Please enable JavaScript.
Coggle requires JavaScript to display documents.
WEEK 11 - Power Supply Chain Paradigm Shift in the AI Compute Revolution,…
WEEK 11 - Power Supply Chain Paradigm Shift in the AI Compute Revolution
AI DISCLOSURE PERCENTAGE RATE (%)
WEEK 11 - Career Strategy Architecture: From Competence Foundation to AI-Age Industry Alignment (CSA)
01 - Career Strategy: Facing Career Trade-offs with Strategic Thinking
NOKS
Evolution
Career viewed as a continuous journey of compounding personal development, not a static job.
Trade-off
The decision-making process of explicitly prioritizing 'working for passion' vs. 'working for a living'.
Guiding Principle
Strategy realized as daily task selection and decision criteria — the sum of Σ Tasks/Decisions over time.
Career Strategy
A directed, selective set of choices about your professional development over the next decade or longer, made under resource constraints.
Alignment
Tuning your profession, education/training, and interests to maximum intersection — boosting long-term satisfaction and performance.
Course of Action
Concrete action sequences derived from strategic principles, bridging abstract direction and daily work.
04 - Career Roles: Unpacking the Many Faces of PM and PjM
TAMERA
CORE DISTINCTION
Misconception: PM and PjM are a "senior vs. junior" binary
Reality: It is a functional split of "Strategy vs. Execution."
PM TRACK (Strategy Focus: What to build & why)
Variations of the "P" (Context-dependent)
Product (Tech)
e.g., TSMC
Program (MKT/Sales/CS)
e.g., MediaTek
Procurement (Purchasing)
e.g., Delta
Production (Manufacturing)
e.g., Foxconn
Ladder: Junior PM ➔ Senior PM ➔ Group PM ➔ PM Director ➔ CPO
PjM TRACK (Execution Focus: How & when to deliver)
Common Contexts
Tech / Engineering
Manufacturing
Construction
Consulting
Ladder: PCC (Entry) ➔ Junior PjM ➔ PjM ➔ Senior PjM ➔ PMO Director
STRATEGIC WRAP-UP (For Job Hunters)
Don't fixate on titles; look at Industry, Department, and JD substance.
Map K/S/A to the JD and reverse-engineer capabilities to strengthen.
HOW TO EVALUATE A ROLE
Do not focus on title alone
Look at industry
Look at company
Look at department
Read the JD carefully
Responsibilities
Reporting line
Cross-functional scope
Required K/S/A
03 - Job Market: Industry → Company → Department → Job → Competency
TAMERA
INDUSTRY
Definition: The economic domain of companies offering similar products/services
Position: The top tier of the job market
Industry = Σ Companies
COMPANY
DEPARTMENT
JOB
COMPETENCIES
Components: K·S·A (Knowledge, Skills, Abilities)
Strategy: Matching your K/S/A portfolio against JD competence demands
WRAP-UP
(Career Strategy)
2 more items...
JD = Σ Competencies
Job Description (JD)
Role: Key job-hunting tool / Dialogue document between market & you
Lists: Title, responsibilities, required experience, K/S/A, and salary
Department = Σ Jobs (JD)
The Five Functions
Production
Sales & Marketing
Human Resources
R&D (Responsible for new products, technologies, and processes)
Finance & Accounting
Concept: The company's functional skeleton
Company = Σ Departments
(5 Functions)
05 - AI-Age Alignment: The AES-KY Case — LEV & BBU Twin Engines
PAUL
Core Idea: Strategy × Industry Trends
Aligning strategy with industry trends is key to amplifying career compound returns.
AES-KY is used as a case to show how one company can ride two structural growth waves at the same time: LEV and AI data centers.
AES-KY Twin-Engine Business
AES-KY (6781) is a subsidiary of Simplo (6121), created in 2020.
It channels Simplo’s decades of lithium battery module R&D and manufacturing into two non-IT high-margin markets:
LEV (Light Electric Vehicles) – focus on European e-bike market.
BBU (Battery Backup Unit) – server backup battery modules for global data centers.
BBU Growth in the AI Data Center Wave
Traditional data centers used lead-acid batteries + UPS for backup power.
