Переоценка индустриальной цепочки за падением тайваньского фондового рынка — от горячего тренда инвестиций в ИИ до рациональности оценки стоимости

Taiwanese stock market experienced a night of transition from “euphoria to rationality.” Last Friday’s sharp decline in U.S. technology stocks triggered a domino effect in the Taipei stock market. After a gap-down opening, the weighted index accelerated downward, plunging over 500 points to a low of 27,684 points, marking the eighth largest decline this year and signaling the breakthrough of the 28,000-point level.

The most direct victims of this adjustment were electronic blue-chip stocks. TSMC ADR plunged 4.2%, with domestic shares dropping 30 yuan to 1,450 yuan, simultaneously breaking through the monthly line. King semiconductor Siliconware Precision Industries staged intense battles near the 6,600 yuan level, probing as low as 6,590 yuan before finding buying support. Top-tier premium stocks (share price above 3,000 yuan) turned red across the board, a rare sight in Taiwan stock history.

The Real Logic Behind Value Reassessment

The seemingly abrupt stock price crash actually stems from a fundamental shift in market expectations regarding AI investment. On the surface, it’s U.S. technology stock volatility—Broadcom fell an astonishing 11.43% following earnings, with Nvidia declining over 3%—but the deeper change warrants closer attention.

This is not a crisis signal for AI demand. Broadcom’s earnings report clearly indicated that AI-related orders for the next 18 months have exceeded $73 billion, and market perception of industry prospects has not changed. What truly changed is investors’ evaluation framework.

Over the past two years, the “AI concept” itself was sufficient to drive valuation increases, with investment logic simple and crude—merely attaching an AI label could translate order growth into stock price gains. But when industry leaders like Oracle and Broadcom released earnings, the market began questioning a long-overlooked issue: can these massive orders translate into actual profits?

Broadcom’s transformation signal is particularly crucial. The company is transitioning from “selling high-margin chips” to “selling system solutions,” which means gross margins may face pressure. Oracle holds $523 billion in orders, with $300 billion from OpenAI, but the market has begun questioning the certainty of order fulfillment.

Long-term investment recovery cycles combined with lower-than-expected profit margins have made the market cautious toward enterprises deeply bound to OpenAI—including Oracle, SoftBank, Microsoft, and Nvidia, whose stock prices have collectively come under pressure since late October.

Cash Flow Reveals Market’s True Intentions

Notably, today’s adjustment is not a signal of mass capital flight from the market. From market performance, oil, electricity, and gas stocks rose 3.09% against the trend, while networking stocks and shipping stocks rose 1.33% and 1.25% respectively, whereas glass stocks fell 2.59% and other electronics stocks fell 2.15%.

This phenomenon indicates capital is reallocating direction—flowing from crowded AI midstream supply chain sectors toward stable assets with clear cash flows, valuations not yet extremely inflated, and insensitivity to interest rate environments. The market has not negated AI industry prospects but is seeking more certain investment anchors within industry chain differentiation.

Winners and Losers in Industry Differentiation

Not all enterprises appear dim in this adjustment. Xcur Therapeutics performed brilliantly, with share prices surging over 8% to a new high of 2,370 yuan. The company benefits from next-generation smartphone and high-end tablet stocking momentum, achieving consolidated revenue of 4.415 billion yuan for the first 11 months, up nearly 40% year-on-year, with markets optimistic about sustaining double-digit growth for the full year.

King semiconductor Siliconware Precision Industries similarly demonstrated resilience. Supported by smooth supply chains and better-than-expected shipments, the company raised guidance for this quarter’s operations for the second time, optimistic that 2025 will be the peak year with order visibility extending into Q2 next year.

These two examples reflect future industry differentiation trends: enterprises that merely rely on “AI concepts,” have single-customer structures, and lack profit support will face sustained valuation pressure; while enterprises with core technologies, stable profitability, diverse customer bases, and clear growth pathways will stand out in the market’s rational screening process.

Year-End Variable Stacking Tests the Market

Taiwan’s stock market currently faces stacking uncertainties. U.S. index volatility directly affects foreign investor allocation, and last Friday’s crash naturally triggered today’s chain reaction. More noteworthy is the implementation of IFRS 17 and TW-ICS accounting regime conversion for life insurance companies next year, which may trigger active or passive selling pressure.

This life insurance adjustment logic warrants deeper investigation. Under the new regime, if stocks are classified as FVOCI, future sales at high prices cannot enter P&L but only shareholders’ equity, effectively cutting off the past practice of beautifying EPS through stock sales. Therefore, life insurers tend to convert accumulated unrealized gains into realized profits before regime switching.

The Bank of Japan interest rate hike expectations during this week’s “super central bank week” may also trigger carry trade exits, further exacerbating market volatility.

From Bubble Panic to Industry Maturation

When the market uses “bubble” to describe this adjustment, it actually misreads the market’s true signal. This is not bubble burst but a necessary phase in industry maturation.

From a medium to long-term perspective, the real watershed lies in corporate competitive fundamentals. Google controls OpenAI’s scarcest resources: cash flow and complete industry chain. Its projected 2026 capital expenditure accounts for 56% of operating cash flow, the most efficient among giants. The cost advantages brought by this vertical integration are extremely evident—Google’s TPUv7 TCO is approximately 44% lower than Nvidia’s GB200 server.

In the future, AI sector differentiation will become the norm. Those enterprises with genuine core technologies, stable profitability, and diverse customer bases will achieve steady growth in the market’s rational screening, while those lacking substantive fundamental support for speculation may face continued valuation adjustments.

Taiwan’s stock market adjustment essentially represents the process of market returning from euphoria to rationality, and a necessary price for investors to learn to distinguish between “stories” and “fundamentals.”