Ideas
Focus on earnings momentum in tech.
For 2026, focus on companies whose revenue growth exceeds consensus and whose profitability improves simultaneously. The AI value chain and technology sector currently best fit this earnings-momentum filter, and for general investors it is still more efficient to stay concentrated in tech/AI than to diversify into lagging sectors.
Memory is core AI shortage.
AI service traffic is the key driver, and memory is the most traffic-sensitive part of the AI value chain. Demand for HBM, server DRAM, KV-cache/vector-database memory, and NAND/SSD is exceeding expectations, supply is limited to a few companies and countries, and earnings estimates are still rising. Korean memory makers satisfy both revenue growth and margin improvement and should be a core portfolio allocation.
Korean market stays memory-tech led.
The KOSPI is at record highs and the Korean market should remain strong in 2026, led by memory and technology, because Korean memory leaders dominate the AI memory supply chain and are showing upward earnings revisions.
AI optical parts face shortage.
Optical components are among the next AI value-chain segments expected to face shortage conditions in 2026, so related suppliers should be considered for investment.
Alphabet's AI position still strengthens.
Even after Gemini 3.0 excitement marked a short-term price peak, he remains positive on Alphabet and expects Google to strengthen its position in the AI industry, implying further upside over time.
AI PCB shortage favors Korea/US.
AI server architecture changes, including Nvidia's Rubin CPX and prefill/decode separation, add new midplane and infrastructure boards, increasing PCB content, layer count, area, and CCL grade. AI service traffic also boosts server demand. Chinese PCB suppliers face exclusion from the AI value chain, tightening supply and favoring U.S. and Korean PCB suppliers with pricing and margin upside.
800V shift lifts SiC/GaN.
From 2027, AI data centers are expected to adopt Nvidia's 800V DC power architecture, eliminating multiple AC/DC conversions and step-down losses. This requires SiC and GaN power semiconductors rather than silicon-based devices. China's exclusion from AI supply chains should benefit U.S., Korean, and European power semiconductor suppliers, creating new demand and earnings momentum from 2026.
800V DC data center transition.
AI data centers are shifting to an 800V DC power infrastructure from 2027, a concept set by Nvidia and likely followed by hyperscalers. The change removes repeated AC/DC conversions and voltage step-downs, reducing power losses and copper usage, and creates major investment opportunities in data-center power infrastructure.
AI data centers need more ESS.
In the new 800V DC AI data center architecture, the role previously played by battery-based UPS systems shifts to ESS. As AI data centers are built out, ESS demand from this end market can surge.
This 3PRO TV (삼프로TV) video, published January 02, 2026,
features Jeong Hee-seok
discussing XLK, AI value chain, SMH, 000660.KS, 005930.KS, MU, EWY, AI optical components, GOOGL, AI PCB, Korean PCB suppliers, U.S. PCB suppliers, Power semiconductors, SiC/GaN, AI-SECTOR, 800V DC power infrastructure, ESS, AI data center ESS.
9 trade ideas extracted by AI with direction and confidence scoring.
Speakers:
Jeong Hee-seok
· Tickers:
XLK,
AI value chain,
SMH,
000660.KS,
005930.KS,
MU,
EWY,
AI optical components,
GOOGL,
AI PCB,
Korean PCB suppliers,
U.S. PCB suppliers,
Power semiconductors,
SiC/GaN,
AI-SECTOR,
800V DC power infrastructure,
ESS,
AI data center ESS