AI data centers, with power-hungry GPUs and high rack density, must shift to lithium BBUs inside the server cabinet.
This structural shift expanded AES-KY’s BBU share from under 10% of revenue to around 70% by 2025, while LEV (originally over 90%) still grows.
Three Career Lessons from AES-KY
AES-KY’s stock reaching NT$1,200 reflects market recognition of the twin-engine model, with BBU most directly amplified by the AI wave.
The company did not “chase AI”; it kept deepening its core capability (lithium-battery modules) and was ready when the trend arrived → capability is the root, opportunity is the application.
For careers: choosing companies on structural growth tracks lets industry tailwinds boost your individual compound returns; going against headwinds needs much more effort.
02 - Competence Foundation: The Watershed Between Talent and Labor
NOKS
Competence
Knowledge: domain expertise (theory, concepts, industry literacy)
This formula determines whether you are paid hourly as Labor or monthly as Talent.
Labor - where there is a shortage in one of the three (K, S, A)
current minimum hourly wage is approximately NT$196
Talent - all three (K, S, A) are present
average starting salary NT$40000 monthly
Skills: executable capability (programming, languages, communication, data analysis)
Attitude: work ethic, sense of responsibility, and willingness to learn
Golden Investment Window
Ages 18 to 22 (undergraduate) extending to 24 (graduate study)
choosing courses that build knowledge (K)
engaging in projects or internships that sharpen skills (S)
cultivating long-term commitment in the face of challenge (A)
GROUP 1 MEMBERS
NOKS
SETS
TAMERA
5%
10%
0%
JOSHUA
PAUL
TINA
0%
0%
5%
01 - What is a BBU? From Smartphone Power Banks to Data Centre Power Hearts
SETS
Fundamentals
Rack-scale temporal power bank
Replaces UPS; ensures zero interruption
Detects power loss & switches instantly
Mandatory shift driven by "Risk Premium''
Market Sizing
TAM: US$15B by 2030 (Total possible demand)
SAM: US$6.75B (North American premium tier)
SOM: US$850m during 2026-2028 (Winnable
BBU vs UPS
BBU: System-level, longer duration
BBU: Built directly into infrastructure racks
UPS: Centralized, standalone, short duration
UPS: Not suitable for AI racks (100kW+)
Market Drivers
High power racks (5kW to 100kW+)
High downtime cost ($10k+ per min)
High surge frequency (Violent GPU loads)
03 - AES-KY's NT$16B Revenue Leap — Dual-Engine Structure and Margin Expansion
TINA
Record-breaking in 2025
Revenue NT$16.001B
Structural revaluation
BBU's revenue share is climbing rapidly
Revenue Share: 21% → 70%+
Product ASP ↑ 3x
First-half EPS of NT$19.36
Dual-engine business
Core risk Resilience
Primary Engine
AI server BBUs
Secondary Engine
Light Eletric Vihicle, LEVs
Focus on high-end EU/US e-bikes & e-moto
Avoids China’s price war
three-pillar( multi-product long-term) structure
Third growth engine
Telecom backup
Industrial vehicles:AGV/AMR/forklift
2025 Exceptional 34.9% GM, Driven by 3 Factors
ASP jump
Generic BBU ~US$350
AI premium BBU ~US$1,000
High technical barriers
Cross-subsidisation
BBU and LEV act as complementary growth drivers.
lnvestment Reminder
pretty financials ≠ guaranteed price gains
02 - From UPS to HVDC + BBU — A Century-Defining Shift in Power Architecture
SETS
Industry Paradigm Shift
AC UPS to DC-native battery architectures
Direct DC distribution lowers PUE (better efficiency)
Driven by Nvidia high-density racks (GB200/GB300)
Power Usage Effectiveness (PUE)
DC distribution reduces the number of power conversions
This shift is heavily pushed by high-density racks like Nvidia's GB200 and GB300.
Metric used to measure data center energy efficiency.
