You are on page 1of 25

International Journal of Innovation and Technology Management

Vol. 15, No. 4 (2018) 1850031 (25 pages)


#.c World Scienti¯c Publishing Company
DOI: 10.1142/S0219877018500311

Business Modeling and Public Policy in High-Tech


Industries: Exploratory Evidences from Two Brazilian
Semiconductor Support Programs

Alex da Silva Alves*


Int. J. Innovation Technol. Management Downloaded from www.worldscientific.com

Graduate Program in Management


ao Paulo (ESALQ/USP)
University of S~
a dua Dias, 11, 13418-900 – Piracicaba, SP, Brazil
by UNIVERSITY OF SAO PAULO on 06/25/18. For personal use only.

Av. P
alexds.alves@usp.br

Antonio Jose Junqueira Botelho


Graduate Program in Political Sociology
Candido Mendes University
Rua da Assembleia, 10 sala 709 – Centro
Rio de Janeiro – RJ, CEP: 20.011-901, Brazil
ajjbotelho@iuperj.br

Virgínia Duarte
Association for the Promotion of Excellence in Brazilian Software – SOFTEX
Rua Irm~ a Sera¯na, 863 – 6  andar – Edif{cio Sada Jorge
Centro, Campinas-SP – CEP: 13015-914, Brazil
virginia@softex.br

Received 15 June 2016


Accepted 22 June 2017
Published 7 June 2018

This paper adopts an exploratory analysis based on a multiple case study to investigate the
interplay of Small and Medium Enterprises (SMEs) business modeling strategies and Science,
Technology and Innovation (STI) policies in a mid-income country e®orts to develop a national
semiconductor industry. Speci¯cally, we analyze how 22 Brazilian SMEs supported by two
comprehensive federal programs designed and implemented business models and strategies in
an attempt to, on the one hand, meet the scope and timing of public funding resources and, on
the other hand, develop capabilities to enter the industry's global value chain. Drawing on a
wide body of literature and on the evidence collected, we identi¯ed and categorized the ¯rms'
business models into ¯ve groups: Pure Play IP, Pure Play Design, Fabless, Fabless Plus and
Captive. We then analysed the ¯rms' within the speci¯cities of each group. The paper shows
that, in addition to business models and strategies oriented to meet short-term survival needs,
the ¯nancial bene¯ts for ¯rms were low and unstable, with recurrent and unresolved frictions
between policy-makers, planners and ¯rms.

Keywords: Semiconductors; science and technology policy; innovation ¯nancing; Brazil.

*Corresponding author.

1850031-1
A. da S. Alves, A. J. J. Botelho & V. Duarte

1. Introduction
The semiconductor value chain consists of materials and equipment suppliers,
manufacturing and other service suppliers, semiconductor device vendors, distribu-
tion channels, original equipment manufacturers (OEMs), associated service and
content providers (design houses and fabless ¯rms) and equipment end users. Al-
though the authors recognize the technically broader meaning of semiconductor vis-
a-vis integrated circuit (IC), we use in this paper the terms semiconductor, IC and
chip interchangeably to describe the industry and product category under study.
In the semiconductor value chain, small businesses carry out the initial stages of
design, as this requires comparatively smaller investment than that needed to
operate in the front-end and back-end of manufacturing. On the other hand, design
Int. J. Innovation Technol. Management Downloaded from www.worldscientific.com

companies demand trained and quali¯ed human resources, with a solid background
in electronics engineering and IC design, such as very large-scale integration (VLSI)
by UNIVERSITY OF SAO PAULO on 06/25/18. For personal use only.

and analog/mixed signal (AMS). In this segment, design houses and fabless ¯rms
play a leading role. The ¯rst refers to enterprises engaged in chip design. The second
is an engineering company that specializes in developing solutions in ICs. Fabless
¯rms have products under their own brand and delegate chip-making to a foundry,
which manufactures IC components. A fabless ¯rm is engaged in the concept, design
and development of ICs under its own brand. A design house, on the other hand, acts
only in the design and can also perform IP-licensing services. The manufacture and
assembly of chips can be outsourced to specialized suppliers such as foundries and
assembly ¯rms, respectively.
Semiconductor ¯rms are facing signi¯cant challenges, as the increasing com-
plexity involved in the design and manufacturing processes of ICs, long development
cycles and time-to-market concerns considerably increase their R&D costs, thereby
compromising ¯rms' business models and strategies [Lange et al. (2013)].
In response to them, on the one hand, companies seek to create a more robust
R&D environment to increase productivity with optimized designs in terms of their
complexity, development cycles and budget constraints. In addition, IC ¯rms strive
to obtain skills in embedded software and the management and licensing of intel-
lectual property   IP [McKinsey (2013)].
On the other hand, due to the uncertainty and long development cycles in the
IC industry, public resources play a decisive role for technological infrastructure
composition and in supporting R&D infrastructure [van Marion (2014); Ernst
(2015)]. Therefore, several countries have deployed considerable e®orts to develop
national semiconductor industries by supporting startup creation, incentivizing
investors with public co-¯nancing mechanisms for semiconductor-based projects
and by including semiconductors' design and development in sponsored programs
for target areas such as health, biotechnologies, agriculture and information and
communication technologies [Mays (2013); van Marion (2014); Fuller (2014);
Lojek (2007)].
In light of the structural de¯cit of Brazil's balance of trade in microelectronics,
the Brazilian government has strongly emphasized the consolidation of a national
semiconductor industry [ABDI (2014)]. The country devised several support

1850031-2
Business Modeling and Public Policy in Brazil's Semiconductor Business

mechanisms through institutional partnerships among the Ministry for Science,


Technology and Innovation (MCTI), the National Bank for Economic and Social
Development (BNDES), Brazil's Innovation Agency (FINEP), the Ministry of Fi-
nance, the National Research Council (CNPq) and the Brazilian Agency for Industrial
Development (ABDI). They focus on the training of human resources and on providing
access to ¯nancial resources. More recently, a set of tax incentives for startups and
mature ¯rms willing to invest in semiconductors was also put in place.
Milestones in this context are the National Microelectronics Design Program
(PNM Design) and the IC-Brazil Program (IC Brazil).a The institutional framework
created by these programs produced so far the following results: (i) 22 new IC design
¯rms; (ii) over 600 IC designers trained; (iii) two foundries in the process of opening
and (iv) two ¯rms in the assembly and IC testing phases   HT Micron and SMART
Int. J. Innovation Technol. Management Downloaded from www.worldscientific.com

Modular Technologies   invested in new plants in the country.


by UNIVERSITY OF SAO PAULO on 06/25/18. For personal use only.

Desired outcomes, however, depend on factors beyond the sphere of governmental


action that are equally crucial for the growth strategies of these companies, such as
business models and market entry strategies developed by the companies supported
by these programs. While the quality of the institutional framework and the public
policies for the sector play an important role, the business models engendered by
companies bene¯ting from these policy instruments also play a fundamental role in
their business performance [Lange et al. (2013); Teece (2010)].
There is little knowledge in the available literature concerning how Brazil's IC
design industry and its semiconductor development policy agenda are evolving.
Notwithstanding, attention to the connection between business models and Small
and Medium Enterprise (SME) innovation is neglected in Brazil's policy. Further,
the literature exploring how public policy in Science, Technology and Innovation
(STI) can help business model e®ectiveness in enterprise development is scarce, even
though there is a revealed need for more e®ective STI policies based on an under-
standing of the contexts in which technology design and business models unfold [Villa
and Antonelli (2009); Arend (2013); Lehoux et al. (2014)]. In emerging countries, this
issue appears to be more emblematic, as high-tech SMEs are strongly dependent on the
prevailing STI framework conditions to increase chances of survival in their ¯rst years
of existence [Tsuji and Kuchiki (2010); Alves et al. (2015); Botelho and Alves (2015)].
In Brazil, studies exploring the link between business models and SME innovation
are virtually non-existent. Equally scarce are studies suggesting new policy frame-
works in Brazil that address, on a timely and market-oriented basis, the myriad of
societal implications of wrong or unfeasible business models that SMEs envisage as a
response to particular policy instruments [Botelho and Almeida (2010)]. This
appears as a recurring aspect of STI policies for semiconductors in other mid-income
countries such as China [Ernst (2005); Chu et al. (2014)] and India [Fuller (2014)].
Since governmental calls, grants and seed capital are the most signi¯cant source of
¯nancing for ¯rms in early growth stages, the funding provision for startups and
assimilated high-tech ¯rms in Brazil is heavily dependent on the public support

a The original names of these programs in Portuguese are, respectively, \Programa Nacional de Micro-
eletrônica" and \Programa CI-Brasil".

