가보지 못한 길...우리가 놓치고 있는 게 있습니다 (김학균) | 인포맥스라이브 2026년 8월7일 방송

Finance-Focused Summary (Markets, Investing, Valuation, Risk)

Market Drawdown & Historical Analogies

• KOSPI saw a steep correction: ~38% down over 27 trading days, with the lowest point cited as July 30 versus the peak last June.
• The magnitude is described as rare / first in the dataset:
  • No prior instance of roughly a 38% drop over 27 trading days in Korea’s stock market history is cited.
• Comparable periods mentioned:
  • Dec 1997: largest drop ~32% over about 27 trading days
  • 2008 Global Financial Crisis: about ~33% decline following Lehman’s collapse
  • COVID: cited around ~34% during the comparable period
• Interpretation offered: the selloff resembles a “knockout” pattern—after enough damage/pessimism accumulates, the final leg falls when a system breaks.
• Global/macro context: the speaker argues Korea is hit uniquely hard:
  • US (S&P 500 / Dow): portrayed as not showing the same broad “risk-off”
  • Europe (Germany/France): described as not collapsing as much
  • UK and Spain: referenced near all-time highs
  • Japan & Taiwan: corrections <10%
  • Conclusion: Korea’s move is largely tied to the memory semiconductor downturn

Semiconductors: What Broke, and Why Korea Fell More

• Korea’s index concentration is emphasized:
  • On June 25, Samsung Electronics + SK Hynix weight is cited as ~60%.
  • As a result, memory-cycle earnings fears transmit quickly to the broader market.
• Drawdowns from prior highs:
  • Samsung Electronics: -42%
  • SK Hynix: -54% (described as “cut in half”)
  • Other global memory peers (examples):
    • Kioxia: -64%
    • SanDisk: -56%
    • Micron Technology: -40%
• Memory-cycle framing:
  • Debate: whether the memory semiconductor cycle has ended
  • Speaker’s claim: Korea’s crash is worse because Korea is more exposed to memory cycle earnings risk

Valuation Framework & Key Multiples (Cheapness vs Cyclicality)

• Valuation lens referenced (conceptually):
  • Stock price is essentially driven by EPS × PER, but in semiconductors the profit cycle distorts outcomes.
• Example “cheapness” multiples cited:
  • Samsung Electronics PER:
    • ~4.8 (this year’s earnings)
    • ~3.4 (next year’s earnings)
  • SK Hynix PER: ~4.3
• Core caution: even low valuation can stay risky because analysts can be wrong when supply and capex cycles reverse quickly.
  • Past pattern described:
    • Undervaluation + continued capex → supply increases → unit prices/margins fall → profits collapse

Past-Cycle Risk: Why “Cheap” Can Stay Wrong

• Samsung example used to show forecast error:
  • Around the 2018 optimism peak, forecasts existed for Samsung’s 2021 operating profit (speaker references numbers such as 53T KRW, plus mentions 63T and 69T for later years).
  • Actual profits are described as later being cut substantially by 2023, with the earlier rebound projections failing.
• Investment takeaway:
  • The market can price very low PER, but the earnings distribution is highly skewed—when it breaks, earnings can fall dramatically.

“When to Sell” / Timing Logic (Explicit Signals)

• Sell-signal logic given:
  1. First sell signal: when upward EPS revisions stop (revision momentum deteriorates)
  2. Stronger conviction sell: when evidence shows supply increasing
• Current condition mentioned (as described):
  • For Samsung / SK Hynix, the speaker claims operating profit estimates for this year and next year haven’t declined “enough,” so consensus may look “fine,” making timing less obvious.
• Recommendation tilt:
  • Earlier “sellers were right” for the initial correction, but whether it collapses further is uncertain.

Long-Term Recovery Expectations (Timeline)

• Recovery example:
  • Investors who bought Samsung around 90,000 KRW in 2021 needed ~4 years to recover principal.
• Bottom call nuance:
  • The speaker implies the memory bottom might be near/early, but recovery speed is unknown.

Hyperscalers, AI Capex, and Cross-Risk From GPUs to Memory

• Key concern: whether AI hyperscalers (cloud/data-center investors) keep funding capex.
• Speaker’s stance:
  • Alphabet, Meta, Microsoft (and Amazon) are viewed as financially able to continue investing.
  • However, there’s still risk that shareholder returns (buybacks/dividends) could be reduced.
• Tickers mentioned (not exhaustive):
  • Meta, Apple, Amazon, Alphabet (Google), Microsoft, Oracle, Nvidia, AMD
  • Indirect reference: TSMC (via “triangular alliance”)
• Risk-management framing:
  • “Uncertainty isn’t solved” until OpenAI stabilizes (framed around profitability and equity outcomes).

Free Cash Flow / Capex Coverage & Vulnerability Example (Oracle)

• Cash-flow logic (conceptual):
  • FCF is clarified as roughly operating cash flow minus capex.
  • If capex > operating cash flow, the firm must rely on:
    • shareholder funding, or
    • borrowing / issuing equity
• Example vulnerability:
  • Oracle is cited as vulnerable due to capex intensity relative to profit cash generation.
• Credit risk proxy:
  • Oracle CDS ~214 bp, earlier referenced as “down to ~200 bp”
  • Spike comparisons referenced around ~2008 and COVID.

OpenAI as the “Weakest Link” in the AI Capital Chain

• Narrative claim:
  • OpenAI is framed as the most vulnerable due to weak cash flow, making IPO/financing less certain.
• Chain description:
  • Hyperscalers build data centers used by “Jaminy/Entropic/OpenAI” (OpenAI is the clearest explicit ticker-level reference).
• Forecast:
  • Uncertainty may persist until OpenAI achieves quarterly profit, possibly via refinancing or equity expansion.