Power Sidecar
Dedicated supply placed next to compute rack
Integrates: PDU, Power Shelf, BBU, Supercapacitors
Adds buffering/backup without redesigning load
Modes: Grid (Normal), Inject (Peak), Takeover (Outage)
07 - Risks, Moats, and Strategic Thinking for the Next Five Years
PAUL
Company Status & 5-Year Keywords
From: battery module house
-->
To: AI compute power solutions provider (rack-level BBU, future HVDC)
2025 “Triple Breakthrough”
NT$16B revenue
35% gross margin
70%+ BBU purity (revenue share from AI BBU)
Four-Dimensional Moat
Customer: AWS & Meta tief, Microsoft/Google in sampling, high switching costs
Technology: BMS decoding, HVDC mass production H2 2026, patents & know-how
Supply chain: Simplo Group, Samsung SDI & Murata, North America China‑decoupling
Financial: near-share-capital EPS, 35% gross margin → R&D & capacity expansion
Three Long-Term Risks
Customer concentration
Geopolitics: 25% tariff threat, Vietnam < Texas-style local manufacturing
Competitive escalation: Sysgration, Tier 1 awakening, 800V Rubin Ultra → AES-KY must keep evolving
Investment Thinking Framework (PMCS)
Top-down: AI compute wave → power supply chain transformation → rack-level BBUs → AES-KY
Bottom-up: financials (revenue mix, gross margin), orders, technology
Scenarios:
base: 30% BBU growth, smooth HVDC mass production
bull: Microsoft/Google production conversion, margin expansion
bear: 25% tariff, customer migration, escalating competition
Career & Learning Takeaways
Toolbox: BBU, HVDC, AMC framework, dual-engine structure, China+1 strategy, four-dimensional combo attack
6 dimensions: industry trends, company financials, customer structure, technology roadmap, competitive dynamics, geopolitics
Supply chain thinking: geopolitics, customer localisation, upstream alliances
Technical depth = pricing power (ohne BMS SW–HW integration: 15% GM cap)
Choosing the right race: AI halls vs consumer electronics
04 - North American CSP Positioning and the China-Decoupling Supply Chain
TINA
Potential Oligopoly Supplier in NA Power Infra
four hyperscalers
AWS,Amazon Web Services
Meta
Microsoft
Google Cloud
Partnership strategy
Samsung SDI (Korea)
Technical
Japanese and Korean matching AI demand
Murata (Japan)
Geopolitical
Aligning with North American 'China-free'
China+1
Vietnam subsidiary
Responding to 2026 Tariff Threat 25%
:warning:geopolitical discount risk
Compared with Sysgration's Texas Plano fully automated line
100% Made in USA
NoTariff
How to shorten physical delivery distance to North American customers
⚔️ 06. Tier 1 vs Tier 2: The AMC Framework & Disruptors
📊
Market Overview & Stratification
🏢
Tier 1 Integrators
(Delta, Vertiv, Schneider)
🔑 Excel at
turnkey
one-stop solutions
🎯
Tier 2 Specialists
(AES-KY, Dynapack, CATL)
💎 Focus on
premium
high-tech products
🚀
Disruptors
(Sysgration)
🧩 Deploy a 4D combo strategy
🧠
AMC Framework Analysis
(Awareness × Motivation × Capability)
🐘
Tier 1 Giants
👀
Awareness:
High (See the BBU opportunity)
📉
Motivation:
Low (Protecting high-margin vendor lock-in)
⚓
Capability:
High, but hindered by organizational inertia
👑
Tier 2 Leader
(AES-KY)
👀
Awareness:
Strong
📈
Motivation:
High (Eager to break Tier 1 monopoly)
🦾
Capability:
High (Backed by Simplo Group's leverage)
⚠️
Other Tier 2s
(Dynapack, Topower)
📈
Motivation:
High
📉
Capability:
Lower (Squeezed by price compression)
🧨
The Disruptor Case: Sysgration
(5309)
⚡
4D Combo Strategy
🛠️ 1.
Hardware:
LIC hybrid storage Sidecar (microsecond absorption)
💻 2.
Software:
'Universal AI-BMS' decoding proprietary protocols
💰 3.
Capital:
Pegatron alliance ('BMS Inside' model)
🏭 4.
