MSCI Asia Pacific ex Japan Index. Their weighting is higher than China’s (18.8%) and India’s (10.9%), despite being far smaller economies.² Returns have been distorted: over the year to date, the index has risen ~24%, but strip out Korea and Taiwan and it would have fallen by ~3%.³ 

As investors in the major chipmakers, our Asia strategy has benefited from their rise over the past 12 months. Nevertheless, the narrowing of markets makes us cautious. 

Interest rates may go up, and debt funding may become much more expensive. Equity investors may question potential return on data center capex. The capex cycle may turn. Or it may just carry on. We don’t profess to know, although as a team we discuss and debate these topics to death. 

But if and when conditions change, discipline and diversification will matter. What’s more, quality companies in many “non-AI” areas of the market are currently under-owned and under-analysed, making this a good time for active, bottom-up stock pickers to look past the dominant narrative and find attractive opportunities with good absolute returns. 

In this letter, we will highlight recent drivers of performance, set out the thinking behind our current positioning, and explore other holdings in the portfolio, such as a collection of leading companies in China’s industrial and advanced manufacturing sectors. While they are attracting less attention than the region’s chipmakers, these innovative businesses are wielding increasing clout on the global stage.

Unsurprisingly, over the past 12 months the list of positive contributors to performance of the strategy is headed by SK Hynix, Samsung Electronics and Taiwan Semiconductor Manufacturing (TSMC). Combined, these three names and their holding companies now account for nearly a third of the MSCI Asia Pacific ex Japan Index, and they have contributed an outsized proportion of the overall market return in 2026. 

TSMC remains our largest portfolio holding, as it has been for years now; its unrivalled position at the most advanced process nodes makes it a natural beneficiary of the expected development of AI “productisation” into end-devices, which should provide a long-term structural tailwind even if data-centre spending normalises. 

By contrast, we have been trimming our exposure to both SK Hynix and Samsung; both Korean firms have shifted capacity toward high-bandwidth memory (HBM) chips for AI applications, which has caused a price spike in conventional DRAM and NAND memory and contributed to a short-term earnings uplift. 

The three memory makers – Hynix, Samsung and Micron in the US – are expected to earn roughly US$675bn in net profit by 2027. This far exceeds the amount of profit (we aren’t talking about cash flows!) that their “big four” hyper-scaler customers are expected to earn combined – or even the combined earnings of the biggest compute chip makers (Nvidia, TSMC, Intel, AMD and Broadcom). Margins have never been higher. But this could change if supply comes online faster or if memory demand softens as prices for consumer electronics rise. 

Controlling our positioning in some of the highest-growth names has meant we have lagged the benchmark of late. While this is uncomfortable, we have long believed in the importance of questioning market assumptions. 

This approach is what led us to invest in other, lesser-known Taiwanese technology names before the AI boom. An example is Realtek, a designer of connectivity chips. Previously unloved because of its links to mature personal computer (PC) and other consumer electronics markets, it has since re-rated as the AI infrastructure buildout fuels demand for connectivity components. We have reduced our position as its share price has risen.

Another example is chip designer MediaTek. We have been shareholders in the company for many years and have seen it navigate various cycles. Under the leadership of CEO Rick Tsai (formerly of TSMC), MediaTek has smartly repositioned itself, adapting its cost structure, building closer relationships with customers and boosting research and development (R&D) beyond smartphones in order to diversify its business. We maintained our conviction in 2022-2023 when the share price fell amid weaker spending on smartphones.

More recently, however, we have been reducing our stake as shares have re-rated thanks to the company’s progress in application-specific integrated circuit (ASIC), or customised AI chips. While the ASIC business is driving faster revenue growth, and MediaTek’s core smartphone business should capture a share of productisation going forward, AI excitement is increasingly reflected in the company’s valuation. 

With all the focus on hardware companies in Taiwan and elsewhere, we believe investors may be underappreciating where the technology can create truly sustainable value. In this respect, our long-standing holding Tencent represents a more contrarian way to participate in the generosity of AI capex. 