Nvidia as an Analogy for Memory (Timing Trigger in Late August)

• Nvidia as a proxy for timing/volatility:
  • Nvidia cited as roughly -43% from its peak.
  • Example path mentioned:
    • peak around $212
    • low around ~$85
    • recovery attempt near ~$153 (Jan 2025), followed by subsequent declines
• Macro risk transfer logic:
  • If hyperscalers slow, Nvidia GPU demand suffers (GPU market framed as >80% of Nvidia’s relevance/market exposure).
• Timing catalyst:
  • Speaker highlights end of August (earnings/conference call timing) and whether “Rubin” (next-gen platform referenced) delivers and meets expectations.
  • Subtext: end of August could be an inflection point for AI/data-center capex, and thus memory.

Shareholder Return & Dividend/Capital Allocation (SK Hynix Numbers)

• SK Hynix dividend
  • 375 won per share (quarterly cash dividend)
  • Dividend yield described as ~0.02% (very low)
  • Company review referenced:
    • “Actively reviewing plans to offer additional shares” to enhance shareholder value
    • Details finalized in the 3rd quarter (timeline discussed around July/August/September, with an announcement “scheduled” around mid-August)
• Interpretation:
  • Speaker is described as disappointed that the dividend is “too weak” to materially move the stock price.
• Samsung shareholder return
  • Dividend yield referenced as ~8.9% (subtitles described as noisy, but the claim is treated as explicit in the text)
  • Debate noted: buybacks vs dividends as the more meaningful support mechanism.

Volatility Regime Shift (Risk Management Theme)

• Volatility metrics cited:
  • ~33 days in the year with price moves >5% (up and down)
  • Over the last 15 years, many years previously had no such days
• Conclusion:
  • Structurally higher volatility is framed as the “new normal” tied to an “untrodden path” and broken reference points.
• Portfolio implication:
  • Investors must decide whether they can endure volatility and hold quality names (“stocks worth getting stuck with”), since index-like exposure may feel safer but offers less opportunity per stock.

---

Instruments / Tickers / Sectors Mentioned

Indices & markets

• KOSPI, S&P 500, Dow
• Global indices referenced: UK, Germany, France, Spain

Companies / tickers

• Samsung Electronics
• SK Hynix
• Nvidia
• Micron Technology
• Kioxia
• SanDisk
• AMD
• TSMC (indirectly referenced)
• Meta
• Apple
• Alphabet (Google)
• Microsoft
• Amazon
• Oracle
• OpenAI

AI/other entities mentioned (unclear tickers)

• Jeep AI, Moonshot AI, Kimi K3, DeepSeek (mentioned via “Deep Chic/Deep Chic Wave” phrasing)

Macro / financial concepts

• Semiconductors (memory), GPUs
• CDS / credit spreads (Oracle CDS cited in bp)
• Long-term supply contracts
• AI / data centers / cloud capex
• Dividends & share buybacks
• Free cash flow / operating cash flow / capex

---

Methodology Frameworks Explicitly Referenced

• Valuation-to-price lens
  • Stock price ≈ EPS × PER, with adjustments required for cycle realities
• Sell-timing framework
  • Sell when:
    • EPS upward revisions slow/stop
    • or when supply increases becomes evident
• Cash-flow vulnerability logic
  • If capex >> operating cash flow, the firm must rely on:
    • shareholder funding, or
    • borrowing / equity issuance
• Volatility interpretation
  • Current market framed as structurally higher volatility; success depends on endurance and selecting “stocks worth holding”

---

Key Numbers & Recommendations / Cautions (As Stated)

Market / drawdown

• -38% over 27 trading days (Korea; cited low on July 30 vs June peak)
• Historical analogs:
  • ~32% (Dec 1997)
  • ~33% (2008 Lehman period)
  • ~34% (COVID period)

Company drawdowns

• Samsung -42%
• SK Hynix -54%
• Kioxia -64%, SanDisk -56%, Micron -40%

Index concentration

• Samsung + SK Hynix weight ~60% (June 25 cited)

Valuation multiples

• Samsung PER ~4.8 (this year) / ~3.4 (next year) (speaker wording)
• SK Hynix PER ~4.3 (speaker wording)

Recovery timing

• ~4 years for Samsung buyers around 90,000 KRW in 2021 to recover principal

Nvidia analog

• Nvidia: -43%
• Example range: $212 peak → ~$85 low

Volatility

• 33+ days with >5% swings

Dividend / capital return

• SK Hynix: 375 won quarterly, yield ~0.02%
• Samsung: dividend yield ~8.9% (per subtitle claim)

Credit risk

• Oracle CDS ~214 bp, then “down to ~200 bp” (comparison to crises)

---

Explicit Cautions / Uncertainty Statements

• The speaker repeatedly emphasizes uncertainty due to an “untrodden path” with earnings/cycle timing risk.
• Even if valuations look cheap, semiconductor earnings can be distorted by supply/capex dynamics, so “cheap” may not equal safety.
• AI capex risk:
  • Hyperscalers appear resilient, but shareholder returns could be reduced.
  • OpenAI is framed as the weakest link.

---

Disclosures / Disclaimers

• No clear “not financial advice” or formal disclaimer text is visible in the provided subtitles.

---

Presenters / Sources Mentioned

• 김학균 (Kim Hak-kyun)
• Program/Channel: 인포맥스라이브 (Infomax Live)
• Another speaker is referenced as “Director/center director”, but no name is provided.