1850031-3
A. da S. Alves, A. J. J. Botelho & V. Duarte

throughout the ¯rms' development [Alves et al. (2015)]. Therefore, understanding


how STI policies interact with the business model strategies pursued by ¯rms in their
early growth stages becomes central. This paper aims to ¯ll such research gap by
evaluating the business models of IC Design ¯rms supported by IC-Brazil and the
PNM Design programs. The authors believe that innovation policy research in Brazil
and other middle-income countries can thus bene¯t from a better understanding of
the interplay between business models and STI policies. The study asks two ex-
ploratory research questions. First, in the context of an important middle-income
country typically faced with extreme market uncertainties, how are business models
for complex system technologies such as semiconductors shaped in response to
government funding schemes? Second, how do recurring tensions between ¯rms and
Int. J. Innovation Technol. Management Downloaded from www.worldscientific.com

government o±cials inhibit both parties from converging their expectations?


To answer these questions, the authors performed an analysis of the strategies for
by UNIVERSITY OF SAO PAULO on 06/25/18. For personal use only.

competitive entry into the semiconductor market to evaluate the current situation of
the SMEs supported by IC-Brazil and the PNM Design since 2006. From December
2013 to March 2014, the authors interviewed representatives of public institutions,
potential customers, public o±cials in charge of PNM Design and IC-Brazil's par-
ticipating companies and partners. In addition, the authors conducted three work-
shops with ¯rms and stakeholders in di®erent Brazilian cities. Funding for the
research leading to this paper came from ABDI.

2. Review of the Literature


Chip design tasks are innovation-oriented activities in the upstream of the semi-
condutor industry [Lojek (2007)]. Firms in this segment need substantial e®orts to
stay ahead of competition, project new ways to increase competitiveness with better
performance, reduce costs, introduce technological breakthroughs and increase
productivity. The pursuit of sustainability in this industry is crucial and requires
business models that help ¯rms maintain their position in the leading edge. The mix
between the de¯nition of an R&D agenda for the next product development cycles,
intellectual property management strategies, wafer manufacturing and foundry
alignment are business strategies that derive from managers' decision to pursue a
business model in chip design ¯rms, such as design houses and fabless ¯rms. Semi-
conductor ¯rms worldwide seek new ways to stay ahead in this highly uncertain
value chain. The exploration of such strategies, along with the required interplay
with public policy, contributes to a broaded understanding of business models and
strategies for ¯rms in the semiconductor design segment in Brazil and related
middle-income countries. In light of the experiences analyzed in the literature, we
discuss these points in the following.

2.1. The brazilian STI policy framework


Over the last couple of decades, Brazil developed guidelines aimed at strengthening
the national system for STI promotion in a range of segments. Highlights include the
Sectoral Funds (1999); the Industrial, Technological and Foreign Trade Policy
(PITCE), 2003; the Productive Development Plan (PDP), 2008; the Greater Brazil

1850031-4
Business Modeling and Public Policy in Brazil's Semiconductor Business

Plan (PBM), 2011; the National Strategy for Science, Technology and Innovation
(ENCTI) of the MCTI, 2012–2015 and then for 2016–2019. In addition, others were
launched to strengthen innovation and technology development in speci¯c areas,
such as semiconductors, health, agriculture and biotechnologies. We herein mention
the names of the programs in English but employ their original Brazilian Portuguese
acronyms.
The science and technology policy framework in Brazil has changed considerably
with the advent of the Innovation Law in October 2005. The law gave Brazil a
governance mechanism to promote new funding schemes for innovation and scien-
ti¯c research and technology development to strenghen the national production
environment. It also established a legal framework to facilitate the transfer of
Int. J. Innovation Technol. Management Downloaded from www.worldscientific.com

researchers between the science and technology institutions and business ¯rms.
Further, its main mechanisms and guidelines focused on the promotion and funding
by UNIVERSITY OF SAO PAULO on 06/25/18. For personal use only.

of university–industry innovation projects. So far, however, the law has not proved
as e®ective as envisaged. Technological development in some strategic sectors, such
as health, biofuels, and oil & gas   which are at the core of the country's industrial
production  
 have not yet managed to take o® as expected [Pinto and Feldmann
(2016); Botelho and Bastos (2010)].
The Innovation Law lacked detailed regulation on key aspects such as the
treatment of intellectual property resulting from university–industry cooperation
and failed to provide a comprehensive interpretation of the legal status of researchers
from public universities and R&D institutions involved with private enterprises,
including startup creation and management. Thus the ensuing intense debate over
the last decade in Brazil led to a revision of the law in January 2016 with provision of
new incentives to scienti¯c and technological development, research, and innovation.
It is still too early to assess whether law revision will address all those issues blocking
innovation development, as further speci¯c regulations at the federal and state levels
are required. Under the prevailing innovation support framework in Brazil, both
state and federal agencies provide subsidies for R&D, innovation and other related
functions (e.g. grants for business planning, etc.) to universities, research organi-
zations and ¯rms. The available instruments can be reimbursable funding, non-
refundable funding (grants) and ¯scal subsidies, such as tax incentives. Other
instruments include equity investing like seed and venture capital, either from direct
public funds or privately managed funds whole or partially public invested. For
example, the Brazilian Development Bank, BNDES, along with private equity fund
managers, provides seed funding for ¯rms on an equity basis of up to USS 2 million.
The comprehensive policy framework for innovation in the country also counts on
the centrality of the federal government in the procurement of goods and services
and on regional policies for the promotion of entrepreneurship [Botelho and
Pimenta-Bueno (2008); Botelho and Alves (2015); Alves et al. (2017)].
Tax incentives include deductible expenses for equipment purchase, imports of
raw materials and other R&D-related expenditures that can also be exempted from
Value-Added Taxes (VATs) or income taxes. The most comprehensive tax incen-
tives in Brazil are \Lei da Inform atica" (Information Technology Law, or IT Law)
and \Lei do Bem" (Law of Fiscal Bene¯ts). In broad terms, the former aims to foster

1850031-5
A. da S. Alves, A. J. J. Botelho & V. Duarte

digital inclusion and to increase the productivity and competitiveness of Brazilian


hardware manufacturers by cutting back on the tax load levied on the sale of per-
sonal computers. The latter seeks to extend the bene¯ts of the former to any
manufacturing ¯rm willing to carry out R&D [Botelho (2012)].
A related, unresolved issue is that Brazil has two di®erent tax regimes, what
signi¯cantly prevents small and some medium-sized ¯rms from accessing R&D tax
incentives. There are two processes for calculating taxes owed by companies. The
¯rst is the \real pro¯t" regime, whereby the annual net accounting of the net pro¯t
with adjustments is the baseline for levying income taxes at a global rate of 34%. The
second is the `presumed pro¯t' regime, whereby the tax baseline is estimated and
extracted from the company's quarterly income, charging companies an e®ective
Int. J. Innovation Technol. Management Downloaded from www.worldscientific.com

rate around 3.08% for industrial companies and up to 10.88% for service companies.
The presumed tax regime mainly applies to ¯rms with current pro¯ts higher than
by UNIVERSITY OF SAO PAULO on 06/25/18. For personal use only.

