NextFin

Citi's Poon Werkun Weighs AI Trade Unwind and China as the Fed Turns Hawkish

Summarized by NextFin AI
  • Citi's Asia Pacific Head of Research Erica Poon Werkun identifies three colliding forces for global investors: the crowded AI trade unwinding, the Fed turning hawkish with a near-certain December rate hike, and China's markets facing a year where policy and export clarity outweigh AI momentum.
  • The AI trade unwind is a cyclical positioning reset, not the end of capex; Citi downgraded U.S. equities to neutral, and the S&P 500 saw its largest weekly decline since April 2025, erasing roughly $2.3 trillion from the Nasdaq 100 in one session.
  • The Fed is tightening, not easing, with annual headline inflation around 4% and core at 2.8%; markets price a 100% probability of a 25-basis-point hike next month, keeping the dollar strong as a headwind for Asian equities and currencies.
  • China's new economy now exceeds 18% of GDP, structurally offsetting property drag, but Citi downgraded China to Neutral from Overweight, citing less favorable EPS revision dynamics and a lackluster macro picture despite the Hang Seng trading near 24,339.

NextFin News - Erica Poon Werkun, Asia Pacific Head of Research at Citi, delivered a three-part assessment for global investors this week: the crowded AI trade is unwinding, the Federal Reserve is turning hawkish with markets now pricing a near-certain December rate hike, and China's equity markets face a year where policy and export clarity will matter more than the AI momentum that powered 2026. The three forces collide squarely in Asia, and the interaction between them - not any single theme - is what will drive returns into 2027.

Speaking with Stephen Engle on the sidelines of the 21st Annual Citi China Conference in Shanghai, Poon Werkun framed the unwinding of the AI trade, the outlook for U.S. monetary policy, and China's markets and economy as the interconnected questions defining positioning. The conference, Citi's flagship China event, runs November 10-13 across Shanghai and Shenzhen and is expected to draw 1,500 global institutional investors, policymakers, and corporate leaders.

The AI Trade Unwind: A Positioning Reset, Not the End of Capex

The timing of Poon Werkun's comments is deliberate. They come just days after Citi's global strategy team, led by Beata Manthey, downgraded U.S. equities to "neutral" from "overweight" and cut its stance on global and European technology to neutral from overweight. The shift followed Democratic victories in the U.S. midterm elections that sparked fears of stricter artificial-intelligence regulation, triggering the S&P 500's largest weekly decline since April 2025 and erasing roughly $2.3 trillion from the Nasdaq 100 in a single Thursday session.

"We worry that positioning on key indices, such as the Nasdaq, could reset lower into net-short territory," Manthey wrote in the November 9 client note. "While the selloff has been most aggressive in Tech, volatility has been broad-based."

The first question investors must answer is whether this unwind is a cyclical correction or a structural regime change. The evidence points to cyclical positioning unwinding layered on top of a structural capital-expenditure cycle that is not over. Citi's research has argued the pullback was driven less by the election itself than by AI financing worries and oil prices - a financing and positioning shock rather than a demand shock. The bank still expects data centers to get built given strong demand for computing power, but says their location, design, and pace of construction may change.

That distinction is the mechanism. Demand for compute is structural, underpinned by enterprise AI adoption that a Federal Reserve report put at roughly 41% work-related generative-AI usage in the United States as of late 2025, with non-work usage at 50%. But the financing of that demand - valuation multiples, leverage, and ownership concentration - is cyclical. When financing conditions tighten or regulation looms, the multiple compresses even while the underlying units keep shipping.

The second-order implication is what much of the market is missing. An unwind in the U.S. AI trade does not mean an unwind in AI globally. It means a rotation: capital exiting the most crowded, most expensive U.S. picks-and-shovels names and searching for AI exposure with cheaper valuations and lower regulatory risk. That is where Asia enters. Taiwan and South Korea's semiconductor supply chains - which Poon Werkun has noted carry 70% to 80% concentration in technology accounts - become both the beneficiaries of continued capex and the most exposed to a deeper positioning reset. A 10% drawdown in the U.S. leaders can become a 25% drawdown in the more concentrated Asian suppliers, even if the end demand is identical.

