Written by Konda — Paddy & Water Inc. / Strategy

— Organizational Models, Industry-Specific Strategies, and Advanced Cases —

Centralization and Decentralization ／ Center of Excellence ／ Open Innovation ／ Glocalization

July 2026

This paper systematically explains the state of globally deployed research and development (R&D) organizations, drawing on typologies of organizational models, industry-specific strategies, and case studies of leading companies. Given the growing complexity of technology, the geographic dispersion of talent, and the rise of emerging markets, R&D can no longer be sustained through a home-country-centric approach alone. Centered on advanced cases including Whirlpool, this paper clarifies the design philosophy and management essentials of global R&D.

Table of Contents

・Chapter 1　Why Global R&D?

・Chapter 2　Organizational Models (Typology) of Global R&D

・Chapter 3　Glocalization and Reverse Innovation

・Chapter 4　Open Innovation

・Chapter 5　Global R&D by Industry

・Chapter 6　Detailed Analysis of Advanced Cases

・Chapter 7　Essentials of Global R&D Management

・Chapter 8　The Reality of R&D at Major Global Companies (Company Names, Investment Amounts, and Site Types)

・Chapter 9　Comparison of National R&D Policies

・Chapter 10　Strategic Options and Execution Roadmap for Japanese Companies

・Chapter 11　Conclusion

Chapter 1　Why Global R&D?

1-1　Four Drivers of R&D Globalization

・Increasing technological complexity:A single company or single site can no longer cover all technologies across AI, software, materials, biotechnology, and more. Access to knowledge worldwide is essential

・Geographic dispersion of talent:Top research talent is unevenly distributed across technology clusters around the world (Silicon Valley, Shenzhen, Bangalore, and others)

・Market diversification:Adapting products to the preferences, regulations, and usage environments of diverse markets, including emerging economies, requires local R&D

・Speed and cost:24-hour development (“follow-the-sun”) shortens development time, while regional cost optimization is achieved

1-2　The Fundamental Dilemma of Global R&D — Centralize or Decentralize?

Designing global R&D is an exercise in balancing two opposing forces. “Centralization” excels at economies of scale, technological synergy, and IP management, but is weaker in market adaptation and talent access. “Decentralization” excels at market adaptation and talent acquisition, but tends to invite duplicate investment, coordination costs, and fragmentation of knowledge. Leading companies overcome this trade-off through the “network-type” organizational model described in the next chapter.

Perspective	Centralized (Home-Country-Centered)	Decentralized (Regional Autonomy)
Strengths	Economies of scale, technological synergy, IP management	Market adaptation, local talent, speed
Weaknesses	Market disconnect, inability to leverage local talent	Duplicate investment, coordination costs, fragmentation of knowledge
Suitable areas	Basic research, common platform technology	Applied development, product localization

Chapter 2　Organizational Models (Typology) of Global R&D

Global R&D organizational forms can broadly be organized into four types. Companies design a combination of these according to their technology strategy and market strategy.

Model	Characteristics	Strengths / Weaknesses	Representative Examples
Home-country-centralized	R&D concentrated at home-country headquarters	Efficiency, IP management ◎ / Market adaptation ✕	Traditional manufacturing industries
Regionally decentralized/autonomous	Each region independently carries out R&D	Market adaptation ◎ / Duplication, fragmentation ✕	Unilever, GM
CoE network	Sites divide roles based on their strengths and collaborate	Specialization + collaboration ◎ / Requires coordination	DuPont, 3M, Whirlpool
Technology-focused coordination	Global sites are organized by technology area	Technological synergy ◎ / Requires central control	Microsoft, Novartis

2-1　Center of Excellence (CoE) Networks — The Contemporary Mainstream

The mainstream approach among today’s leading companies is a network model in which each site is given an “area of expertise” (Center of Excellence) and these are connected globally. Each CoE accumulates deep specialization as the global site responsible for a specific technology or product category, and shares its results company-wide. Its distinguishing feature is that it avoids duplication while making the fullest use of regional strengths and talent.

※ The success of a CoE network is determined by “clarity of role division” and “mechanisms for knowledge sharing.” The key is to define which site holds global responsibility for what, and to circulate results company-wide through digital infrastructure such as PLM.

Chapter 3　Glocalization and Reverse Innovation

3-1　Glocalization

Glocalization refers to reconciling the “global × local” duality — leveraging a global technology base while adapting products to the market, culture, and regulations of each region. It requires an ambidextrous approach that addresses the technological, commercial, and cultural dimensions simultaneously. For example, a Korean home appliance manufacturer optimizes products for the local market at its R&D site in India, thereby increasing its market share in that region.