Tencent’s competitive advantage lies in its sticky ecosystem, which provides a formidable moat and is deeply embedded in everyday life in China: WeChat’s social media platforms, integrated payments, content and gaming. The market has been disappointed by Tencent’s perceived delay in embracing large language models (LLMs), but recent management meetings left us comfortable with its measured approach. Tencent has hired talent in a bid to create a state-of-the-art LLM while simultaneously deploying AI to monetise its existing products. This has helped it improve advertising targeting and efficiency, and boost both production and engagement across its gaming portfolio. 

Over the longer term, Tencent sees great potential in integrating AI agents into WeChat: the company has started testing an AI assistant that can interact with WeChat’s suite of “mini-programmes”. It’s worth remembering the company has a track record of coming from behind to compete in new technologies such as e-payments, and we would back the management team to execute a repeat when it comes to AI.

At a time when a handful of AI-linked names have dominated index returns, we have been encouraged to see solid contributions from holdings in other sectors.

A good example is Midea Group, a Chinese home appliance manufacturer. In a market fixated on data centres, the world of air-conditioners and kitchen gadgets might seem mundane, but Midea stands out for its technical expertise, marketing and branding acumen, and a management team with a strong track record. 

Midea is also building its business-to-business operations. Having closely studied the development of leading global peers such as Siemens, the company is becoming an increasingly sophisticated provider of automation and energy-management systems, as well as a manufacturer of advanced industrial components such as compressors, motors and electric vehicle (EV) parts. This means it is aligned with broader structural tailwinds such as import substitution in China (see “Investing in China’s industrial innovation story”, below). 

The result is a highly cash-generative business: Midea earns a return on equity (ROE) of 20% and consistently converts over 100% of earnings into free cash flow, operating with negative working capital and minimal fixed asset requirements. Midea’s management is mindful of shareholder value and distributed RMB44bn (US$6bn) to shareholders in 2025 alone through dividends and buybacks.⁴

Another solid compounder in the portfolio is Singapore’s OCBC. The bank has benefited from higher, more normalised interest rates in recent years, which have structurally boosted its net interest margins and ROE. At the same time, OCBC is reaping the benefits of its two decade-long investment in building a wealth management franchise, which is driving fee income. Like Midea, the bank is using its strong cash generation to directly reward investors: it has launched a S$2.5bn (US$1.9bn) capital return plan, combining special dividends and share buybacks over a two-year period.

Midea and OCBC are exactly the kinds of well-run, resilient, shareholder-orientated businesses we aim to own across our portfolios. They might not post explosive growth, but they steadily compound wealth, providing diversification and balance at a time when cyclical trends are contributing to market distortions.

Looking at the detractors from performance over the last 12 months is a reminder that there have been significant developments beyond the world of algorithms and data centres. Most notably, the war in Iran has interrupted the passage of oil and liquefied natural gas (LNG) through the Strait of Hormuz, a key conduit for global energy flows. 

Net energy-importing economies in South and Southeast Asia have been particularly hard hit. As the world’s third-largest buyer of crude, India has faced rising inflation, foreign capital flight and currency pressure. Our holdings ICICI Bank, HDFC Bank and Kotak Mahindra Bank have been de-rated as the market frets about the potential knock-on impact on the financial sector.

These banks’ recent earnings showed few signs of trouble: non-performing loans (NPL) remain well under control and continued to decline through the first quarter of 2026. We are monitoring for signs of a deterioration in asset quality or a slowdown in credit demand, but we believe all three banks should remain resilient. They entered the crisis from a position of strength, thanks to their conservative management approach and well-managed loan books, and could gain market share as weaker rivals come under pressure.

HDFC Bank has faced an additional, idiosyncratic challenge. Its shares weakened significantly in March following the sudden resignation of chairman Atanu Chakraborty, who cited practices “not in congruence with [his] personal values and ethics”. 

Our subsequent research into the circumstances has been reassuring. The episode appears to have reflected friction at the board level over the boundaries of a non-executive role, along with other strategic disagreements, rather than deeper operational issues or a breakdown in core governance. Some uncertainty inevitably remains, but we welcomed the appointment as interim chairman of Keki Mistry, who has a long record of stewardship in Indian financial services.

