Research
Table of Contents
Research and Projects#
Selected work demonstrating analytical rigour, strategic thinking, and technical capability.
Academic Research#
Geographic Concentration in Global Semiconductor Manufacturing#
MSc Dissertation | University of Exeter | 2025-2026
Research Question: Why has semiconductor manufacturing become MORE geographically concentrated despite USD 95B in global diversification subsidies?
The Concentration Paradox
Despite massive policy interventions (US CHIPS Act, EU Chips Act), market concentration increased rather than decreased:
METRIC 2022 2024 CHANGE
------ ---- ---- ------
Top 3 Producers Market Share 79.2% 83.7% +4.5%
Taiwan Global Export Share 31% 31% Stable
Taiwan Advanced Logic (3-7nm) 90% 90% Dominant
Taiwan Annual Capacity Value -- $91.6B --
Original Analytical Framework: Dual-Risk Model
Developed methodology separating Strategic Risk Index (SRI) from Environmental Vulnerability Index (EVI) to explain manufacturer location decisions:
RISK TYPE CONCENTRATION KEY INSIGHT
--------- ------------- -----------
Strategic Risk 84.8% Firms ACCEPT geopolitical exposure
for ecosystem access
Environmental Risk 39.6% Firms AVOID infrastructure and
natural disaster risk
THE GAP 45.2 pts Geopolitical risk viewed as
acceptable trade-off for
operational efficiency
Interpretation: Manufacturers systematically accept geopolitical exposure (Taiwan Strait tensions) because the daily efficiency gains from established ecosystems outweigh the low-probability risk of conflict.
10-Year Total Cost of Ownership Analysis
Built comprehensive TCO model comparing manufacturing economics across regions:
COST CATEGORY TAIWAN USA EUROPE
------------- ------ --- ------
Initial Investment $20.0B $22.0B $21.5B
(+$2B) (+$1.5B)
10-Year Labor $4.5B $15.75B $13.5B
(3.5x) (3.0x)
Supplier/Coordination $1.8B $5.4B $4.5B
(3.0x) (2.5x)
Yield Loss (Waste) $1.2B $3.6B $3.0B
(3.0x) (2.5x)
─────────────────────────────────────────────────────────
TOTAL 10-YEAR COST $39.5B $61.75B $57.9B
COST PREMIUM Baseline +56% +47%
Why Diversification Policies Fail
Identified three structural reasons why USD 95B in subsidies haven’t reversed concentration:
1. Infrastructure vs. Ecosystems Subsidies target infrastructure where it’s already excellent. They fail to address 30+ years of supplier networks and process knowledge in Taiwan.
2. Cost Structure Misalignment
CHIPS Act Coverage: 25% of initial capital (~$5.5B)
Capital Share of TCO: 35-40%
Result: Subsidies address <10% of total 10-year cost
3. The Expertise Gap
FACTOR TAIWAN USA/EUROPE
------ ------ ----------
Labor Cost Multiple 1.0x 3.5x
Supplier Proximity 200+ within Dispersed
1-hour radius
Manufacturing Yield 98% 90-94%
Yield Cost per 1% -- $60M additional
Engineering Graduates 8,000/year Bottlenecked
Post-Subsidy Economics
Even after applying CHIPS Act grants:
US Facility 10-Year Cost (Pre-Subsidy): $61.75B
CHIPS Act Grant (25% of CapEx): -$5.50B
────────
US Facility 10-Year Cost (Post-Subsidy): $56.25B
Taiwan 10-Year Cost: $39.50B
────────
REMAINING GAP: $16.75B
Conclusion: One-time capital subsidies cannot overcome perpetual operational disadvantages.
Strategic Recommendations
For Policymakers:
- Shift from 5-year funding cycles to 15-25 year ecosystem development timelines
- Focus on “technology-appropriate” diversification: mature nodes (28nm+) to new regions while accepting advanced nodes (3nm) will remain concentrated
For Industry:
- Implement strategic semiconductor reserves (3-6 month buffers)
- Pursue international production-sharing agreements rather than geographic relocation
Technical Implementation:
- Data Sources: UN Comtrade, World Bank, proprietary cost models
- Tools: Python (pandas, numpy), SQL, Tableau
- Analysis: Trade flow analysis, regression modelling, scenario planning
Skills Applied: Economic Analysis, Financial Modelling, Data Visualisation, Policy Analysis, Strategic Research
Strategic Consulting#
Market Entry Strategy | Blackmont Consulting#
Engagement: Strategic analysis for market entry in a USD 400M+ sector
Challenge: Client needed to evaluate market opportunity, competitive landscape, and optimal entry strategy for a fragmented B2B market with multiple incumbent players.