3-2　Reverse Innovation — Innovation Originating in Emerging Markets

Another important trend is “reverse innovation,” in which low-cost, highly functional products and technologies developed locally for emerging markets are conversely rolled out to developed markets. The ingenuity born from operating in highly constrained markets flows back to strengthen competitiveness globally. This phenomenon symbolizes the value of decentralized R&D.

Chapter 4　Open Innovation

Global R&D is shifting away from a self-reliant (closed) approach toward open innovation that actively incorporates external knowledge. As technology grows more complex and moves faster, it is inefficient for a company to develop everything entirely on its own.

・Collaboration with universities and research institutions:Joint research with universities in technology clusters in cutting-edge fields such as biotechnology, materials, and AI

・Startup partnerships and CVC:Investing in and collaborating with promising technologies through corporate venture capital

・M&A and site acquisition:Acquiring companies or sites with advanced technology and incorporating them as global CoEs

・Partner ecosystems:Connecting regional clusters with the global organization through cross-industry open development programs

[Case] Cross-border open innovation by a medical device manufacturer:A medical device manufacturer headquartered in Shanghai acquired an orthopedics business in Memphis, USA, and designated that site as a global innovation hub (CoE) for the orthopedics field. This is a good example of cross-border product innovation achieved by combining the strengths of the home country and the acquired site.

[Case] Open program for a semiconductor cluster:A global compound semiconductor manufacturer has built an open program that nurtures a regional semiconductor cluster while co-creating new markets with partners worldwide, achieving both local cluster formation and global collaboration.

Chapter 5　Global R&D by Industry

The global design of R&D varies significantly by industry characteristics. Industries where market adaptation is critical tend to lean toward decentralization, while industries where core technology is the source of competitiveness tend to lean toward centralization.

Industry	R&D Focus	Characteristics of Global Design	Site Strategy
Consumer goods/appliances	Market adaptation + common technology	Local adaptation via regional CoEs, common technology base	Regional CoEs + global technology centers
Pharmaceuticals/life sciences	Basic research + regulatory response	Bio-cluster collaboration, global clinical operations	Concentrated research + regional regulatory-response sites
Automotive	Software-defined vehicles/CASE response	Emphasis on access to software talent	New technology sites in places like Silicon Valley
Industrial machinery/equipment	Customer proximity, applied development	Application development close to customer sites	Regional application-development sites

5-1　Consumer Goods/Appliances — Regional CoEs + Global Technology Centers

For consumer goods and appliances, competitiveness is shaped by adaptation to regional lifestyles and preferences, while also pursuing economies of scale through common platform technology. As a result, the dominant model is a two-tier structure: “local adaptation through regional CoEs, with core technology consolidated at global technology centers.” Whirlpool, discussed in the next chapter, is a representative example of this approach.

5-2　Pharmaceuticals/Life Sciences — Cluster Collaboration and Regulatory Response

In pharmaceuticals, basic research carries substantial weight, making collaboration with global bio-clusters (such as Boston) essential. At the same time, regional sites are needed to address each country’s regulatory authorities and conduct global clinical trials, requiring a design that balances “concentration of research” with “decentralization of regulatory and clinical functions.”

5-3　Automotive — Proximity to Software Talent

In automotive, software has become central to competitiveness through CASE (Connected, Autonomous, Shared, Electric). Companies are establishing R&D sites in technology hubs such as Silicon Valley to access software and semiconductor talent. The traditional home-country-centered hardware development is evolving to connect with decentralized software development sites.

5-4　Industrial Machinery/Equipment — Application Development Close to Customers

For industrial machinery and equipment, application engineering closely tied to the customer’s usage environment is critical. An effective design consolidates core technology while placing applied-development sites near customers in key markets, enabling rapid response to local requirements.

Chapter 6　Detailed Analysis of Advanced Cases

6-1　Whirlpool — A Model of Regional CoE × Global Technology Center

Whirlpool, one of the world’s largest appliance manufacturers, is a representative example of global R&D’s “regional CoE + global technology center” model. The company operates Centers of Excellence (CoEs) in each of the North American, Latin American, European, and Asian regions, entrusting them with day-to-day regional innovation activities, project execution, and talent development. While these regional CoEs handle adaptation and execution for local markets, specific technologies and product categories are consolidated at global sites.

