Author: Claude, Shenchao TechFlow
DeepInsight Summary: On May 11, Nomura released a research report making a key observation: “At least in the U.S. stock market, the AI rally may be taking a breather.” On the same day, the KOSPI surged 4.32% to 7,822.24 points, triggering a buy-side cascade; SK Hynix jumped 11.98% in a single day, surpassing Eli Lilly in market cap for the first time to become the world’s 14th-largest company. The report’s prediction of “the next leg driven by FOMO in Korea” coincided almost exactly with the surge in Korean equities. The momentum behind U.S. AI trading is shifting from “short squeeze” to “retail FOMO.”
The U.S. AI market rally is not over. The S&P 500 has risen approximately 16.6% over 28 trading days, but the source of the funds driving the rally and the remaining fuel for further gains are undergoing subtle shifts. Nomura’s assessment is that the phase driven by short covering and institutional rebalancing is nearing its end; for the AI trade to continue, a new wave of capital must step in. The Korean market provided a perfect example on the day the report was released: the KOSPI surged past three key thresholds—7,000, 7,400, and 7,800—within a week, as retail investors succumbed to “Hynix FOMO” and overseas funds collectively increased positions in chip stocks via DRAM ETFs. The narrative is now shifting from the Nasdaq to the KOSPI.
U.S. stocks appear to be operating normally, but the unusual combination of rising spot prices and rising volatility has already signaled a warning.
The surface readings for AI trading in U.S. equities remain robust. Saxo’s May 11 options briefing shows the VIX closed at 17.19, up 0.64% on the day. While this level sits below the historical average, the fact that the VIX is rising alongside record-high index levels is itself an unusual signal. The CBOE SKEW Index rose to 138.21 (+1.54%), and the VVIX, which measures the volatility of the VIX, increased to 96.78 (+3.39%). The simultaneous rise in all three indicators suggests institutional investors have not reduced their hedging positions despite the index reaching new highs.
Nomura described this combination as an "anomalous trend" for U.S. tech stocks in its research report on May 11. The report noted that the Nasdaq has moved into a scenario where both spot prices and volatility have risen, while the VIX has continued to decline and the VXN (Nasdaq volatility) has shown a clear rebound; the options skew for U.S. tech stocks (the difference between the implied volatility of 1-month 25-delta put options and 25-delta call options) has rapidly declined to near historical lows, returning to levels seen around October 2025. The decline in skew indicates that the premium for put protection relative to call options has been compressed, signaling increased market congestion in pricing for upside moves in tech stocks.
More noteworthy is the structure of this rally. According to a return attribution chart cited by TECHi from Nomura’s Vol team, of the S&P 500’s approximately 16% gain since March 30, ten stocks accounted for 69%: Alphabet, NVIDIA, Amazon, Broadcom, Intel, Micron, Apple, AMD, Microsoft, and SanDisk. The remaining 490 components contributed only 31%. Goldman Sachs’ U.S. Equity Strategy Head, Ben Snider, also noted that current market breadth has narrowed to one of its lowest levels since the dot-com bubble. Goldman Sachs has identified “the AI build-out” and “Iran conflict” as the two most clearly defined market risks over the coming weeks.

The short squeeze is over—who will drive the next move?
The real killer in the Nomura report is not the "abnormal combination" itself, but the analysis of funding conditions: quantitative funds' equity exposure has returned to near-neutral levels, with the prior forced buying and short-covering process largely complete. CTA (commodity trading advisor) funds have largely returned to fully long positions, and the marginal demand from volatility-controlled strategies is also diminishing.
In other words, the three main buying forces that drove AI stocks higher over the past few weeks—short squeezes, CTA position increases, and volatility decline triggering vol-control leveraged buying—are now nearing their limits. If AI stocks are to continue rising, they can no longer rely primarily on the force of short squeezes.
Note that Nomura’s estimates for positions in quantitative funds, CTA funds, macro funds, and others are based on model projections, not actual measured holdings. This means it is better suited as a barometer for marginal changes rather than a precise snapshot of positions. Nevertheless, the direction is clear: institutional algorithmic buying is nearing its limit, and future upward momentum will need to rely more heavily on retail investors and sentiment-driven capital.
Goldman Sachs’ trading desk aligns closely with Nomura’s assessment. Rich Privorotsky, head of Goldman Sachs’ One-Delta trading desk, previously described the current momentum as “semi-irrational FOMO,” drawing a parallel to 1999, when surging orders for telecommunications equipment created a “physical bottleneck narrative” similar to today’s scarcity of AI computing power. Goldman’s volatility trading desk has characterized the recent market dynamics as “spot price rising alongside volatility,” which has constrained further position building by systematic strategies.
This suggests that AI trading in U.S. stocks has not collapsed, but the script of continuing to push prices higher by squeezing short sellers is nearing its end.
Korean stocks provide the answer: On the day the Nomura report was released, the KOSPI rose 4.32% in a single day, triggering a buy-sidecar.
Another observation in the Nomura report is: If AI trading is to experience another leg up, the true continuation signal will depend on whether FOMO reappears in Korea.
