COMPLEXITY AND INTERNATIONALISATION OF INNOVATION Why is Chip Design Moving to Asia? Dieter Ernst, East-West Center, Honolulu, Hawaii WIR 05 Geneva January 05
Internationalization of Innovation Unresolved Issues Drivers Impacts New innovation clusters (Asia) Complexity and mobility (organizational; geographical)
Role of Complexity Established Wisdom Complexity constrains the internationalization of innovation Physical proximity is advantageous for innovative activities that involve highly complex technological knowledge Firms concentrate innovation in home country, to facilitate exchange of complex knowledge Complexity explains why innovation remains an important case of non-globalisation Exception: policy factors
Chip design as a test case value creator & differentiator High and rising cognitive complexity High geographic concentration (US, Japan, Europe) But recently: growing geographical decentralization Massive dispersion to leading Asian electronics exporting countries Why is chip design moving to Asia, despite its extraordinary complexity?
Research Methodology Exploratory, semi-structure interviews (2002 and 2003) 60 companies & 15 research institutions that are doing SoC design (US, Taiwan, Korea, China, Malaysia) Sample contains global and regional carriers of chip design in Asia
Carriers of Asian Chip Design System companies Integrated device manufacturers (IDM) Contract manufacturers (EMS; ODM) fabless chip design houses Chip contract manufacturers ( foundries ) Tool vendors (EDA; design testing) Chip packaging & testing chipless licensors of silicon intellectual properties (SIPs) Design implementation services
Research findings Rapid growth of chip design investment in Asia All interviewed firms are planning to expand such activities Design implementation continues to play dominant role System specification is gaining in importance Substantial progress in complexity of design
Chip Design Flow Chart Market Analysis & Product Planning Define Standards (Upstream) Market Specification System/Application Level Specification Speed, Quality, Flexibility, Cost Asia s Advantage Advantage of Advanced Countries Behavioral Level Design RTL Level Design Gate Level Design Circuit Level Design Physical Level Design Post-Layout Verification Foundry Manufacturing, Packaging and Testing Semiconductor Material, Manufacturing Equipment EDA Software Support IC Design Implementation Flow (Downstream) Process Technology
Why is chip design moving to Asia, despite its extraordinary complexity?
Explanations (1): Pull and Policy Factors Pull ( supply; demand) Policy (incentives, regulations & public goods that affect differences in the cost of conducting design across locations) Both factors explain what attracts design to particular locations But they do not explain what tilts the balance in favor of geographic decentralization
Annual Cost of Employing a Chip Design Engineer* (US-$), 2002 Location Silicon Valley Canada Ireland Taiwan South Korea China India Annual Cost 300,000 150,000 75,000 <60,000 <65,000 28,000 (Shanghai) 24,000 (Suzhou) 30,000 *=including salary, benefits, equipment, office space and other infrastructure Sources: PMC-Sierra, Inc. Burnaby, Canada (for Silicon Valley, Canada, Ireland, India); plus interviews (Taiwan, South Korea, China) Dieter Ernst
China Market World s largest market for telecom equipment (wired & wireless) (test bed for 3G) Ditto for handsets (launch market) Third largest market for semiconductors Sophisticated markets for digital CE (#2) & computers Leading export market for US, Japan, Taiwan and Korea
Explanations (2): Push Factors Dramatic changes in design methodology (SoC) to improve productivity & performance Unexpected result: further increase in design complexity Organizational response: vertical specialization within global design networks Stages of chip design are outsourced to specialized suppliers (dis-integration of design value chain) and relocated across national boundaries (geographic dispersion)
Global Design Network multiple layers EMS IDM SIP Licensors Fabless Tool System Company Design vendors support services Chip assembly Foundry ODM Dieter Ernst
GDN diversity Chinese system company defines system architecture contract manufacturing of electronic equipment (Taiwan) American IDM provides design platform European SIP provider fabless design houses (US; Taiwan) foundries (Taiwan, Singapore and China) chip packaging companies (Taiwan; China) tool vendors for design automation and testing from the US and India design support service providers( various Asian countries)
Drivers of relocation: skill requirements and work organization designer bottleneck : US has failed to train enough engineers for the next generation Global market for design training Asian designers are trained using the latest tools and methodologies Global firms seek to bypass resistance to design automation
Undergraduate Degrees in Engineering, 2002 China 460K India 292K Japan 103K Russia 82K US Korea 73K (40K foreign nationals, India & China) 45K Sources: U.S. Census Bureau; National Center for Educational Statistics; Morgan Stanley
Enabling Factors production - production & SCM -CRM -KIBS - EMS/ turnkey - ODM suppliers ICT-enhanced Information Mgmt communication tools & libraries remote control audio-visual codified> tacit Vertical Specialization disintegration dispersion modular re-use & recombination Innovation Global Flagship Networks re-integration concentrated dispersion diverse & asymmetric governance -design - licensing ( IP trade ) - standard consortia - R&D collaboration - SI: complex systems Knowledge Diffusion tacit> codified Transnational Knowledge Communities professional peer group networks diaspora of skilled migrants reverse brain drain IT mercenaries Dieter Ernst
Conclusions (1) Complexity is no longer an absolute constraint to internationalization of innovation Geographic proximity can be a disadvantage when design requires a large number of designers with diverse capabilities Pull and policy factors explain what attracts design to particular locations Changes in design methodology and organization explain what tilts the balance in favor of geographic decentralization
Conclusions (2) GDNs are long-term arrangements that are shaped by corporate strategies (not arms-length frictionless contracting) But: Limits to Modularity (?) We need to lock up the hardware designers and energyaware software designers with nano-aware architects and CMOS scalers under one roof until they talk to each other. IDMs have an advantage because they have all the people under one roof. Hugo de Man, co-founder of IMEC, Louvain
Host Country Impacts and Policy Responses Conceptual Building-Blocks
Typology of Innovations changed architectural radical Architecture unchanged incremental modular unchanged changed Components Dieter Ernst
Innovative Capabilities R&D is important BUT - equally important are complementary capabilities skills, knowledge & management techniques needed to design, produce, improve & commercialize artefacts artefacts = products, services, equipment & processes
Strategies Definition Capabilities Comments Catchingup enter after growth stage lower-cost producer operational assimilate & improve foreign tech s decreasing returns razor-thin margins Fast- Follower enter early during growth stage quick market response process development prototype development footloose investment weak marketing skills flexible production system cost control where to move to? (paradigm shift) Dieter Ernst
Strategies Definition Capabilities Comments Technology Diversification recombine (mostly known) technologies to create new products & services economics of scope (technology) applied research international knowledge sourcing build on proven capabilities higher margins & limited uncertainty new opportunities (vertical specialization in GFNs) latecomer advantages IPs Technology Leader sets standard during introduction of new products/service basic research pure science defining standards higher margins strong entry deterrents high cost (R&D; regulations) superior IPs lower-cost imitators disruptive technologies Dieter Ernst