・Regional CoEs (North America, Latin America, Europe, Asia):Sites responsible for executing day-to-day regional innovation, consulting, and talent development

・Global Technology and Engineering Center (Pune, India / GTEC):One of the largest research sites, responsible for future technologies such as AI, IoT, and Industry 4.0

・Global R&D site for refrigeration and dishwashing (Wrocław, Poland):Develops category-specific technology for all brands as a global responsibility

・Global Food Institute for food technology (Cassinetta, Italy):A CoE for food technology research applied to cooking appliances

・Institutionalizing innovation:Company-wide programs such as the Innovation Challenge embed innovation into the organization as a “culture”

※ Whirlpool’s example offers three key insights: (1) clearly assigning “global responsibility (category/technology)” to each site to avoid duplication; (2) using regional CoEs to drive market adaptation and talent development; and (3) embedding innovation into the organization as an institution and culture rather than relying on individual talent.

6-2　Automotive — Silicon Valley R&D Sites

Major automakers have established R&D sites in Silicon Valley to access the new competitive technologies of software and semiconductors. By connecting their home-country hardware development assets with local software talent, startups, and ecosystems, they secure the development speed required in the CASE era.

6-3　Korean Appliance Makers’ R&D in India — Glocalization in Practice

Korean home appliance manufacturers leverage their R&D sites in India to optimize products for the local market, thereby increasing their market share in the region. This is a practical example of glocalization — tailoring global technology to local usage environments, price points, and preferences.

Chapter 7　Essentials of Global R&D Management

Global R&D does not succeed simply by establishing sites. Success or failure is determined by the skill of management in bringing together dispersed knowledge, avoiding duplication, and circulating results company-wide.

・Governance and role division:Clarify which site holds “global responsibility” for what, to prevent duplicate investment and turf wars

・Knowledge-sharing infrastructure:Share design and technical knowledge across all sites through PLM, digital threads, and similar tools, to avoid reinventing the wheel

・Talent and careers:Use regional CoEs as hubs for talent development, and circulate knowledge through global talent exchange and rotation

・Intellectual property (IP) management:Establish rules for IP ownership, management, and protection in decentralized development

・Digital and AI utilization:Accelerate cross-site collaborative development through generative design, simulation, and digital twins

⚠ The greatest enemy of decentralized R&D is “fragmentation of knowledge (siloing).”Development that remains closed off at each site breeds duplication and inefficiency. Without a design for role division and knowledge sharing enabled by digital infrastructure, global R&D falls into becoming “costly decentralization.”

Chapter 8　The Reality of R&D at Major Global Companies (Company Names, Investment Amounts, and Site Types)

This chapter organizes the reality of global R&D among leading companies across industries, based on public disclosures (annual reports, financial results, and R&D site descriptions). Site deployment can broadly be organized into three types, and competitiveness is determined not by the “absolute amount” of funding but by “where, on what, and how flexibly it can be deployed.”

8-1　Three Types of Site Deployment

Type	Typical Companies	Objective	Site Deployment and Strengths/Weaknesses
Home-country-centralized	TSMC, Tesla, Intel	Protecting core IP, technological consistency, yield management	Centered on a home hub, with overseas sites for customer support/applied development. Technological depth ◎ / local responsiveness can lag
Hub-and-spoke	Roche, AstraZeneca, Microsoft, Samsung, Siemens Energy	Talent, university partnerships, technology scouting, regulatory proximity	Specialized centers across multiple continents. Scouting/recruiting/co-creation ◎ / organizational complexity and duplicate investment
Market-complete decentralization	Volkswagen, Toyota	Market adaptation, regulatory response, customer-driven development	Design, testing, and certification are handled within each region. Development speed ◎ / platform standardization can erode

In recent years, a combination of the second and third types — “concentrating core technology while completing product fit and verification locally” — has become the mainstream approach. What matters is that leading companies establish sites to “go and capture knowledge flows,” rather than choosing research locations based on simple labor-cost differentials.