The strongest counter-thesis deserves its weight: the AI trade is not unwinding at all. This is a healthy correction inside a secular bull market, and investors calling an unwind are simply late. That case rests on two pillars - hyperscaler AI capex guidance remains strong, and semiconductor-chain earnings continue to beat. The falsifying signal for the unwind view is specific and observable: if Nasdaq positioning fails to reset toward net-short territory and AI capex guidance for 2027 is reaffirmed or raised through the January earnings cycle, the "unwind" framing is wrong and this was merely a drawdown in a continuing uptrend.

The Fed Pivot: Tightening, Not Easing, Changes the Asia Equation

The second theme is the Federal Reserve, and here the picture has shifted materially from the narrative that dominated earlier in 2026. The Fed is no longer in an easing cycle. Inflation has come in above expectations for a third successive month, with annual headline inflation around 4% and core inflation at 2.8% - well above the 2% target. The central bank delivered its first rate rise of the cycle in September, and markets now price a 100% probability of a 25-basis-point hike when policymakers meet next month, with some officials preparing the ground for more than one increase.

This is the single most important correction to the consensus view that has built up around Poon Werkun's framework. In Citi's 2026 outlook, she laid out a clear transmission chain: a Fed easing cycle, combined with a global soft landing, has historically been a powerful catalyst for emerging-market equities. The mechanism runs through the dollar - Fed cuts weaken the greenback and cheapen dollar funding, which pulls capital back into emerging markets and lets EM equities re-rate. Citi expected emerging markets to deliver the strongest earnings growth of any major region in 2026, with consensus forecasting around 17% EPS growth, and the Fed was to be the enabler of that re-rating.

That transmission chain now runs in reverse. Every delay in easing - or every hike - keeps the dollar stronger for longer, and a strong dollar is the single most reliable headwind for Asian equities and currencies. The two-year Treasury yield jumped 19 basis points and the 10-year yield rose 15 basis points to 5.01% as the repricing hit, a move that tightens financial conditions across Asia regardless of local fundamentals. This is why Poon Werkun has noted that H-shares are more sensitive to U.S. dollar rate cuts than A-shares: H-shares are dollar-funded and foreign-owned, so they re-rate first when the dollar weakens - and sell off first when it strengthens.

The cyclical-versus-structural call on the Fed is clear: this is a cyclical support factor that has flipped from tailwind to headwind. Rate hikes do not change China's earnings trajectory or resolve its property overhang; they change the discount rate and the currency backdrop. The adversarial case is that inflation proves transitory - that falling oil prices, now back around $92 a barrel, and sliding gasoline prices pull inflation back toward target fast enough for the Fed to pause rather than hike further. The falsifying signal for the "Fed stays hawkish" thesis is quantifiable: if core PCE prints below 0.2% month-over-month for two consecutive months and the next payroll report adds fewer than 100,000 jobs, the December hike is off the table and the dollar-strength thesis for Asia breaks.

China: A Structural Transition Riding a Cyclical Rally

The third theme is China, and this is where the cyclical-versus-structural distinction becomes the entire investment question. On the surface, 2026's China rally looks cyclical: policy stimulus, an export rebound, and AI enthusiasm lifted the CSI 300 above 4,000 and the Hang Seng Index through 2026. As of November 11, the Hang Seng stood at 24,339, down 0.89% on the day and roughly 5% year-to-date, trading within a 52-week range of 22,518 to 28,056. The STAR 50, the mainland's technology-heavy benchmark, was reported up roughly 29% year-to-date on the AI and chip theme.

But Citi's research argues the deeper story is structural: the "new economy" - technology, advanced manufacturing, green energy - has grown large enough to offset the property drag. China's new economy, as measured by the National Bureau of Statistics, is now over 18% of GDP. "So, in a broader sense, the new economy can now fully offset the property's drag," Citi's economists argued in April. That is a structural claim: the growth engine has changed, and the old framework centered on property misses how much the landscape has shifted.

There is a catch, and it is the catch Poon Werkun's November comments surface. The 2026 rally was led by AI and technology - the same theme now unwinding in the United States. If the AI trade unwind is cyclical and contained, China's tech complex can decouple on domestic AI adoption and self-reliance policy under the 15th five-year plan. If it is structural and deep, the STAR 50 is the most exposed index in Asia. Citi's own positioning reflects this ambivalence: the bank downgraded China to Neutral from Overweight for 2026, citing "less favorable EPS revision dynamics and a lackluster macro picture," while upgrading Taiwan to Overweight on stronger AI and technology visibility. Yet the strategists also cautioned that "sizable China policy easing or China-specific AI headlines (ie, akin to a 'DeepSeek moment') could challenge this relative call." In other words, the China call is conditional on policy, not on fundamentals alone.