what is presumed for business   particularly bene¯cial for small ¯rms  


 so that
the limit of gross revenue to be able to opt for the presumed income is up to R$ 78
million of gross revenue in the preceding calendar year. The tax incentives available
for ¯rms involved in R&D do not apply to ¯rms opting for the presumed tax regime,
which is the case of most SMEs in the country.
Over the last decade, although signi¯cant gains have been achieved in Brazil's
innovation institutional framework, one can still encounter fragmented and dis-
functional policy mechanisms that hamper an e®ective participation of universities
and ¯rms in the creation and transformation processes of knowledge into interna-
tionally competitive innovations [Dutrenit and Arza (2015); Alves et al. (2015);
Del-Vecchio et al. (2014); Cui et al. (2016)]. Areas of improvement are recognized by
policy-makers, as featured in the document of the \ENCTI from 2016 to 2019" of the
MCTI [Ministry for Science Technology and Innovation (2015)], which calls atten-
tion to the high Brazilian dependence on foreign technologies and innovations in key
areas as microelectronics, defense, health and information and communication
technologies.
These issues in°uence the Brazilian development agenda, giving rise to concerns
on what increases the vulnerability of the national innovation system, thereby
bringing to light the need to deepen the country's scienti¯c knowledge base. Despite
the institutional gains achieved when compared the current situation with that of
two decades ago [Kasahara and Botelho (2016)], a challenge that persists remains
overcoming the fragile national production structure. Therefore, semiconductors are
at the core of the STI policy concerns due to its pervasive impacts in other areas.
The Brazilian government has a signi¯cant role in the development of a socially
and economically important production base to increase the technological density of
high-technology segments. Brazilian government actions aim to induce the dynamics
of a dense and complex production chain to embed solutions in chip design,
manufacturing, packaging and testing, with greater emphasis in the development of
local ¯rms in the design chain [ABDI (2014)]. The industry framework pillars are two
federal programs for supporting the growth of a national semiconductor industry.
Established in 2002, the PNM Design program aims to promote the growth of IC
design companies (design houses) in Brazil. The policy focus on design was based on

1850031-6
Business Modeling and Public Policy in Brazil's Semiconductor Business

the fact that design activities o®er lower technological and economic barriers to
entry [ABDI (2014)]. The requisite public investment in design activities is signi¯-
cantly lower than in foundries. Further, the human resources requirements for design
activities are more compatible with the technological infrastructure available in the
country.
The IC-Brazil Program, launched in 2005 as a component of the PNM Design, has
the following objectives: (i) train designers of ICs and (ii) support the creation and
development of Brazilian IC design companies. Participation in the program is open
to graduates in computer sciences and related engineering ¯elds. The IC-Brazil
Program has two training centers (TCs) with a total capacity of 180 students per
year. The running costs of TCs are US$1.4 million per year. In complement to the
Int. J. Innovation Technol. Management Downloaded from www.worldscientific.com

PNM Design and IC-Brazil programs, there are earmarked public funds managed by
FINEP and BNDES for the development of related innovation and business projects
by UNIVERSITY OF SAO PAULO on 06/25/18. For personal use only.

that directly or indirectly make use of semiconductors. Funds are in the form of seed
¯nancing (equity); sunk cost basis and/or on a reimbursable basis at subsidized
rates.

2.2. Business models in semiconductors


Chesbrough [2010, p. 354] argues that \companies commercialize new ideas and
technologies through their business models". While companies may have extensive
investments and processes to explore their new ideas and technologies, business
model innovation is more critical and depends on an extensive set of strategic
components. However, business models and business strategies do not refer to the
same thing. Zott and Amit [2013] comparison of business models with product
market strategies concluded that they complement each other, rather than substi-
tuting one for the other.
Managers are generally asked to ¯rst think in terms of their businesses' value
proposition, current or future competencies required to explore and create the
desired value and the revenue streams and cost structure that will permeate the ¯xed
and intangible assets base by which ¯rms generate and deliver value. These com-
ponents are necessary before managers can develop, adopt or even modify their
strategies [Newth (2012)]. A business model is, therefore, how a company operates in
a broader sense (including its organizational form, activity systems, ecosystem, value
chain, etc.), whereas a business strategy de¯nes how a company will compete [Teece
(2010); Zott and Amit (2013); Newth (2012)]. Further, as Christensen [1997] and
Tidd and Bessant [2013] demonstrate, the same idea or technology taken to market
through similar business models can lead to di®erent business strategies.
The literature highlights two distinct forms of production and labor organization
that culminate in two very di®erent business models in the semiconductor industry.
One is a localized phenomenon centered in SMEs focusing on the provision of R&D-
intensive business services and the other focuses on the organization of ¯rms of
diverse sizes around design networks dispersed globally.
The ¯rst model is known in the literature as Knowledge-Intensive Business Ser-
vices, or KIBS. KIBS share two main characteristics in their business models [EC

1850031-7
A. da S. Alves, A. J. J. Botelho & V. Duarte

(2011)] (i) the solution provided to the customer is the result of a cumulative
learning process that evolves from a deep interaction with the user of the solution
and (ii) is a consulting activity in which the KIBS ¯rm adapts its expertise to the
customer's needs.
The challenges faced by KIBS in the semiconductor industry are not di®erent
from the challenges faced by KIBS ¯rms in other high-tech sectors. Thus KIBS
business models evolve to: (i) reduce the risk in the technology development process
while meeting the needs of customers; (ii) e®ectively plan the organizational struc-
ture and product development cycles and (iii) develop new strategies [Probert et al.
(2013); EC (2011)].
The city of Cambridge in the UK presents one of the worlds' largest concentra-
Int. J. Innovation Technol. Management Downloaded from www.worldscientific.com

tions of KIBS in the semiconductor industry [Kirk and Cotton (2012)]. In the
Cambridge cluster, Probert et al. [2013] observe a mind-set of managers and their
by UNIVERSITY OF SAO PAULO on 06/25/18. For personal use only.

teams that evolves toward collaborative teamwork. For managers, responsibility for
training and managing functional teams is crucial, as well as judgments regarding
the feasibility of proposed technological paths, as the timing for entering and exiting
given markets makes a substantial di®erence in a ¯rm's market orientation and,
therefore, a ¯rm's chosen business model.
KIBS' business models entail project-based structures, which in fact seem
appropriate for business models with value proposition strategies stemming from the
gradual accumulation of knowledge and the development of market understanding.
As a KIBS ¯rm accumulates more projects from a wider range of customers, its
credibility typically grows in the market [Probert et al. (2013)]. And as customer
relationships evolve, ¯rms are enabled to take on projects of greater complexity and
magnitude for their clients.
The second model relates to recent events in the global semiconductor industry
that paved the way for a reorganization of this segment, thereby bringing about
another form of enterprise and production organization. Ernst [2005, 2013] calls
them Global Design Networks (GDNs).
In the mid-1990s, the advance of semiconductor process technologies sparked a
trend of miniaturization, allowing the realization of the System on a Chip (SoC)
concept. SoC allows the integration of an electronic system onto a single chip,
embedding both hardware and software. The advent of SoCs accelerated the spread
of new products, such as consumer goods, mobile phones, automotive applications
and medical equipment. The design of SoC components requires designers with
multiple skills. When chip design groups are concentrated in one place, especially
within local contexts, they can become very in°uential and thus can indirectly un-
dermine productivity gains throughout the complex stages encompassing chip
design. However, geographical dispersion also creates challenges. Geographically
dispersed design requires extremely demanding coordination to manage the multiple
layers and interfaces of chip design. Vertical specialization in a GDN is an attempt to
provide a °exible environment for organizational knowledge-sharing throughout
communities of designers who are not in the same location [Ernst (2005)].
There are several possible con¯gurations for a GDN. A GDN is a network that
incorporates strategic groups of companies operating on three levels. In the ¯rst, the