8-2　Pharmaceuticals/Biotech — Proximity to Knowledge Clusters

・Roche:R&D investment of CHF 12.0 billion (2025). Innovation centers in 17 countries. Strengths include the co-location of Pharma and Diagnostics and dense collaboration across US-Europe clusters. Operational complexity from large-scale decentralization is a weakness

・AstraZeneca:R&D of $14.232 billion (2025). Six strategic R&D centers across the UK, US, Sweden, and China, with a presence in over 50 countries. High-density collaboration with universities, hospitals, and bio-clusters. The heavy investment burden of late-stage development is a weakness

8-3　Semiconductors/Hardware/Software — Core IP Concentration with Peripheral Decentralization

・TSMC:R&D of NT$246.4 billion (2025). Research is divided between “central research” and “fab-site research.” Advanced nodes, 3D stacking, and SIP are handled centrally, while yield improvement is handled on-site. Concentration in Taiwan is a strength but carries geopolitical concentration risk

・Intel:R&D of $13.774 billion (2025). Domestic US R&D and manufacturing are being reinforced in line with the CHIPS Act. Proximity between research and manufacturing is a strength. R&D spending cuts and restructuring in 2025 are a warning sign of weakness

・Samsung:R&D of KRW 34.9981 trillion (2024). A three-tier structure of proximity development, mid-to-long-term research by each division, and future technology research at SAIT. Prioritization is a challenge given the many sites and businesses involved

・Microsoft:R&D of $32.488 billion (FY2025, up 10% year-on-year). Strong connection between research and productization, backed by its own cloud infrastructure. Capital-intensive AI investment is making cross-organizational coordination more difficult

8-4　Automotive/EV — Development Speed Through In-Market Completion

・Toyota:R&D of ¥1.5228 trillion (FY2026). Discloses a regional allocation of roughly 42% for Japan and roughly 29% for North America, with the remainder for Europe/Asia. Maintains a core in Japan while responding to regional needs. Balancing this with platform standardization is a challenge

・Volkswagen:Total R&D costs of approximately €21 billion (2024). Positions its VCTC in Hefei, China, as “the most comprehensive R&D hub outside the home country,” integrating software, hardware, and vehicle testing. A reflection point has been the restructuring needed to close the gap in its response to China

・Tesla:R&D of $6.411 billion (2025, up 41% year-on-year). A strongly centralized, US-centered model that operates AI, software, and hardware as one. Strengthening energy storage at the Shanghai Megafactory. Large-scale AI investment and dependence on the home country are weaknesses

8-5　Chemicals/Energy — Central Platform with Business-Unit Application

・BASF:R&D of €1.176 billion (2025, BASF SE). Group Research handles common themes while business units handle industry-specific applications, a clear division of roles. Prioritization across a broad business portfolio is a challenge

・Siemens Energy:R&D of €1.210 billion (FY2025). Collaborates with regional ecosystems through four innovation centers in Berlin, Orlando, Shenzhen, and Abu Dhabi. Managing commercialization priorities is important

※ The way to read individual cases is to ask “what is being centralized, and what is being localized.” What Japanese companies should compare is not the total R&D spending figure itself, but “the logic of resource allocation and the design of authority.”

Chapter 9　Comparison of National R&D Policies

The focus of policy competition has shifted from subsidies alone to the comprehensive design of “tax systems, visas, university partnerships, demonstration infrastructure, and connections to mass production.” Tax incentives alone create little differentiation; the real differences come from ease of entry for researchers, connections to universities, hospitals, and demonstration facilities, proximity to regulators, and connections through to mass-production ramp-up.

Country/Region	Main Incentives/Tax Systems	Research Infrastructure/Talent	Implications for Japanese Companies
United States	CHIPS: $11 billion for R&D + $39 billion in manufacturing subsidies	NSTC, advanced packaging facilities; attracting researchers via O-1/EB-1 visas	High priority for US sites in semiconductors, advanced manufacturing, and AI infrastructure
EU	Chips Act, Horizon Europe, Chips JU	Chips for Europe, joint research/prototyping	High value in research collaboration, standardization, and regulatory response
Germany	Research allowance (FZulG), strengthened industrial siting	Integration with existing industrial clusters	Well suited to joint research in materials, manufacturing, and industrial machinery
France	CIR at 30% (up to €100 million)	Deeptech and university innovation policy	Attractive integration of tax policy and deeptech policy
Netherlands	WBSO, Innovation Box	English-speaking talent, RVO support	Well suited as a European site for software/hardware applied development
China	Very high R&D expense super-deduction rates, huge market	Universities and industrial clusters	Effective for designing market-complete cells (IP management must be strict)
South Korea	National strategic technology tax system	Concentration of semiconductor and battery industries	Suited to joint research and demonstration in advanced manufacturing
Singapore	Integrated operation of research infrastructure and corporate incentives centered on RIE2025	Research institutions, talent	Well suited as an Asia-wide research and demonstration hub
India	Large-scale fiscal support for semiconductor manufacturing	Scale of talent pool	Suited for software development and market-adaptation sites

※ Claims under France’s CIR grew from €1.8 billion in 2007 to €6.5 billion in 2018, and the Ministry of Finance’s evaluation estimated this would raise GDP by 0.5 points after 15 years. The competitive factor is not a “one-off subsidy,” but the environment itself that allows continuous research, recruitment, experimentation, and commercialization.

