Access the full Bloodstone Capital Research platform — AI-powered intelligence, portfolio tracking, real-time market data and more.

Enquire →
Asian Equities5 November 2026 · 2,175 words · 10 min read

Asia briefing — 2026-11-05

japankoreataiwansemiconductorsequitiesnikkeitopixkospitaiexsamsunghyperscalersnovember-2026

Japan, Korea and Taiwan all fell on Thursday, and the easy reading is that Asia sold off on the US midterm result. The sessions' own internals do not support treating them as one event. In Tokyo, 161 of the Nikkei's 225 constituents rose while the index lost 1.8%. In Taipei, 872 listed shares fell. Those are entirely different markets, and the distinction between them is the most useful thing the day produced.

TL;DR

  • The Nikkei fell 1.8% while the TOPIX rose 1.2%, with 161 of 225 Nikkei constituents advancing and 25 of 33 Tokyo industry groups higher.
  • Three heavily weighted Japanese technology names did almost all the damage. Advantest fell 14.7%, Tokyo Electron 9.4% and SoftBank Group 6.9%.
  • Korea was a genuine market-level decline, with the Kospi down 5.28% to 5,383.80 and a sell-side sidecar halting program trading for five minutes.
  • Taiwan's selling was broad, not concentrated. The Taiex closed down 3.20% at 41,574.72, with 872 declining shares against 170 advancing and 69 unchanged.
  • Two catalysts arrived within 24 hours — disappointing Qualcomm and Arm results, and post-election expectations of tighter scrutiny of data-centre construction. No source apportions them.
  • The large cloud operators rallied on the same news, with Microsoft closing up 6.7%, Amazon 6.1% and Alphabet 3.2% in New York on Wednesday.

Japan — The Index Misread

Japan produced the session's clearest lesson in benchmark construction. The Nikkei 225 closed 1.8% lower while the TOPIX gained 1.2%, with 25 of 33 Tokyo Stock Exchange industry groups advancing. Transport equipment rose 11.8%, real estate 7.2% and retail 5.1%. Only shipping, down 6.9% on expectations that freight rates fall if the Middle East conflict eases, joined technology on the downside.

The explanation is price weighting. The Nikkei divides the sum of its constituents' adjusted prices by a divisor, so a high-priced share carries influence out of all proportion to the size of the company behind it. Going into Thursday, Advantest carried an index weight of roughly 12%, Fast Retailing 10.1%, Tokyo Electron 7.9% and SoftBank Group 5.6%. The TOPIX, weighted by free-float market capitalisation, distributes influence very differently.

That is sufficient to explain the divergence. Advantest's 14.7% fall, Tokyo Electron's 9.4% and SoftBank's 6.9% came from three of the index's most heavily weighted members, and between them they outweighed everything the 161 advancing constituents contributed. Fast Retailing's 4.7% gain, from a weight of comparable size to Advantest's, was not enough to offset them. The same session, measured by capitalisation rather than by share price, was a rising market.

Toyota illustrates the mechanism from the other side. It gained 16.7% after reporting a 34% rise in quarterly operating profit and raising its annual forecast, and it is Japan's largest listed company — but at its share price it moves the Nikkei far less than it moves the TOPIX. That is also a reminder that Thursday's advancing stocks were not simply beneficiaries of cheaper oil or rotation out of artificial intelligence. Toyota rose because of its own results.

Korea — Concentration and a Genuine Decline

Korea cannot be described in Japan's terms. The Kospi closed down 5.28% at 5,383.80, and the fall was severe enough to trip an exchange mechanism: a sell-side sidecar suspended program trading in Kospi-listed shares for five minutes from 10:16, triggered when the Kospi 200 Futures Index fell 5% or more for at least a minute. SK Hynix lost 9.46% and Samsung Electronics 7.57%, and Korean reporting puts the two together at roughly half the index's capitalisation. Concentration amplified the move, but it did not manufacture it.

There were pockets of strength consistent with rotation rather than flight. Celltrion gained 7.5%, Hanmi Pharmaceutical 9.8% and Samsung Epis Holdings 14.33%. That supports a sector-rotation reading without establishing that the wider Korean market held up.