1850031-8
Business Modeling and Public Policy in Brazil's Semiconductor Business

system company de¯nes the concept and can outsource the rest of the chip devel-
opment process. Examples are Huawei, Apple, Bosch and Boeing. Next, in the
current market con¯guration, there are also Integrated Design Manufacturers
(IDMs), such as IBM, Samsung and Toshiba, who integrate the stages of conception,
design, fabrication, packaging and testing, and o®er their products to the market
under their own brand. Finally, The SoC design can be made by the system com-
pany, the IDM, the fabless ¯rm, or by all these entities working in conjunction.
A GDN is therefore a complex network whose con¯guration has undergone sig-
ni¯cant changes over the last decade, experiencing substantial disintegration within
and geographical dispersion of chip design teams, especially throughout the 1990s,
with a high concentration of chip design companies moving to Asia [Ernst (2005,
Int. J. Innovation Technol. Management Downloaded from www.worldscientific.com

2013)]. The main motivation for geographical relocation of chip design companies to
Asia has been to help manage the challenges faced in closing contracts with global
by UNIVERSITY OF SAO PAULO on 06/25/18. For personal use only.

customers (system companies) which are located elsewhere. Given the pressures
brought by increasing costs and competition in the semiconductor industry, system
companies are becoming increasingly more conservative in adopting new IC-based
solutions into their production processes [McKinsey (2013)]. Consequently, chip
design ¯rms have turned their attention to the Asian market.
Asian system companies were more receptive to new chip designs that helped
them capture market share from global leaders because they are interested in gaining
new market share and expanding operations [Ernst (2013)]. Asian system companies
like Huawei and Samsung are more likely to take on risks and bear part of the chip
design costs that can bring them productivity gains and/or ensure faster placement
of new products into global markets [Mays (2013)]. Consequently, the governance
structure and coordination of GDNs may still experience signi¯cant changes in the
coming years.
However, SMEs not located in Asia may encounter di±culties to enter a GDN
[Ernst and Kim (2002)]. Such ¯rms need to devise business models strictly focused on
price competitiveness, reliability and a solid international penetration strategy,
which demands a portfolio of proven solutions by relevant clients. Firms should
therefore have an existing established presence in the local market and a history of
successful projects to enter the global procurement of large clients. Thus, a primary
step for a fabless or design house to integrate into a GDN is to have a strong local
client base.

3. Research Methods: Rationale, Data Collection and Analysis


This study makes an original empirical contribution to research on semiconductor
policy and SMEs business models in high technology segments in emerging countries.
In Brazil, the last comprehensive study sponsored by the National Bank for Eco-
nomic and Social Development (BNDES) exploring the potential of the country in
semiconductor business dates from over a decade ago [Gutierrez and Leal (2004)].
That study ¯ndings informed the design of the current policy framework in support
of the semiconductor segment. Recently, however, similar comprehensive studies
have been carried out in China and India [Mays (2013); Fuller (2014); Ernst (2013,

1850031-9
A. da S. Alves, A. J. J. Botelho & V. Duarte

2015)], showing the di®erences and idiosyncrasies of semiconductor development


models in these countries.
The scenario of STI policies in Brazil presents challenges that are not di®erent
from other similar middle-income countries. Fuller [2014], for example, shows
that the IC design industries in India and in China have increased their global
market share, activities and technical capabilities in the area, with India more service-
oriented and China more manufacturing-centric. Ernst [2015], on the other hand,
sustains that China's policy challenge in semiconductors is to move from an invest-
ment-driven catching up strategy to an innovation-driven development model.
In Brazil, in spite of institutional advances, there is a persistence of a top-down
policy framework [Kasahara and Botelho (2016)] built in an industrial ecosystem
Int. J. Innovation Technol. Management Downloaded from www.worldscientific.com

highly dependent on the state for subsidized ¯nancing, general infrastructure pro-
vision, technical and scienti¯c work-force training, and tax exempts [Barros (2016);
by UNIVERSITY OF SAO PAULO on 06/25/18. For personal use only.

Brainard and Martinez-Dias (2009)]. The e®orts of some government units to design
modern and responsive policies, particularly in support of startups and other high-
tech segments, have produced internal and external con°icts and tensions, thereby
a®ecting the audience supported by the policy apparatus. In the end, we observe a
repeated policy see-saw between state-centered and more market-oriented policies
which re°ect recurring tensions between two con°icting objectives, similar to a dy-
namic identi¯ed by Ernst [2015] for China. On the one hand, as a latecomer to the
semiconductor industry, Brazil strives to develop and consolidate its industrial pro-
duction and innovation system. On the other hand, in parallel, Brazilian ¯rms are
impatient to enter the industry's value chain to obtain the bene¯ts arising out of global
knowledge sourcing. Therefore, the overall framework conditions presented shape the
context whereby the PNM Design and the IC-Brazil programs took place and evolved.
The IC-Brazil and the PNM Design programs established an ecosystem for par-
ticipants that remains fragmented. Design houses and fabless ¯rms participating in
the programs are based in 8 of Brazil's federation of 23 states. The states of São
Paulo (Southeast region) and Pernambuco (Northeast region) comprise 60% of the
¯rms. Six of the nine companies in the State of São Paulo are in the city Campinas,
showing a strong geographical density.
Generally, ¯rms have a small number of employees. Even prestigious institutions,
such as federal labs and private research facilities that bene¯t from the policy
incentives presented, have a relatively small number of people dedicated to IC
design. Fourteen of the 22 companies are non-pro¯t organizations that include ¯rms
legally linked to public universities and ¯rms within the structure of private and
public research organizations. The latter group has been setup as non-pro¯t orga-
nizations so that clients can bene¯t from the underlying tax-incentive mechanisms
enjoyed by these organizations, as described above. Out of the remaining eight for-
pro¯t organizations, three are subsidiaries of multinationals and two inhabit
university business incubators.
The identi¯cation and analysis of the strategies that engendered the de¯nition of
the business models of participating companies bene¯ted by the IC Brazil Program
was done in three steps. First, we performed a literature review to identify trends
and the contemporary frames of business models and strategies for high-tech SMEs,

1850031-10
Business Modeling and Public Policy in Brazil's Semiconductor Business

particularly in the semiconductor industry. The second consisted of three work-


shopsb with an audience of executives from di®erent sectors that make use of
semiconductor-based solutions, entrepreneurs participating in the IC-Brazil, aca-
demics and government o±cials. In the ¯rst, we discussed the markets and business
models of the companies involved; in the second, we discussed human and techno-
logical resourcing topics and in the third, we addressed current issues in public
policy, institutional frameworks and investments. In the third step, we interviewed
entrepreneurs participating in the IC-Brazil Program, government o±cials and
potential customers of semiconductor-based solutions established in Brazil. Each
interview initiated a process through which an interviewed company or government
representative suggested other government o±cials and potential customers for
Int. J. Innovation Technol. Management Downloaded from www.worldscientific.com

interviews.
The methodology employed had two main objectives: enable the identi¯cation
by UNIVERSITY OF SAO PAULO on 06/25/18. For personal use only.