Chapter 10　Strategic Options and Execution Roadmap for Japanese Companies

10-1　Three Strategic Options

There are three realistic options for Japanese companies, and rather than being mutually exclusive, a combination of “core concentration + market-complete cells” offers the highest reproducibility.

・① Japan-core-maintained model:Keeps basic research, common platforms, core IP, standardization, and development of key talent in Japan, while placing customer co-creation, regulatory response, demonstration, and manufacturing transfer overseas. Most robust for semiconductors, materials, core software, and advanced medical technology

・② Dual-hub model:In addition to the Japan hub, establishes another advanced research hub in North America or Europe to strengthen recruiting and joint research. Effective for AI, drug discovery, digital manufacturing, and energy

・③ Market-complete cell model:Operates product fit, testing and evaluation, certification, and application development locally in key markets such as China, India, ASEAN, and Europe. Highly effective for automotive, medical devices, and industrial machinery

10-2　Recommended KPIs — Designed at the Portfolio Level

KPI Area	Recommended Metric	Why It Matters
Speed	Days from theme selection to PoC completion; days from PoC to mass-production transfer	The value of a new site is judged by speed
Business contribution	Share of revenue and contribution to operating profit originating from overseas R&D	Prevents research sites from becoming mere “cost centers”
Talent	Fill rate for key talent recruitment, turnover rate, days to obtain a visa	Measures the effectiveness of policy and location
Intellectual property	Number of key patent families, FTO clearance rate	Focuses on quality over quantity
Co-creation	Number of joint projects with universities/startups, commercialization rate	Measures the effectiveness of external connections

10-3　Risk Mitigation — Tiering Technology and the 100-Day Rule

・Three-tier technology structure:Tier 1 = home-country only / Tier 2 = shared with two designated regions / Tier 3 = local discretion permitted. Vary information access rights, the scope of shared source code/formulations/processes, and joint research contract templates by theme

・100-day rule for new sites:Within 100 days of establishing a new site, put in place an “authority matrix, IP rules, KPIs, local partner policy, and crisis-time suspension/alternative procedures”

・In-house governance:Rather than treating legal affairs, export controls, cybersecurity, and data governance as external add-ons to R&D, embed them within the R&D PMO (a lesson borne out by the cases of Intel, TSMC, VW, and Roche)

10-4　Execution Priority — “Redefinition” Before “New Establishment”

The first step should be redefinition, not the establishment of new sites. Many Japanese companies already have overseas sites, but their functions remain frozen in historical circumstances.

・Step 1:Clarify in writing which of “exploration/application/regulatory/customer co-creation/mass-production transfer” each site is responsible for, and eliminate duplicate themes

・Step 2:Cultivate a priority hub in either North America or Europe to lead talent recruitment and joint research

・Step 3:Design China, India, and ASEAN as cell-type sites according to market speed

⚠ Do not get the order wrong:Increasing the number of sites first only causes governance debt to build up first. In the short term, redefine the roles of existing sites; in the medium term, add redundancy through a second region; and in the long term, transition to an integrated operating model spanning R&D, manufacturing, and customer demonstration.

Chapter 11　Conclusion

The essence of global R&D lies in integrating the trade-off between “the efficiency of centralization” and “the adaptability of decentralization” through CoE networks and digital infrastructure. Giving each site clear global responsibility, driving market adaptation and talent development regionally, incorporating external knowledge through open innovation, and connecting all of this globally — advanced cases including Whirlpool concretely demonstrate this design philosophy.

※ Final message: Global R&D is not a contest over “where to place sites,” but over “how to bring knowledge together and circulate it.” It is not the number of sites but the mechanisms for collaboration and knowledge sharing between sites that determine competitive advantage in global innovation.

(This paper was prepared based on general theory of global R&D management — including CoE networks, glocalization, and open innovation — as well as public information on Whirlpool and other companies and literature from 2025-2026. The quantitative figures in the case studies are representative examples based on public information.)