Asia-Pac Japan
Portfolio

July was defined by dispersion rather than regional direction. Hong Kong was the world’s strongest major equity market, with the Hang Seng rising 13.1%, while the Nikkei 225 fell 8.1% and Korea recorded its worst month since the global financial crisis. The divergence reflected a rotation out of the Japanese and Korean AI-semiconductor complex and into offshore Chinese equities, rather than a broad withdrawal from Asia-Pacific risk.

Japan suffered its first negative month in four and a much more violent path than the closing return suggests. The Nikkei declined 12.8% from its 1 July high to the 29 July trough before rebounding 4.0% on the final session. Semiconductor and memory exposures led the correction, with Kioxia Holdings falling 48% over the month. The Bank of Japan held its policy rate at 1.0% by an 8–1 vote, although one member favoured 1.25%, and continued to warn that underlying inflation could remain above its 2% objective.

Japanese rates and currency volatility became the more important macro signal. The 10-year government bond yield reached 2.88%, its highest level in 29 years, as investors confronted mixed policy signals, fiscal expansion and unfunded tax-cut proposals. USD/JPY rose to 163.84 before reported intervention and a US rate check helped drive the pair back to 157.47 at month-end. The episode suggests the effective intervention zone is approximately 162–165, rather than a single exchange-rate threshold.

Korea represented the month’s tail event. The Kospi fell 10.8% on 28 July and triggered another market-wide circuit breaker as Samsung Electronics and SK Hynix declined sharply on concerns about Chinese memory competition and the returns available from AI capital expenditure. Samsung finished July down 21% and SK Hynix lost 35%, before an 18% index rebound on the final Friday. With the two companies now representing approximately half of the Kospi, the benchmark increasingly behaves as a leveraged memory-cycle exposure.

China absorbed much of the capital leaving the regional AI complex. Offshore technology and consumer exposures rallied strongly, while domestic A-shares lagged. The bid was driven primarily by policy and stimulus expectations rather than improving fundamentals, with second-quarter GDP growth slowing to 4.3%. CXMT’s 466% debut surge nevertheless delivered an important competitive signal: investors are increasingly assigning value to China’s domestic semiconductor and AI supply chain while marking down incumbent Japanese and Korean hardware producers.

Australia was the region’s quieter outperformer. The ASX 200 gained 2.0% for a fourth consecutive positive month, led by energy as oil prices strengthened. Inflation eased modestly, and the Reserve Bank did not meet in July after holding the cash rate at 4.35%. Leadership remained concentrated in larger companies, with the ASX 20 gaining approximately 4% while the Small Ordinaries declined 3%.

Currency translation materially affected AUD-reported outcomes. AUD/USD appreciated 1.5%, creating a headwind for Hong Kong and other USD-linked holdings, while AUD/JPY declined 1.6%, cushioning Japanese losses. On an approximate translated basis, the Nikkei’s 8.1% local-currency decline became a 6.6% loss for an AUD investor, while the Hang Seng’s 13.1% gain moderated to around 11.4%.

For the portfolio’s Nexus thesis, July was not a repudiation of AI-infrastructure demand. It was a supply-side re-rating of who captures memory-sector economics. Chinese capacity was interpreted positively for domestic platforms and semiconductor producers but negatively for Japanese and Korean incumbents. The investment implication is to distinguish durable infrastructure demand from competitive margin risk across memory, semiconductor equipment and the regional AI supply chain.

APAC/Japan Energy Transition AI Nexus Q2 2026: China’s Abundance vs. Japan’s Scarcity

The defining contrast across the region is not policy direction, but starting conditions. China is addressing the AI power challenge from a position of energy abundance, while Japan is solving it from structural scarcity. That divergence should shape portfolio positioning across the two markets differently from the approach used in North America or Europe.

China’s advantage is structural, not merely a function of scale. Cheap electricity for data centres reflects both excess generating capacity and low utilisation across parts of the power system. China built generation so rapidly that AI load is now absorbing capacity that was already available. The equipment supply chain reinforces that advantage: domestically produced transformers can reportedly be delivered in approximately 48 weeks, compared with an average of roughly 143 weeks in the United States. That is close to a threefold speed advantage for one of the components most responsible for Western interconnection delays.

The strategic dichotomy is increasingly clear: the United States retains leadership in advanced chips, while China controls much of the renewable energy equipment supply chain required to power those chips at scale. For investors, the implication is that China’s AI infrastructure advantage extends beyond compute hardware into transformers, solar modules, batteries, power electronics and grid equipment.