Overall, we remain confident in the quality of HDFC's franchise. Its deposit-gathering capability and underwriting culture continue to stand out among regional peers, and its asset-quality metrics remain among the strongest in the sector. Valuations have retreated to levels that, on a price-to-book basis, compare favourably even with the lows reached during the GFC and the Covid-19 pandemic. Given the potential for the bank to compound book value at mid-teens rates or above, with further scope for ROE and return on assets (ROA) to improve, we see this as an attractive opportunity and have been adding to the position.

The most rewarding opportunities are rarely found where attention is concentrated. While markets have been fixated on AI hardware, a transformation of comparable scale is underway in China’s sophisticated industrial and advanced manufacturing sectors. 

In 2000, China’s share of R&D spending was just 4% of the global total; that has risen to more than 27% (see Figure 1). This reflects both a sustained national commitment to moving up the industrial value chain and bottom-up innovation among a set of entrepreneurial and – increasingly – globally relevant Chinese companies. These include our holdings Shenzhen Inovance, an automation specialist; Contemporary Amperex Technology (CATL), the dominant global EV battery maker; and XCMG, China's leading construction equipment manufacturer.

While they have benefited from regulatory tailwinds and government support to differing degrees, these businesses have distinguished themselves through their own sustained investment in engineering capability and a near-obsessive attention to their customers’ practical needs. Just as important from our perspective, they feature management teams whose interests are well aligned with our own.

At Inovance, the focus on people is reflected in the company’s Chinese name, Huichuan, which roughly translates as “rivers converging”. The founders, a team who formerly worked at tech giant Huawei, chose the name to convey their desire to bring together China’s best young engineering talent. To this day, each of Inovance’s production lines is named after a different watercourse, signalling its enduring commitment to this goal.

Figure 1: Top gainers and losers of global R&D share, 2000-2024

Source: World Intellectual Property Organisation, Eurostat, OECD, RICYT, and UNESCO UIS, December 2025.

Starting in 2003 as a small electrical inverter maker, Inovance won market share in elevator inverters before diversifying into other areas (this strategy proved wise following the collapse of China’s property sector in 2021, which severely hit demand for lift components). Thanks to disciplined investment in R&D, Inovance has since grown into the dominant force in China's industrial automation market, where it is increasingly overtaking global heavyweights like Siemens, ABB and Schneider Electric. For example, it now commands a 28% domestic market share for servo systems (high-precision motion control solutions), nearly three times that of its closest global competitor. 

The journey has not always been smooth. Management admits to over-expanding early on – such as in 2012, when they anticipated a demand shift from Japan that never materialised – but the team has since shown greater discipline in capital allocation. On the whole, Inovance has been successful in targeting adjacencies and mastering technologies in new segments where it can compete on both quality and cost.

Over the past five years, Inovance’s revenue has compounded  at an average annual rate of over 31%, while ROE averaged 21% over the same period.⁵ Its valuation is not cheap, at nearly 25 times forward earnings, but we believe this premium is justified by its superior execution, strong net cash balance sheet and the long-term growth potential in robotics and other areas, such as digital energy management.

As with Inovance, the investment case for CATL starts with the quality of its management. The company is led by CEO Robin Zeng, who previously founded Amperex Technology, a supplier of batteries for consumer electronics. CATL was spun out of Amperex’s EV division in 2012. 

At a meeting last year, we asked Zeng about his philosophy, which he distilled into a set of core principles: common sense, vision, insight, courage, and an appreciation of competitors. This mindset is evident across the company and is part of the reason CATL has risen into the global leader in EV batteries over the past decade, with over 40% share domestically and an increasing footprint in Europe. CATL also has a growing business in energy storage solutions (ESS).

A central part of this story is CATL’s technological edge, which it maintains through a substantial annual R&D budget of around RMB20bn (US$2.9bn), and an engineering workforce that numbers more than 20,000. A recent visit to its plant in Ningde gave us an insight into its manufacturing capabilities, which run on custom-made and highly automated equipment. This in-house capability, combined with CATL’s scale and supply chain control, translate into a clear cost advantage over peers, and consistently superior margins.