The more revealing observation sits a day earlier. On Wednesday the Kospi closed 0.62% higher at 5,683.80 while foreign investors sold a net ₩2.51tn, with domestic retail buying ₩1.14tn and institutions ₩1.01tn. Foreign selling was already substantial before Thursday's decline, absorbed by domestic buyers and concealed by a positive close. That is evidence of institutional repositioning under way ahead of the shock, not proof that foreign investors anticipated it.

Taiwan — A Broader Sell-Off

Taiwan is the counterexample that prevents the Japanese finding from being over-generalised. The Taiex closed at 41,574.72, down 1,374.19 points or 3.20%, on turnover of NT$926.02bn. Its internals look nothing like Tokyo's: 170 shares advanced, 872 declined and 69 were unchanged, and Delta Electronics fell 7.38%, well outside the most prominent chip manufacturers.

So the same regional catalyst produced a narrowly concentrated index decline in Japan and a genuinely widespread one in Taiwan. Any account of Thursday that attributes Asian benchmark losses purely to index-construction effects is wrong about Taiwan, and any account that treats the region as uniformly sold is wrong about Japan.

Two Catalysts, Not One

The attribution is genuinely contested, and this briefing does not resolve it because the sources do not.

Reuters attributes the Asian chip selling to earnings. Qualcomm's handset chip revenue fell 26% to $5.14bn and the shares dropped more than 9% in extended trading; Arm beat on revenue and adjusted earnings but its royalty line rose only 8%, to $670m against a $720.8m estimate, and the shares slipped about 3%. Reuters does not connect the Asian selling to the election at all.

AP, the Korea JoongAng Daily and Reuters' own Tokyo file attribute the same selling to the midterm result and to expectations that a Democratic Congress slows data-centre expansion. Nomura's Wataru Akiyama framed it as funds rotating out of AI and semiconductors, with Democratic gains likely to delay data-centre plans and cap earnings growth — while adding that US–China strategic competition makes an outright reversal of the construction trend unlikely.

Both channels are supported and neither is quantified. They also do not apply uniformly: Qualcomm's problem is smartphones and customer mix, Arm's is royalty monetisation, and neither demonstrates that hyperscalers are cutting data-centre investment. The most defensible reading is that Thursday repriced expectations and valuations rather than observed demand. The earnings releases and market reports cited here do not establish cancelled projects or an industry-wide reduction in capital expenditure.

What happened to the buyers is the clearest evidence for that reading. Microsoft closed 6.7% higher in New York on Wednesday, Amazon 6.1%, Alphabet 3.2% and Meta Platforms 1.4%, while several specialist data-centre operators fell by double digits. Investors appeared to distinguish established cloud operators, which already control substantial computing capacity, from the companies financing and supplying further expansion. The former rallied while several infrastructure specialists and semiconductor suppliers fell, consistent with expectations that restrictions could constrain future capacity growth.

China, Hong Kong and the Policy Backdrop

Greater China did not follow Korea and Taiwan. The Hang Seng rose sharply, with mainland indices far more subdued, and the divergence is consistent with investors separating domestic Chinese technology and policy-supported sectors from businesses geared to US data-centre spending.

Two People's Bank of China measures sit underneath that, and they are distinct instruments rather than a single easing. The central bank cut the rate on its share-buyback and share-purchase relending facility by 25 basis points to 1%, with bank applications for quota under the scheme exceeding ¥200bn, approximately $30bn: that is targeted equity-market support. Separately, announced on 3 November, it is raising the reserve requirement on banks' foreign-currency deposits from 4% to 6% with effect from 15 November, which tightens foreign-currency liquidity to moderate the renminbi's appreciation. Neither is equivalent to a broad policy-rate reduction.

Mainland closing levels are not settled across the feeds available at the time of writing, so this briefing describes mainland China qualitatively rather than selecting between them.

India, Southeast Asia and the Oil Channel

India extended its outperformance, the Nifty 50 adding 2.01% to 24,442, with Indonesia's IDX Composite up 2.32% to 5,034 and Australia's ASX 200 up 2.6% to 9,065.10. Falling crude and lower US yields — the 10-year near 4.85% and the two-year around 4.56% in Thursday morning trade — are plausible supports, and Brent has fallen below $100 after a four-session decline.

The oil channel needs handling by sector rather than by region. A lower crude benchmark does not translate proportionately into imported refined-product prices, which remain constrained; India's substantial domestic refining industry gives it a different exposure from economies buying finished diesel; and energy producers and energy-weighted equity sectors lose from the same decline that helps consumers. Singapore's banks, Indonesian commodity producers, Malaysian energy exposure and Philippine domestic consumption do not belong in a single regional sentence.