On the day the research report was released, the Korean market responded with an extraordinary surge. The KOSPI closed at 7,822.24 points, rising 4.32% for the day and hitting an intraday high of 7,899.32 points, triggering a buy-side sidecar. SK Hynix surged 11.98% to KRW 1.888 million per share, surpassing Eli Lilly for the first time to become the world’s 14th-largest company by market cap. Samsung Electronics rose 6.33% to KRW 285,500 per share; together, the two companies’ combined market capitalization exceeded KRW 3,000 trillion, accounting for nearly half of the KOSPI’s total market value. The combined market capitalization of Korea’s stock and KOSDAQ markets surpassed KRW 7,000 trillion for the first time, just eight trading days after crossing the KRW 6,000 trillion mark on October 27.
On May 12, during trading hours, the KOSPI surpassed the 3,900-point level (equivalent to the 7,900-point level), setting a new all-time high. However, data from the same day revealed another side of FOMO: among the 948 stocks on the KOSPI, only 186 advanced while 696 declined; approximately 30% of the index components have declined year-to-date. The gains were entirely concentrated in two semiconductor heavyweights: Samsung and SK Hynix.
FOMO among retail investors has become a new market term. Korean financial media use “Hynix FOMO” to describe the divided sentiment among retail investors—on one side, regret over missing out: “I should’ve bought at 800,000 KRW”; on the other, anxiety over whether to jump in now and whether a correction is imminent. Retail communities are flooded with discussions about “Samjeon-nix” (a portmanteau of Samsung and Hynix). This is a classic retail-driven momentum pattern that closely aligns with Nomura’s definition of a “FOMO signal.”
The flow of overseas capital is more telling. According to the Seoul Economic Daily on May 10, the iShares MSCI Korea ETF (EWY) experienced a net outflow of $1.0145 billion between May 1 and 7, signaling passive funds withdrawing from the Korean market. However, during the same period, the Roundhill Active DRAM ETF saw a net inflow of $1.9538 billion, with SK Hynix accounting for 25.94% and Samsung Electronics for 21.62%, totaling approximately 48%. Overseas capital is not selling Korea—it is selling broad-market funds and buying chips, precisely increasing exposure to the AI theme.
There is one detail worth remaining alert to. Nomura’s report on May 11 noted that the KOSPI 200 also experienced “spot price rise alongside volatility rise,” but the call option skew did not increase, suggesting this volatility expansion was not driven by demand for call options fueled by FOMO. In other words, as of the report’s publication, the Korean market had not yet entered the typical “fear of missing out, rushing to buy calls” state. Whether this signal quickly reverses after the KOSPI’s sharp rally today will be key to assessing the sustainability of FOMO.

Korean stocks are an extension of the U.S. AI capital expenditure chain; how long the next leg can last depends on the "top of the pyramid."
South Korean stocks' FOMO is not an isolated event; it is essentially a high-beta extension of the U.S. AI capital expenditure narrative.
Data directly anchors this transmission chain. According to Bridgewater, Alphabet, Amazon, Meta, and Microsoft are estimated to collectively invest approximately $650 billion in AI-related infrastructure by 2026. Goldman Sachs, citing data, notes that the consensus estimate for capital expenditures by the largest cloud infrastructure companies rose by $130 billion in the most recent quarter to $670 billion, equivalent to over 90% of these companies’ projected operating cash flow. Microsoft’s capital expenditures for its third fiscal quarter reached $31.9 billion, Alphabet disclosed $35.7 billion in property and equipment purchases in its Q1 earnings report, and Meta has raised its 2026 capital expenditure guidance to a range of $125 billion to $145 billion.
This capital flow is directed toward data centers, GPUs, memory, networking, power systems, and cloud capacity. SK Hynix and Samsung are positioned at the heart of this flow, as HBM4 memory and HBM high-bandwidth memory are being fiercely sought after by hyperscale cloud providers. According to Reuters, SK Hynix recently received unprecedented order proposals from major tech companies, with some customers even offering to finance new production lines and ASML lithography machines. Chip production capacity is nearly exhausted. This is why the KOSPI’s single-day surge of 4.32% makes perfect sense in narrative terms—Korean equities are essentially the second derivative of the U.S. AI story.
However, this linkage also implies vulnerability. If U.S. tech stocks experience a sharp reversal, Korean equities will be the most direct high-beta assets to absorb the selling pressure. Another risk pathway mentioned by Nomura is rising inflation forcing global central banks to adopt a more hawkish stance; this week’s U.S. CPI report (May 12) is a key event, and currently, option markets are still pricing in a low premium for this event—meaning the market has not yet paid a high insurance cost for this risk.
On a macro level, another variable is the Strait of Hormuz. On May 8, WTI crude closed at $100.09 (+4.89%), and Brent crude at $105.66 (+4.31%), as tensions near the Strait of Hormuz continue to escalate. Nomura’s assessment is that as long as the strait remains blocked and differences persist between the U.S. and Iran over ceasefire conditions, the AI-driven market environment may last longer than expected. Energy price disruptions will elevate inflation expectations, but they will also make markets even less willing to abandon the “AI story” that delivers profits.
Stacking the above clues together, the phase of U.S. tech stocks being driven higher by short squeezes in the AI sector is nearing its end; FOMO in South Korean stocks has been ignited, with retail investors and overseas chip ETFs simultaneously increasing positions, though options skew has yet to catch up. How long the next leg can last depends on whether U.S. tech stocks correct, whether the U.S. CPI signals accelerating inflation, and whether the Strait of Hormuz ultimately cools down. Nomura’s analytical framework has been progressively validated by market moves, with Seoul emerging as the latest epicenter of this AI trading cycle.