While CATL has a significant technical lead for the time being, the cyclical nature of the battery industry can prove challenging. The industry has historically been affected by price swings in commodities. However, CATL has performed well during previous industry downturns. For example, when lithium prices plummeted in 2023-2024 – driving sharp declines in battery prices and compressing margins for many of its competitors – CATL's unit profit remained stable, reflecting its cost control and premium product mix. 

This track record bolsters our conviction in CATL’s long-term resilience, as does its continued success in building share internationally. Overseas revenues grew 17.5% to more than RMB129bn (US$18bn) in 2025, and now account for more than 30% of total earnings.⁶ Geopolitical concerns over China’s involvement in supply chains may limit its opportunities in the US and some other markets, but elsewhere such concerns are likely to be outweighed by the cost advantages CATL offers. The company is expanding its manufacturing footprint abroad, with new battery-making facilities in Hungary and Spain set to bring CATL closer to its customers in Europe. 

Like CATL and Inovance, XCMG is increasingly winning share on the global stage thanks to its engineering expertise and cost-competitiveness in construction equipment. It has become a leader in equipment electrification and benefits from direct access to China’s deep supply chain in batteries and other components. 

In recent years, it has moved into high-end mining equipment, penetrating a lucrative market once dominated by Caterpillar and Komatsu. Its progress is underlined by recent deals including a major strategic partnership with another of our holdings, Australian mining group BHP, to develop next-generation solutions across its global operations – an endorsement of its technical capability.

XCMG is a state-owned enterprise (SOE). We are usually sceptical of SOEs, as their structure means capital allocation can be driven by political priorities rather than shareholder returns. However, XCMG has undergone significant governance reforms in recent years, which have improved execution, operational efficiency and risk control. The company has also introduced an Employee Stock Ownership Plan (ESOP) to ensure management incentives are consistent with the interests of minority shareholders. 

Meanwhile, earnings quality has improved; whereas reported profits previously failed to translate into cash, the ratio of free cash flow (FCF) to net income has now reached 152%. And the company has committed to returning capital to shareholders, with RMB300-600m (US$40-80m) in buybacks planned for 2026.⁷

As with all three companies, geopolitics represents the key risk: trade tariffs and protectionist politics could prevent XCMG from continuing to expand into Europe and Latin America. On the other hand, the same disruptions may also strengthen the long-term investment case. Higher energy prices make battery-powered vehicles more cost-competitive against diesel alternatives and could accelerate demand for energy-efficient industrial automation. 

Looking ahead, we believe these Chinese companies are representative of a broader opportunity set among quality businesses in Asia, as the run-up in AI-related stocks means strong companies in other areas become increasingly compelling from a valuation perspective.

This is reflected in data at the portfolio level. Earnings expectations across the holdings in our Asia strategy are broadly comparable with the index, and balance sheets continue to strengthen. Profitability remains robust, with ROE matching the benchmark and financial holdings delivering higher ROA. Yet the portfolio’s forward price-to-earnings (P/E) multiple has fallen in line with the index, while it trades on a substantially lower price-to-free-cash-flow multiple.⁸

This is unusual – quality strategies typically trade at a significant valuation premium to the benchmark – and underpins our conviction that the portfolio remains attractively valued relative to the quality of the businesses we own.

We thank you as always for your continued support and feedback.

 

Footnotes

¹ “CSOP’s SK Hynix leveraged product overtakes Tracker Fund as Hong Kong’s largest ETF”, The Standard, June 23, 2026.

² FactSet, as at June 30, 2026.

³ FactSet, as at June 30, 2026.

⁴ Midea, as of April 2026.

⁵ FactSet, as at June 30, 2026.

⁶ CATL, as of March 2026.

⁷ XCMG, as of April 2026.

⁸ FactSet, as at June 30, 2026. 

Investment involves risk. Reference to specific securities (if any) is included for the purpose of illustration only and should not be construed as a recommendation to buy or sell the same. All securities mentioned herein may or may not form part of the holdings of First Sentier Group’s portfolios at a certain point in time, and the holdings may change over time.