and analysis of business models and corporate strategies of SMEs participating in the
public support program analyzed and allow in-depth observation of the views held
by government o±cials and representatives of Brazilian companies about potential
customers.
The authors conducted 32 face-to-face interviews with participants' codes to
protect their identities. We asked respondents about their ¯rm's history, existing
and potential clients, their business models and strategies, and their views about the
support provided by policy and on the relationship developed with policy-makers.
Next, we asked them about the market position of known national and international
rivals with similar business models. In parallel, we asked government o±cials about
the evolution of the IC-Brazil Program and the results achieved by the myriad of
support mechanisms for semiconductor ¯rms, including policy issues and the eco-
system for IC design ¯rms. Finally, we addressed three potential customers and
asked them about their perceptions of drivers and barriers to the adoption of pro-
ducts and services from Brazilian design houses and fabless ¯rms, which were then
incorporated into the analysis.
The empirical data were synthesized and all the information were combined for
an integrated analysis. Following the framework developed by Lange et al. [2013], we
summarize in Fig. 1 the emergent data structure utilized in this paper (i.e. employing
terms generated during our interviews and observations).
To avoid con°icting interpretations, data from the interviews were crosschecked
with information from public sources (press, company websites, scienti¯c papers and
consultancy reports), which allowed for the construction and the triangulation of
evidence provided by the multiple case study [Yin (2003)]. The initial coding
resulted in ¯rst-order categories provided by interviewees (Fig. 1). Subsequently, we
constructed mutually exclusive second-order themes and grouped them, which led to
the collapse of the ¯rst-order categories into second-order and, later, third-order
themes that represented more abstract and researcher-induced interpretations. By
following this approach, we developed underlying generalizable constructs and

b The workshops were held, respectively, in the cities of São Paulo, Porto Alegre and Belo Horizonte.

1850031-11
A. da S. Alves, A. J. J. Botelho & V. Duarte

First-order categories (derived from data) Second-order themes Third-order themes Overarching concept
(researcher-induced) (researcher-induced) (researcher-induced)

• Dynamic capabilities and competence brought through a


bundle of public, semi-public and private resources (R&D Strengthening of technology
infrastructure, university-industry cooperation, joint private design efforts
R&D etc.) Business
• Latent value proposition in its market segment Value chain for creating and modelling
• Alignment of internal cost structure with external revenue distributing the offering strategies of
streams to achieve sustainable profits SMEs
• Conflicts between entrepreneurs, investors and users in Managing trade-offs between
terms of perceived differences of the value implied in a value offer and value capture
given IC Design product
Competitive
insertion of
Brazilian IC
Design firms into
the global
semiconductor
• Complement capacitation to build skills in the critical value chain
Incentives for leading
phases of IC Design (VLSI, AMS, testing)
technology development
• Attract MNCs with potential to strengthen local synergies
STI
• Cross-border financing programmes Structure and scope of Semiconductor
Int. J. Innovation Technol. Management Downloaded from www.worldscientific.com

• Public funding for technological infrastructure provision funding programmes Policy


• Signalizing priority areas for STI investments Framework
• Stimulate corporate venture capital and ensure the rule of Strengthen local market for
by UNIVERSITY OF SAO PAULO on 06/25/18. For personal use only.

law, demanding the integration of different policy schemes financing riskier ventures
• Actions and policies for reducing cost of capital and cost of
doing business

Fig. 1. Data structure.

relationships concerning the competitive insertion of Brazilian IC design ¯rms in the


global semiconductor value chain. Further, the framework allowed us to determine
that recurring tensions between the existing policy and the SMEs render the business
modeling process more complex, so that gap-¯lling e®orts by public policy and SMEs
had to be explored, as well [Lange et al. (2013); Lehoux et al. (2014)].
Due to the limited availability of studies linking business models and innovation
with the public policy framework for high-tech SMEs in Brazil, the authors adopted
a \how" and \why" questioning perspective as a theory-building approach [Yin
(2003)]. The approach responds directly to the need to build the foundations of a
theory of the relationship between business models and innovation in Brazilian high-
tech SMEs. Given its exploratory nature, this study avoids generating hypotheses to
test relationships between variables [Edmondson and McManus (2007)]. A multiple
case study approach proved to be the most suitable for the realization of a broad
exploratory analysis and serves as an e±cient approach to the study of organiza-
tional and/or institutional aspects that cannot be captured by quantitative research
methods [Eisenhardt and Graebner (2007)].
Finally, the authors acknowledge that caution is required in making general-
izations based on the results of an exploratory research. However, we expect that the
exploratory nature of this study will not invalidate its merit as an empirical research
e®ort, as this approach is aligned with international best practices in exploratory
research employing surveys or case studies on similar themes [Probert et al. (2013);
Lange et al. (2013); Tongur and Engwall (2014)].

4. Findings and Discussion


In the foregoing ¯ndings and discussion sections, we present the authors' view based
on the evidence collected by the study.

1850031-12
Business Modeling and Public Policy in Brazil's Semiconductor Business

4.1. Findings
Potential clients with su±cient scale to enable the business models of Brazilian IC
design ¯rms are either subsidiaries of multinationals and/or domestic ¯rms engaged
in R&D or New Product Development (NPD) e®orts. Therefore, on the one hand,
the business models of Brazilian IC design ¯rms willing to enter global industry's
value chain need to encompass the challenge of penetrating the international pro-
curement structure of their potential clients. On the other hand, strategies focused
on providing specialized services such as solving clients' R&D and NPD bottlenecks,
which is the case of business models encompassing KIBS, remain di±cult. Because in
Brazil, just a few large ¯rms carry out substantial R&D [Botelho and Alves (2015);
Pinto and Feldmann (2016)] and most resources °owing to the ¯nancing of R&D are
Int. J. Innovation Technol. Management Downloaded from www.worldscientific.com

in the form of subsidies or tax incentives [Kasahara and Botelho (2016)].


From the analysis of interviews with policy-makers and participating companies,
by UNIVERSITY OF SAO PAULO on 06/25/18. For personal use only.

as well as policy documents and the legal framework, we can a±rm that the PNM
Design and IC-Brazil Program focused on consolidating an ecosystem of chip design
¯rms combining elements of both KIBS and GDN models. For KIBS and GDNs,
both present substantial di®erences and challenges (Table 1), thus policies that
support their development may or may not produce the results expected by policy-
makers [Ernst (2013); Rodriguez and Machado (2008)]. The Chinese experience is
very representative of the underlying challenges presented to government o±cials
and entrepreneurs themselves. In the words of Ernst [2005, p. x]: \China's new policy
to upgrade its semiconductor industry through innovation does not represent a
radical break with a deeply embedded statist tradition".

Table 1. Functional aspects of KIBS and GDNs.

Aspects KIBS GDNs

Location High geographical concentration Global, with a strong concentration of


design teams in Asian countries
Firm size R&D-intensive SMEs All sizes
Target segments Multifunction and inter-disciplinary Designers culturally adapted to more
teams rigid labour organization
Business strate- Project-based structure. Revenue model Vertical specialization, with multi-layer
gies varies from hours worked, ¯xed price networks
contract, licensing or royalties
Critical success Management of contracts, IP and the SoC development, complexity in project
factors development cycle of solutions management and time-to-market
Customers Varied sizes (includes startups) Market leaders in electronics equipment
manufacturing and communications
devices
Core business IPs reuse, multidisciplinary team orien- IPs reuse, vertical specialization with
features tation, collaborative teamwork, separation between strategic and
strong client orientation and e±cient routine design functions, large
assessment of the feasibility of cur- teams, skills for managing large and
rent and future technological paths complex teams with multiple layers
and interfaces