The mechanism for China's path runs through three channels: the 15th five-year plan's emphasis on tech self-reliance and export growth; the property market's drag, which the new economy must continue to offset; and the external environment, where export clarity with the United States remains the swing factor. Poon Werkun has said Citi is more optimistic about China's development compared with a year earlier because of "positive catalysts from China market domestically from 15th five-year plan and externally for more export clarity."

The adversarial case against China is the strongest in this piece. China's earnings revision index has remained in net downgrade territory with few signs of a broad inflection. The property overhang is not resolved - it is merely being offset by a sector that is itself now tied to the unwinding AI trade. Demographics are structural, not cyclical, and no five-year plan reverses a shrinking workforce. The falsifying signal for the constructive China view is specific: if China's earnings revision index does not move into net-positive territory by the first quarter of 2027 and the Hang Seng fails to hold the 18,290 medium-term support level, the "new economy offsets property" thesis is a valuation story without earnings to back it.

What It Means: Rotation, Not Collapse

Pulling the three threads together, the base case is a rotation rather than a collapse. The AI trade unwind is cyclical positioning, not the end of AI capex - data centers still get built, but the winners shift from the most crowded U.S. names to the supply chain and to regions with cheaper valuations. The Fed is tightening, not easing, which keeps the dollar a headwind and makes stock selection in Asia more important than index beta. China's new economy is a structural reality, but its equity market remains a cyclical bet on policy that has not yet been confirmed by earnings revisions.

Split by time horizon:

  • Short term (next three months): volatility dominates. Watch the Fed's December meeting for the expected 25-basis-point hike, and watch whether Nasdaq positioning resets to net-short as Citi's strategists expect. A clean positioning reset is actually the setup for a year-end rally, not the prelude to a crash.
  • Medium term (six to twelve months): the dollar path is the gate. If the Fed's hiking cycle proves short-lived and the dollar rolls over, H-shares and dollar-sensitive Asian markets re-rate first. Without an inflection in China's earnings revisions, the rally stays policy-driven and fragile.
  • Long term (two to three years): AI capex is structural, and Asia's semiconductor supply chain is the structural beneficiary regardless of which U.S. names win or lose. China's new-economy transition is real, but it will be measured in earnings, not announcements.

The beneficiaries are the AI supply chain in Taiwan and Korea on continued capex, European equities - which Citi upgraded to overweight as a defensive rotation given AI infrastructure represents roughly 15% of the Stoxx 600 versus about 40% of the S&P 500 - and quality Asian names with dollar revenue. The exposed are the crowded U.S. AI picks-and-shovels names, China's STAR 50 if the AI unwind deepens, and any Asian market that rallied purely on the hope of a Fed cut that is not coming.

The signal that would prove this whole framework wrong is a specific one: if the Fed pauses or cuts in 2027 while China's earnings revisions inflect positive and the AI trade keeps climbing, then the rotation thesis fails and the old regime - a weaker dollar, strong U.S. tech, and broad Asia participation - persists. That is a low-probability but high-impact scenario, and it is why positioning, not conviction, is the right posture.

The market is not being asked to choose between AI and no-AI, or between China and no-China. It is being asked to price a world where AI capex survives its financiers, where the Fed tightens into inflation, and where China grows differently than it did before. That is a harder world to invest in - and a more interesting one.

Explore more exclusive insights at nextfin.ai.

Insights

What drives the AI trade unwind?

How does Fed policy affect Asia?

What is China's new economy share?

Why is the Fed turning hawkish now?

How did Nasdaq 100 recently fall?

What is current US core inflation?

Where does Hang Seng stand today?

How much AI usage is work-related now?

Why were US equities downgraded now?

What is the December Fed rate call?

Will AI capex continue in 2027?

Who benefits from AI supply chains?

Can China offset property drag?

What signals China earnings inflection?

Is AI unwind cyclical or structural?

What risks face Asian semiconductors?

Why is STAR 50 most exposed in Asia?

Does demographics limit China growth?

How H-shares differ from A-shares now?

Why upgrade Taiwan stocks over China?

Search, ask or analyze anything...