Approach:
-
Market Sizing — Built bottom-up and top-down models to validate market opportunity and identify growth trajectories through 2030
-
Competitive Intelligence — Analysed 4 major players across 8 dimensions including pricing, distribution, product range, and market positioning
-
Customer Segmentation — Developed targeting framework based on customer value potential and acquisition feasibility
-
Financial Modelling — Projected Year 1 revenue scenarios with sensitivity analysis on key assumptions
Deliverables:
- Executive presentation to client leadership
- Strategic roadmap with 8 prioritised recommendations
- Go-to-market playbook with phased implementation plan
Impact:
Market Opportunity Identified USD 1.6B by 2030
Competitive Dimensions Analysed 8
Strategic Recommendations 8
Revenue Scenarios Modelled 3 (Conservative/Base/Aggressive)
Skills Applied: Market Sizing, Competitive Benchmarking, Customer Segmentation, Financial Modelling, Executive Communication
Technical Projects#
AI-Enabled Digital Marketing Platform#
Built with: Python, OpenAI GPT-4 API, pandas, python-docx
Problem: Digital marketing campaign planning is manual, time-intensive, and requires coordination across multiple specialists. Small businesses spend less than 1 hour daily on marketing despite recognising it as a key growth opportunity.
Solution Architecture:
USER INPUT AI PROCESSING OUTPUT
| | |
v v v
+----------+ +----------------------------------+ +-------------+
| Product | | Prompt Engineering Layer | | Strategy |
| Audience |--->| - Strategy Generation |--->| Document |
| Platform | | - Platform-Specific Content | | (Word) |
| Budget | | - Creative Ideation | +-------------+
| Duration | | - Calendar Automation | | Campaign |
+----------+ +----------------------------------+ | Calendar |
| | (Excel) |
v +-------------+
+------------------+
| Quality Checks |
| - Readability |
| - Sentiment |
| - Brand Safety |
+------------------+
Key Features:
- Multi-Platform Support — Facebook, Instagram, LinkedIn, Google Ads
- Automated Strategy Generation — AI-generated marketing plans with audience segmentation and messaging frameworks
- Platform-Specific Content — Headlines, descriptions, and creative concepts tailored to each platform’s best practices
- Quality Assurance — Built-in readability scoring (Flesch-Kincaid) and sentiment analysis (TextBlob)
- Budget Allocation Model — ML-trained weights based on platform engagement data
Technical Implementation:
# Budget allocation using trained model
platform_weights = pickle.load('engagement_weights.pkl')
daily_budget = total_budget / duration
platform_allocation = {
platform: daily_budget * platform_weights[platform]
for platform in selected_platforms
}
Results:
Campaign Setup Time Reduced from hours to minutes
Platforms Supported 4 (Facebook, Instagram, LinkedIn, Google)
Output Quality Score Avg. 53+ Flesch Reading Ease
Sentiment Accuracy Positive/Neutral tone maintained
Skills Applied: Python, API Integration, Prompt Engineering, NLP, Machine Learning
Analytical Toolkit#
Technical Proficiency:
SKILL LEVEL APPLICATION
----- ----- -----------
SQL Advanced Complex queries, data pipelines
Python Advanced Analysis, automation, ML
Excel Advanced Financial modelling, dashboards
Power BI Proficient Interactive reporting
Tableau Proficient Data visualisation
Strategic Frameworks:
- Market Sizing (Top-Down and Bottom-Up)
- Competitive Benchmarking
- Customer Segmentation
- Total Cost of Ownership Modelling
- Scenario Planning
- Policy Analysis
Get In Touch#
Interested in discussing any of these projects or exploring collaboration opportunities?