Manufacturing:
Local Texas line + Celxpert cell supply alliance
🔮
Future Battleground & Moats
🔄
The Shift:
From pure hardware to 'power operating system definition rights'
🤖
Next Frontier:
AI Predictive Maintenance (ML fault forecasting)
🏰
AES-KY's Deep Moats:
📦 Simplo Group scale
🤝 Premium cell partnerships (Samsung SDI/Murata)
☁️ Existing production ties (AWS/Meta)
📖
Key Terms Dictionary
🔑
Turnkey:
One-stop integrated solution (cooling, UPS, distribution, BBU)
🔒
Vendor Lock-in:
Binding cells via closed protocols for aftermarket revenue
🛤️
Path Dependency:
Struggling to transition due to past success patterns
🧠
'BMS Inside':
Embedding BMS directly into AI servers (like 'Intel Inside')
💡
Wrap-Up & Core Insights
⚖️
Core Insight:
'Asymmetric Competition'
🐘
Tier 1:
Scale vs. Path-dependency baggage
🎯
Tier 2:
Technical focus vs. Lack of turnkey integration
🚀
Disruptors:
Strategic innovation vs. Limited scale
🎓
Academic Lens:
'AMC × Game Theory × Dynamic Capabilities'
💼
Career Wisdom:
"Don't fight giants where they excel — create value in niches they refuse to serve."
JOSHUA
JOSHUA
🔋 05 - From Component Contractor to AI Power Operating System Definer
🗺️
Overview & Technology Roadmap
🎯
Goal:
Evolve from 'battery supplier' to 'system-level power OS definer'
⚡
Pillar 1:
Deep HVDC positioning (H2 2026 mass production)
🔨
Pillar 2:
Break physical (surge) and protocol (lock-in) pain points
🌐
Pillar 3:
Cross-domain expansion (Cloud BBUs -> Telecom & AGV/AMR)
🤿
Deep Dive: Pain Points & Solutions
🌊
Physical-Layer (Power Surge)
⚠️
Challenge:
AI GPUs cause millisecond violent oscillations
🐢
Limitation:
LFP cells respond >10ms (too slow); micro-cycling kills lifespan
🛠️
Solution:
BMS hardware optimization for short-window large currents
🔮
Future:
Potential LIC (Lithium Ion Capacitor) hybrid buffering
🔒
Protocol-Layer (Vendor Lock-in)
⚠️
Challenge:
Tier 1 giants use closed proprietary protocols
💸
Impact:
'Technical extortion' (~$US38K/month hidden costs for mixing cells)
🛠️
Solution:
Embrace OCP ORv3 open standard
⚔️
Core Weapon:
Intelligent BMS middleware ('mix-and-match freedom')
⏱️
NVIDIA AI Server Timeline
🚀
Driver:
Per-GPU power doubling forces constant redesign
🗓️
2024:
H100 (700W)
🗓️
2025:
GB200/B200 (1000W)
🗓️
2026 H2:
GB300/B300 (1400W) ➡️
AES-KY HVDC Mass Production
🗓️
2027:
Rubin / VR200
🗓️
2028+:
Rubin Ultra
🔀
Diversified Application Expansion
🛡️ Reduces single-CSP dependency & cyclical risk
📡
5G Telecom Backup:
Mass production in 2025 (high-temp endurance)
🏭
Industry 4.0:
AGV & AMR power modules
📖
Key Terms Dictionary
⚡
HVDC:
High Voltage DC Battery (needs high consistency & advanced BMS)
🔋
LFP:
Lithium Iron Phosphate (safe but too slow for AI surges)
🔥
Thermal Runaway:
Uncontrolled heat chain reaction
📉
Micro-cycling:
Short charge-discharge cycles degrading capacity
🧠
BMS:
Battery Management System (core competence for safety/decoding)
🔓
OCP ORv3:
Open standard defining BBU integration
💡
Wrap-Up & Academic Takeaways
🧬
Core Message:
'Software-hardware integration × open standards × diversified applications'
🎯
Strategic Aim:
Embed into NVIDIA standards & break Tier 1 monopolies
🎓
ITBM Lens:
'Technology diffusion × open innovation × platform definition rights'
📈
SMMC Lens:
'Dynamic capabilities × cross-domain extension × moat deepening'