엔비디아 젠슨 황 공백은 위기일까? 적절한 조정일까? | 그로쓰힐자산운용 김태홍 대표 & 체슬리 학습부장 복부장 & 박세익 전무 [별주부전 / 26.11.14.토]

Summary

The discussion centers on whether the semiconductor and AI-stock selloff is a temporary correction or a lasting change in the growth cycle. The speakers see scope for a near-term rebound but caution that it would not necessarily signal a durable trend reversal.

Market and investment views

• Samsung Electronics and SK hynix: Kim Tae-hong said he would actively buy both stocks at current depressed levels, expecting a market rebound and potentially a Christmas/New Year rally. He nevertheless characterized semiconductors as a rebound trade, not yet a confirmed trend reversal. A short-term rally could fade, and prices could revisit prior lows.
• Timing: The speakers discussed a possible November bottom, a rebound into December or January, and uncertainty beyond that. One longer-term view put a potential semiconductor-stock bottom around Q3 2027, with new AI-related demand possibly absorbing oversupply from 2028 and a later application-driven rally potentially around 2029. These were forecasts, not certainties.
• Earnings and growth: SK hynix’s HBM-related results were described as strong, with no major earnings collapse expected through Q2 next year. However, the discussion highlighted slowing sequential profit growth. Samsung Electronics’ consensus profit growth was cited as about 30% for Q3, 19% for Q4, and 8.9% from Q4 to Q1.
• KOSPI valuation: A speaker argued that Korean equities merit a higher price-to-book valuation than past cycle lows, citing a level of at least 1.2× P/B. The discussion also referenced a KOSPI level around 5,000–5,200 next year, though parts of the valuation exchange were unclear in the subtitles.
• NVIDIA: The speakers debated whether Jensen Huang’s eventual absence could create a leadership gap. They stressed the importance of technical leadership at technology companies and compared the risk of leadership or governance problems with Intel’s history. One speaker suggested that a 2–3% earnings beat could trigger a 20–30% relief rally, after which investors might reassess whether further upside is justified. The subtitle’s NVIDIA earnings figure is unclear, so it is not repeated as a reliable estimate.
• Other sectors: Potential beneficiaries of a rotation out of semiconductors included pharmaceuticals and biotech, as well as cosmetic-material companies. Secondary batteries were viewed cautiously, though possibly as a temporary refuge. Robotics and space-related companies were mentioned as thematic areas. The speakers did not identify a clearly dominant replacement sector.

AI investment framework

The speakers framed the AI pullback as a test of investment returns and capital efficiency, rather than proof that AI demand has ended:

• Treat NeoCloud restructuring as a possible removal of duplicate orders and speculative demand, while watching whether weaker operators fail to secure customers or achieve profitability.
• Assess data-center investment by GPU utilization and returns, not just by how many GPUs companies buy.
• Consider whether lower infrastructure costs are being reinvested in computing efficiency, cooling, and other capacity improvements.
• Monitor:
  • Cloud-revenue growth
  • GPU utilization
  • Monetization of AI services
  • Hardware procurement prices
  • Cost per unit of computing
  • Big Tech ROI and cash flow

If these measures improve, slower capital-expenditure growth need not be wholly negative. If they do not, the positive “normalization” thesis is weakened.

The discussion also distinguished the AI infrastructure phase—including GPUs, CPUs, and data centers—from a later phase of AI applications, software, and agents. The speakers saw signs of fatigue in the infrastructure trade but did not conclude that the broader AI cycle was over.

Macro and risk factors

• Foreign selling was attributed partly to crowded semiconductor exposure and leveraged positions, as well as rising rates, borrowing costs, and oil prices.
• The speakers noted that high yields on some corporate bonds—citing roughly 7% for blue-chip issuers and 15% for a higher-yield category whose label is unclear in the subtitles—could compete with equities for capital. They compared this with an estimated 7–8% theoretical U.S. equity return based on buybacks, cancellations, and dividends.
• War-related oil and inflation risks, U.S. interest rates, and political uncertainty were identified as possible drivers of volatility. One speaker said that easing geopolitical pressure and lower oil prices by around Q1 next year could support Korean equities, but this was speculative.
• The speakers cautioned that strong company results do not guarantee rising share prices: markets can price in slowing growth, shifting capital flows, or valuation concerns in advance.

Assets and companies mentioned

• Stocks and companies: NVIDIA, Samsung Electronics, SK hynix, SanDisk, Intel, Apple, Tesla, Hyundai Motor, and SpaceX
• Markets and indexes: KOSPI and U.S. equities
• Semiconductor and AI segments: HBM, conventional DRAM, DDR4, DDR5, GPUs, CPUs, data centers, NeoCloud providers, AI software, and AI agents
• Other sectors: Biotech and pharmaceuticals, cosmetic materials, secondary batteries, robotics, space, and autos
• Other instruments and commodities: Corporate bonds and oil

Disclosures and cautions

One presenter explicitly described the AI-market analysis as a personal perspective. The subtitles contain no clear general “not financial advice” disclaimer. Several figures and company references appear garbled by automated transcription; uncertain numbers are identified above.

Presenters and source

The discussion features Kim Tae-hong, CEO of Growth Hill Asset Management; Park Se-ik, Chesley executive; and Bok director, identified in the title as Chesley’s learning director. The program is Byeolju-bujeon on ChesleyTV.