1850031-13
A. da S. Alves, A. J. J. Botelho & V. Duarte

As discussed earlier, India and China have taken quite di®erent routes in their
semiconductor industry developmental paths. As stressed by Fuller [2014], Mays
[2013] and Ernst [2013, 2015], their contrasting routes were informed by di®erent
perspectives on development, brought to life by distinct co-evolution of the domestic
institutional setup and the opportunities brought by the global IC design industry.
The Indian route is characterized by the dominance of MNC-owned design opera-
tions and Indian design service ¯rms, whereas the Chinese route is characterized by
the dominance of local fabless and design houses concentrating on creating their own
ICs. China bene¯ted from the massive in°ow of Chinese returnees from the US
and Europe, who established their own IC design services for local IDMs, OEMs and
ODMs, and who could count on the availability of local venture capital funds and
Int. J. Innovation Technol. Management Downloaded from www.worldscientific.com

proximity to foundries. Indian design service ¯rms, by contrast, counted on the


learning curve provided by the presence of MNCs, which trained local engineers who
by UNIVERSITY OF SAO PAULO on 06/25/18. For personal use only.

later became design leads or established their own IC design services with MNCs as
the main client. In this regard, Fuller [2014] mentions that, today, the most im-
portant foundries in Asia have o±ces in India to accelerate contracts and negotia-
tions with Indian IC design ¯rms. In both cases, IC design ¯rms became engaged in a
dense network that later expanded into other markets, with the o®er of standard-
ized, high-scale IC solutions that are highly price-competitive, thereby following a
GDN orientation.
KIBS' business models rely on the R&D agenda of their clients. The more outside
R&D a company procures, the more likely specialized KIBS can work to overcome
bottlenecks that may appear in a client's R&D process (Table 1). Therefore, to avoid
the endless price competition strategy with design ¯rms from Asian countries, IC
¯rms in countries like Sweden, the United Kingdom, the Netherlands and Germany
envisage growth routes focused on the supply of more elaborate and customized
R&D-intensive services that encompass IC design activities. Countries like Ireland,
Spain, Italy and Portugal also pursue similar routes [EC (2011); Kirk and Cotton
(2012); Probert et al. (2013); Rodriguez and Machado (2008)].
As Table 1 shows, the clients for ¯rms engaged on a KIBS-like business model can
be either SMEs or large ¯rms. However, given the Brazilian tax structure, SMEs are
unable to enjoy substantial tax bene¯ts by hiring the services of Brazilian KIBS.
However, large companies operating under the IT Law are required to invest at least
4% of their turnover in R&D with non-pro¯t R&D institutions and 1% on projects
with universities and research labs. This tax policy ends up inhibiting a potential
market for Brazilian KIBS. The sustainability for the business model of Brazilian
KIBS lies largely on their internationalization strategies.
Brazilian public o±cials have made considerable e®orts to devise the best policy
routes to support the growth and competitiveness of Brazilian ¯rms in the concept
and design of ICs, aimed by the two main government programs. Still, for the
government, which shapes and enacts the legal framework and public policies for the
sector's development, devising the best strategies for the international promotion of
Brazilian design houses and fabless ¯rms by inserting them into functional national
and international networks has proved to be a di±cult challenge. The main reason is
that a substantial portion of client procurement is carried on a global scale, involving

1850031-14
Business Modeling and Public Policy in Brazil's Semiconductor Business

high volumes, tight operating margins, high reliability levels, and worldwide ex-
pertise in the ¯rms' respective ¯elds. In addition, the consolidation of an ecosystem
supporting the growth of design houses also depends on the willingness of locally-
established companies to adopt Brazilian IC-based solutions over those of foreign
competitors, which include major ¯rms engaged in substantial R&D and NPD e®orts
as part of their business strategies.
By addressing an empirically observable and relevant innovation management
phenomenon, at the intersections of the STI policy and business modeling literatures
discussed above, we identi¯ed ¯ve types of business models pursued by the partic-
ipating ¯rms, labeled: Pure-Play IP, Pure-Play Design, Fabless, Captive and Fabless
Plus. Table 2 characterizes them in terms of barriers to entry, main business
Int. J. Innovation Technol. Management Downloaded from www.worldscientific.com

interests, existing challenges and opportunities and business risks.


As Table 2 shows, companies in the Pure-Play IP model have a diverse range of
by UNIVERSITY OF SAO PAULO on 06/25/18. For personal use only.

clients. Their primary function is to meet the needs of other players in the semi-
conductor value chain, by licensing IP for design service companies, fabless, foun-
dries and IDMs. They also connect directly to system companies that orchestrate
GDNs.
Not every ¯rm can invest in its own IC-based solutions. Considering the in-
creasing presence of ¯nished IC solutions available on the market, many of these
solutions are geared to meet speci¯c applicationsc because the demand for chip
design is driven by a relatively limited group of customers [Adams et al. (2013)].
Additionally, competition based on price di®erentials tends to favor suppliers from
Asian countries.
The orchestration of a KIBS ¯rm within a Pure-Play Design model can serve as
an alternative to price-based competition with Asian companies. The model adopted
in the Cambridge cluster contributed to concentrating the companies in the local
territory, despite the little inter-¯rm cooperation [EC (2011)]. However, the scat-
tered pro¯le of Brazilian chip design ¯rms along with the limited R&D and NPD
components in the corporate agenda of potential domestic clients may pose some
barriers to their establishment as these ¯rms demand a quali¯ed local ecosystem of
other players [Adams et al. (2013)].
The Fabless model requires substantial investment in the early stage of growth of
¯rms because gains only occur as products are brought to market. Consequently, the
inherent risks for this model are attributable to time-to-market constraints  
 due
to longer development cycles and shorter life cycles – and support from specialized
sources of ¯nancing that understand these critical aspects of the business, such as
venture capital [Lange et al. (2013)]. The compensation for the higher initial product
development costs is achieved through higher margins, which encourage the com-
pany to rely on a strong network of distributors and resellers (Table 2). As the
Fabless model requires a prior knowledge of the market, one can expect a migration
over time of Pure-Play IP and IP Design ¯rms which are more oriented to tailor-
made and customizable solutions into the Fabless model when su±cient economies of
scale and internationalization are achieved.

c Also known by the acronym ASIC or Application-speci¯c Component.

1850031-15
Int. J. Innovation Technol. Management Downloaded from www.worldscientific.com
by UNIVERSITY OF SAO PAULO on 06/25/18. For personal use only.

Table 2. Business models and strategies of companies in the IC-Brazil Program.

Denomination Barriers to entry Business foci Challenges Risks Advantages

Pure-play IP Low Licensing of IP and design Fragmented market; low Productivity dependent on Can bene¯t from compre-
services, diversi¯ed client operational margins; team's skill set hensive and diversi¯ed
portfolio demands large customer ecosystem
base
Pure-play Design Low Design services; demands in- Select foundry partner; con- Productivity dependent on Can bene¯t from the pres-
clusion in global networks vince clients out of the team's skill set ence of a local
to achieve scale local production chain manufacturing equip-
A. da S. Alves, A. J. J. Botelho & V. Duarte

ment and devices


industry
Fabless High Commercialization of pro- Designing own products; Dependence on distribution Can bene¯t from the pres-
ducts with their own achieving economies of channels ence of foundries and

1850031-16
brand scale and manage time- design houses (pure-play
to-market IP)
Captive Not applicable Operated by subsidiaries of Appropriation of the advan- Integration with the ecosys- Enhance learning curve,
multinationals tages related to attract- tem may be low if local especially in nations with
ing multinational institutional setting is a late entry in the semi-
companies not prepared conductor industry
Fabless Plus High Complete solutions, inte- Financial health and HR to High ¯xed costs (laboratories In time can become system
grating hardware and translate technological and HR); mitigation companies
software challenges into industry strategies demand higher
needs operational margins
Business Modeling and Public Policy in Brazil's Semiconductor Business