China’s cheap power story nevertheless carries a significant caveat: curtailment. National wind and solar curtailment reached approximately 5.4% during the first nine months of 2025, above the original 5% policy target and 2.1 percentage points higher year on year. The problem reflects inflexible long term contracts, barriers to inter provincial electricity trading and provincial preferences for local generation over imported power.

This is the same underlying challenge seen in the United States and United Kingdom the grid cannot move power efficiently to where it is needed but it manifests as wasted renewable output rather than stalled interconnection queues. China is responding by connecting data centres more directly to renewable generation. One example is the Ningxia project pairing 500 MW of solar, alongside 1.5 GW of planned wind, with the Zhongwei cloud base and shifting compute workloads toward locations where renewable power is most abundant.

Coal remains the honest answer to what powers Chinese AI today. Data centre electricity supply is currently estimated to be almost 70% coal, compared with roughly 20% renewables and 10% nuclear. Wind and solar are projected to supply around 40% of total Chinese generation by 2030 and overtake coal for the first time, but that remains a medium term rather than immediate transition. Timing therefore matters: near term beneficiaries are likely to be equipment, grid and component suppliers supporting the buildout, while the full decarbonisation benefit arrives later.

Japan is the mirror image: scarcity driven, restart dependent and geographically mismatched. Unlike China’s abundance model, Japan’s path requires faster nuclear restarts alongside renewable deployment because renewables are expected to reach only around 17% by 2030. Much of the country’s large scale renewable and nuclear capacity is located in Hokkaido and Kyushu, far from the Tokyo and Kansai data centre clusters that require the power.

Japan’s phased “Watt Bit Collaboration” framework is designed to address that mismatch through welcome zones, transmission upgrades and, ultimately, greater co location of compute and generation. The practical consequence is that the Japanese opportunity is likely to unfold through nuclear restarts, power equipment, transmission, industrial automation and data centre infrastructure on a longer runway extending into the early 2030s.

Q3 watch list: measurable progress in China’s provincial power trading reforms and their effect on renewable curtailment; the pace of additional Japanese reactor restart approvals following Kashiwazaki Kariwa; and the financing structure of Japan’s ¥5.2 trillion data centre power project. Whether that project is led by public utilities or funded directly by hyperscalers will determine how Japan’s model differs from the US pay your own way framework and China’s state directed approach.

APAC/Japan Volatility Regime — July 2026

July shifted Asia-Pacific into a high-dispersion, event-driven volatility regime. The Hang Seng gained 13.1% while the Nikkei 225 fell 8.1% and the Kospi suffered its worst month since the global financial crisis. This was not conventional regional de-risking: volatility was concentrated in Japanese and Korean semiconductor exposures while capital rotated into offshore Chinese technology and consumer equities.

Japan’s realised volatility materially exceeded the month-end index return. The Nikkei declined 12.8% from its 1 July high to the 29 July trough, then rallied 4.0% in a single session at month-end. That path, combined with a 48% monthly fall in Kioxia Holdings, confirms that crowded AI and semiconductor leadership had become the principal source of portfolio-level convexity.

Rates and currency were equally important volatility channels. The 10-year Japanese government bond yield reached a 29-year high of 2.88%, while USD/JPY traded to 163.84 before reported intervention and a US rate check drove it back to 157.47. With the Bank of Japan holding at 1.0% but retaining a tightening bias, Japanese equity, rates and currency volatility are now tightly linked. The approximate 162–165 intervention zone should remain a key event-risk trigger.

Korea delivered the clearest tail-risk signal. A 10.8% one-day Kospi decline triggered a market-wide circuit breaker, followed by an 18% rebound on the final Friday. Samsung Electronics and SK Hynix now represent roughly half of the benchmark, meaning index protection is increasingly a proxy hedge for a concentrated memory trade rather than diversified Korean equity exposure. Single-name and semiconductor-sector risk therefore require explicit treatment.

Hong Kong presented the opposite skew profile. Strong gains in offshore Chinese equities were driven by policy expectations and domestic semiconductor momentum rather than broad earnings confirmation. This creates upside participation risk for underweight portfolios but also leaves the market vulnerable to abrupt reversals if stimulus delivery disappoints. Defined-risk call spreads are preferable to chasing spot exposure after a double-digit monthly rally.

For the ECATS Vol Carry & Skew framework, broad regional hedges are unlikely to capture the risk efficiently. The preferred implementation is selective: use elevated semiconductor volatility to finance collars and put spreads in Japan and Korea; concentrate event protection around Bank of Japan meetings, JGB yield shocks and the USD/JPY intervention zone; and retain measured upside convexity in China-linked exposures. AUD/JPY and AUD/USD should also be monitored alongside equity hedges because currency moves materially altered July’s AUD-reported returns.