What This Means for Asian Equities

The valuation question is the one Thursday leaves open and the one that matters most. A sell-off of this magnitude in semiconductors is not evidence of an equivalent deterioration in expected earnings. It is at least as consistent with investors reducing high multiples and crowded positioning as with any change in forecast cash flows, and the day's data cannot separate the two.

Nor should the affected exposures be treated as one. AI compute demand and hyperscaler capital expenditure are different variables. Memory pricing and high-bandwidth-memory demand are different again, as are semiconductor equipment orders, smartphone and consumer-electronics demand, and policy risk attaching to data-centre construction. Qualcomm's handset disappointment says very little about SK Hynix's high-bandwidth memory outlook, and Thursday's tape did not distinguish between them.

One further observation belongs on the record. Early-session prints understated the damage materially: Samsung was down 5.7% intraday against 7.57% at the close, SK Hynix 7.4% against 9.46%, Advantest around 11% against 14.7%. Selling accelerated into the close rather than stabilising, with Lasertec down 12.5% and Kioxia 9.2% in early trade.

Outlook and Key Risks

The central question into next week is whether Thursday was a repricing or the start of a de-rating. Four observations will separate them: whether the Kospi stabilises or triggers further sidecars; whether any data-centre rule is actually proposed, given that the incoming Congress does not take office until January; whether hyperscaler capital-expenditure guidance confirms reduced spending, which is the only series that settles the argument directly; and whether Korean foreign outflows continue once domestic absorption is exhausted.

The risks run in both directions. Codified US restrictions on data-centre development would validate Thursday's move and extend it. Equally, Akiyama's point stands: strategic competition with China makes a genuine reversal of the construction trend unlikely, which would leave a market that has repriced an expectation rather than an outcome.

Intelligence Monitoring Points

Korea Exchange sidecar and circuit-breaker frequency, since repetition distinguishes forced selling from a single repricing. Korea Exchange foreign and institutional flow data, to establish whether domestic absorption continues. Nikkei and TOPIX divergence, as the cleanest available measure of whether selling broadens beyond the index heavyweights. Taiwan Stock Exchange daily breadth, for the same reason in reverse. US hyperscaler capital-expenditure guidance, and any formal federal data-centre rule proposal.

FAQ

Why did the Nikkei fall when most Japanese stocks rose? The Nikkei 225 is price-weighted, so high-priced constituents carry disproportionate influence. Advantest, Tokyo Electron and SoftBank Group between them outweighed the 161 advancing constituents. The capitalisation-weighted TOPIX rose 1.2% on the same session.

Was this an Asia-wide sell-off? No. Korea and Taiwan fell on genuine breadth, Japan's decline was concentrated in a handful of index heavyweights, and Hong Kong, India, Indonesia and Australia rose.

What caused the semiconductor selling? Two catalysts arrived within 24 hours: disappointing Qualcomm and Arm results, and expectations that a Democratic Congress slows data-centre expansion. Reuters attributes the Asian selling to the earnings; AP and Korean reporting attribute it to the election. No source quantifies the split.

What is a Kospi sidecar? A five-minute suspension of program trading in Kospi-listed shares, triggered when the Kospi 200 Futures Index falls 5% or more for at least one minute. One was activated at 10:16 on 5 November.

Has any AI data-centre regulation been passed? No. The expectation rests on state ballot measures and anticipated congressional scrutiny. No federal rule has been proposed and the incoming Congress does not take office until January.

Report Methodology

Index levels and percentage moves are closing figures for the 5 November Asian session unless identified as intraday, and US equity figures are closes for the 4 November session. Reported closing levels for the Nikkei 225 differ materially between data providers for this session; this briefing therefore cites the percentage change reported by Associated Press and Reuters and does not state a closing level or a numerical index-point decomposition. Where feeds carried differing mainland Chinese closing levels, mainland China is described qualitatively rather than by selecting between them. Taiwanese market breadth figures are as reported by CNA. Korean flow figures are for the 4 November session. Early-session observations are labelled as such and are not presented as closes.

Sources


This research is provided for informational purposes only and does not constitute investment advice, an investment recommendation or an offer to buy or sell any financial instrument or commodity. Data and estimates may be revised. Readers should undertake their own independent assessment of market conditions and risks.