In the Captive model, the challenge to attract multinational companies is in how


to appropriate bene¯ts to incentivize investment within the host country. The dif-
¯culty becomes greater with low levels of interaction between the subsidiary and the
local production system [Ernst (2013); Thun (2007)]. As Fuller [2014] noted, Indian
design houses were unable to capture as much value as Chinese ¯rms because the
design centers within multinational subsidiaries in India were not independent
pro¯t-and-loss units. This prevented Indian design houses from capturing value from
the chips they designed, thereby dampening revenue capture due to the larger
concentration of MNCs in India following a Captive model. Chinese design ¯rms, on
the other hand, have their own products with several having their own brand and,
therefore, can capture more value than the design service strategy within a Captive
Int. J. Innovation Technol. Management Downloaded from www.worldscientific.com

model that many Indian ¯rms have followed. Fuller [2014] and Chu et al. [2014]
suggest that Chinese design ¯rms seem to be strongly oriented towards the Fabless
by UNIVERSITY OF SAO PAULO on 06/25/18. For personal use only.

business model.
The Fabless Plus model is composed of companies with activities that transcend
the links of the semiconductor value chain by o®ering complete solutions to the end
customer through integration of hardware and software. This model resembles KIBS
employed by some European ¯rms, some of which have been so successful that spun-
o® independent design ¯rms, which in turn migrated to Asia while maintaining the
parent KIBS company in the host country [EC (2011); Probert et al. (2013);
Rodriguez and Machado (2008)]. In addition to professionals with design expertise,
the Fabless Plus business model requires expensive programmers and system engi-
neers. In respect to the interaction with the semiconductor ecosystem, the rela-
tionships of Fabless Plus ¯rms tend to be stronger with Pure-Play IPs and foundries.
In fact, most complete solutions of Fabless Plus ¯rms presented by the interviewees
can inhibit market performance of Pure-Play Design and Fabless ¯rms in the in-
cipient Brazilian market. Therefore, business relationships between Fabless Plus,
Pure-Play Design and Fabless ¯rms, according to the interviewees, may prove
unnecessary and at times may be in con°ict.
The business model for Fabless Plus companies is in high risk (Table 2), since the
maintenance of a Fabless Plus structure can be expensive. In contrast, operational
margins can be higher because these companies market more valuable products.
Fabless Plus ¯rms can create a relatively wide scope of R&D intensive products and
solutions for the market by mastering a wider range of skills and producing advanced
laboratories with a myriad of resources. The solutions that Fabless Plus companies
o®er can be complete products or even be part of someone else's solution. The
activities of Fabless Plus companies could bene¯t greatly from a local presence of end
customers who demand innovative applied-electronic solutions. The interaction with
international networks can occur with the support of distributors, dealers of
machinery and equipment, as well as investors.

4.2. Discussion
In Table 3, we present an assessment of IC Brazil participating companies based on
the convergence between the ¯rm and its envisaged business model. Each ¯rm's

1850031-17
A. da S. Alves, A. J. J. Botelho & V. Duarte

Table 3. Evaluation of the strategies of design houses of the IC-Brazil Program.

Level of adequacy to the business model


envisaged by entrepreneurs Characteristics of the business model

Public policy dependence

Uncertainties to business
International orientation
Fabless Plus Private
Fabless Plus Public
Pure-Play Design
Pure-Play IP

consolidation
Captive
Fabless
Firm
Int. J. Innovation Technol. Management Downloaded from www.worldscientific.com

Firm 1 1 High Low High


Firm 2 1 High Low High
Firm 3 3 High Low Low
by UNIVERSITY OF SAO PAULO on 06/25/18. For personal use only.

Firm 4 2 High Low High


Firm 5 3 High Low Low
Firm 6 1 High Low High
Firm 7 4 Average Average Average
Firm 8 1 High Low High
Firm 9 1 High Low High
Firm 10 1 High Low High
Firm 11 5 Low High Low
Firm 12 3 Average Low Average
Firm 13 4 High Low Low
Firm 14 4 High Average Average
Firm 15 1 High Low High
Firm 16 1 High Low High
Firm 17 1 High Low High
Firm 18 3 Average Average High
Firm 19 3 Average Low High
Firm 20 2 High Low High
Firm 21 1 High Low High
Firm 22 3 High Low High

Notes: 1  very incipient; 2 


 incipient; 3 
 under development; 4 
 mature; 5 –
consolidated.
Source: Author developed based on interviews.

business model categorization is based on a score ranging from 1 to 5. We built this


categorical scale upon considerations elicited from the analysis of Table 2, thereby
focusing on the following aspects: survival capacity in the market without govern-
ment support; international penetration; uncertainties relating to business consoli-
dation; company history and the pro¯le and motivation of the partners and market
positioning strategies. \Very incipient (1)" means that the ¯rm's current business
model, in respect to the aspects analyzed, is still far from the envisaged one.
\Incipient (2)" signi¯es that the ¯rm's current business model reveals some of these
aspects. \Under development (3)" means that the ¯rm's current business model
shows some aspects towards the envisaged one. \Mature (4)" state means that the
¯rm's current business model presents all aspects, even though they may not all be
fully sustainable. And ¯nally, \Consolidated (5)" states that the ¯rm's current

1850031-18
Business Modeling and Public Policy in Brazil's Semiconductor Business

business model is fully developed and sustainable in the current competitive context.
For analytical purpose, companies included in the Fabless Plus model were sepa-
rated into public and private. The information presented in Table 3 help us un-
derstand the resulting adequacy to the business model envisaged by each ¯rm along
with other important aspects such as dependence from public funding, interna-
tionalization e®orts and level of uncertainties evidenced throughout their growth
paths. We then discuss the results achieved and the challenges ahead of these ¯rms.
Although part of the PNM Design's policy e®orts has shed light on and
strengthened IC design companies, we have observed that the least e®ective business
models were those that remained attached to the Pure-Play Design model.
Two companies reviewed experienced success by combining the supply of chip
Int. J. Innovation Technol. Management Downloaded from www.worldscientific.com

design services with the delivery of IP services by adopting the Pure-Play IP model.
The Pure-Play Design strategy proved to be the least sustainable among the
by UNIVERSITY OF SAO PAULO on 06/25/18. For personal use only.

business models. Firms that remained committed to this model were those in public
universities. A strong academic bias and thus a reduced capability to deal with the
market contributed to their di±culties in reviewing and adapting their strategies.
This helps to explain the shutdown of some and the incipient results achieved by
others. Moreover, legal barriers to the commercialization of products by non-pro¯t
institutions have restricted some ¯rms' e®orts to adopt a more market-oriented
approach.
While university-based design houses in the Pure-Play Design category failed,
none of the other for-pro¯t companies in that model kept the focus on this model.
One of these, for example, viewed the Fabless model to develop more recurrent
resources, because contracts for design services are not o®ered with the frequency
required to stay in the market.
The race for market penetration has just begun and uncertainties remain con-
cerning the success of ¯rms that specialize in the Fabless model. Except one, ¯rms
have not yet started to bring products to the market. Moreover, one of the challenges
reported by companies in this model is the di±culty entering amenable markets. In
this regard, the Brazilian government requirements for strategic projects, such as
defense, IT, medical equipment, and energy could contribute to the success of
companies that adopt the Fabless model.
Firms in the Fabless Plus Model are private, non-pro¯t research organizations.
Their clients enjoy ¯scal incentives, such as the Law of Fiscal Bene¯ts and the IT
Law, to contract R&D services from them. Despite such advantage vis- a -vis other
PNM participating ¯rms, Fabless Plus ¯rms face trouble expanding their operations
through other quali¯ed sources of ¯nancing, such as Venture Capitalists (VCs) and
angel investors, so that a sustained expansion is still an unresolved issue that
restricts their long-term competitive growth.
Adopting a Fabless Plus model is di±cult for small companies because it requires
a broad set of ¯nancial resources, technological infrastructure and expertise. The
relative self-su±ciency of companies operating along the value chain of the semi-
conductor industry does not su±ciently strengthen the ecosystem for Brazilian
design houses and fabless ¯rms. The current ecosystem is fragile in Brazil and
Fabless Plus ¯rms have legal advantages and enjoy special government allowances.

1850031-19
A. da S. Alves, A. J. J. Botelho & V. Duarte

Fabless Plus ¯rms compete with small for-pro¯t companies in a situation that tends
to give them advantage due to their size, their brand and their ability to respond to
their clients' needs in a comprehensive manner given their pre-existing subsidized
R&D infrastructure. In addition, as design services for ICs have become an integral
part of a supply structure that encompasses larger solutions for an increasingly
demanding market, companies operating under the Fabless Plus model also tend to
be favored by large government projects due to usually encompassing and complex
nature of government demand in certain public areas. Examples are defense tech-
nologies for a continental country, faster diagnostic equipment for tropical diseases
and technologies for remote real-time monitoring and control of deforestation in the
Amazon rainforest.
Int. J. Innovation Technol. Management Downloaded from www.worldscientific.com

The incipient Brazilian high-tech industry also faces search barriers in acquiring
industrial clients that wish to invest in Brazilian semiconductor solutions. The
by UNIVERSITY OF SAO PAULO on 06/25/18. For personal use only.

companies interviewed highlighted the di±culty ¯nding domestic customers. This


situation led many of IC-Brazil's participating ¯rms to abandon the Pure-Play
Design model for other strategies. However, these companies were neither able to
mobilize the domestic market nor to look beyond it. No ¯rm has yet closed a ¯rm
deal with an external customer or has entered a global network. Many interviewed
¯rms said the resources available for client prospecting were insu±cient to ade-
quately assess opportunities in foreign markets.
Likewise, even ¯rms with sta® dedicated to the evaluation of markets and pro-
specting for customers possess limited capability to expand their business prospects
outside the country, which is often expensive and dependent on a dense network of
contacts. Factors such as an inadequate track record and the limited experience of the
entrepreneurs were also aspects identi¯ed as obstacles to accessing global networks.
An important lesson learned by the ¯rms supported by IC-Brazil Program and
PNM Design is that the Brazilian market is not yet ripen for contracting semicon-
ductor design services, nor does there seem to be interest in the acquisition of spare
chips from Brazilian suppliers. This occurs even though there is funding for joint
(supplier þ client) IC development projects.
Companies that were capable of identifying potential customers frequently
sought subsidies or ¯nancial support from governments and private partners to
reduce development and ¯nal product costs. Interviewed companies often mentioned
that their request for project-funding were often rejected due to the lack of under-
standing by public institutions of the diverse organizational and economic nature of
semiconductor-related projects.
Reports by interviewees portray situations in which the design house and the
customer declined to pursue a project already granted by a public funding agency in
the initial stage due to the long evaluation time by a public funding agency. These
issues produced recurrent tensions between participating ¯rms and policy-makers, as
noted in the interviews and observed in the workshops. A similar situation is
reported by Ernst [2015] in his account of China's semiconductor industry unre-
solved friction between state and market, where policy-makers and planners pre-
scribe desired outcomes but fail to take into account the needs of industry for
technology sourcing.

1850031-20
Business Modeling and Public Policy in Brazil's Semiconductor Business

Overall, the ¯rms' market de¯nition occurred without a formal business plan.
Partly because of this lack of strategic preparation from which an appropriate
business model could derive, the search for customers was limited to the region where
the company was located.

5. Conclusions and Implications


The discussion above shows that, in general, the economic and ¯nancial bene¯ts
from the strategies adopted by the companies interviewed were low and discontin-
uous, with a prevalence of business models oriented to survival needs in the face of
adverse market dynamics.
Int. J. Innovation Technol. Management Downloaded from www.worldscientific.com

The slow and intermittent pace of the government program development and
implementation and of the demand that would be fostered by other government
by UNIVERSITY OF SAO PAULO on 06/25/18. For personal use only.

mechanisms contributed to the somewhat inexpressive economic results attained by


participating design houses and fabless ¯rms. A series of factors ranging from the
complexity involved in the scaling up production to governance issues in the various
support institutions and public bodies involved, to repeated budgetary constraints,
seem to have played a key role in engendering the obstacles ¯rms encountered in
¯nancing project development cycles and expanding operations.
The advantages experienced by non-pro¯t ¯rms put for-pro¯t companies in dis-
advantage. For-pro¯t SMEs that had opted for the Pure-Play Design model are
leaving the business or seeking other ways to continue in the market. One alternative
pursued by some of these ¯rms is to transition to a Pure-Play IP or a Fabless
strategy, which is nevertheless precarious given that none of these companies have
yet managed to place products on the market. The Fabless Plus model is not
accessible to for-pro¯t design houses and fabless ¯rms, as Brazilian Fabless Plus
companies are organized as non-pro¯ts so that their customers can reap tax bene¯ts.
This aspect requires policy reconsideration, so that the PNM Design and IC-Brazil
instruments can bene¯t all types of ¯rms regardless.
Government funding and promotion programs in the IC-Brazil and PNM fra-
meworks were not su±cient to support a sustained generation of ¯rms' revenues or
to leverage their business models. But at least they helped extend their lifecycle
which, in turn, allowed them to retain quali¯ed teams and accumulate expertise.
Structured ecosystems for supporting innovation in Brazil are relatively recent.
Therefore, one cannot expect that stimulus for high-tech SMEs will succeed over-
night in Brazil. The set of policies envisaged by the PNM and IC-Brazil programs
represent a valid e®ort towards a more e®ective innovation support system. Al-
though Brazilian IC design ¯rms still lag in terms of innovation and competitiveness
in the global semiconductor industry, these policy e®orts constitute an important
initiative to incorporate innovation into the organizational culture of Brazilian
companies, particularly small ones.
The ¯ndings provide valuable policy lessons that point out to the need for sectors-
speci¯c innovation promotion policies. On the one hand, the study showed that new
policies need to encompass issues emanating from the global industry dynamics along

1850031-21
A. da S. Alves, A. J. J. Botelho & V. Duarte

with the target ¯rms' diverse organizational natures and staged growth ¯nancing
requirements. On the other hand, that policies need to attend to time-sensitive
competitive demand aspects and, related and more importantly, to the target ¯rms'
need to evolve their business model as they launch into growth in the context of
shifting competitive patterns. KIBS' business model's experiences appear to o®er
relevant clues in this direction.
The nature of this study does not allow for broad generalizations, as it is
restricted to a review of relevant literature and based on a multiple case study
approach. However, given its exploratory nature, we believe the results outlined may
o®er support to guide future studies. In middle-income countries, for example,
comparisons of the connection between semiconductor ¯rms' business modeling
Int. J. Innovation Technol. Management Downloaded from www.worldscientific.com

e®orts and the policy framework sustaining their ecosystems could open new policy
horizons that could reduce the recurrent frictions among policy-makers, planners
by UNIVERSITY OF SAO PAULO on 06/25/18. For personal use only.

and market segments. In addition, on the one hand, in Brazil and in assimilated
economies, we believe that this study can assist in the formulation of new hypotheses
about the e®ectiveness of STI policies aimed at the consolidation of competitive
ecosystems for high-tech SMEs. On the other hand, the business model's categories
typology developed can provide a framework for high-tech entrepreneurs and policy-
makers alike to assess the adequacy of their technology forecasting strategies to the
harsh realities of their uncertain domestic and global markets.

Acknowledgments
The authors would like to express their gratitude to the Brazilian Industrial
Development Agency (ABDI) for the ¯nancial support for the realization of the
research project leading to this paper.

References
ABDI. (2014). Avaliacao~ das Estrategias de Negocios
 das Empresas de Projeto de Circuitos
Integrados do Programa CI-Brasil (Evaluation of the Business Strategies of Chip Design
Firms in the IC Brazil Programme). Ag^ encia Brasileira de Desenvolvimento Industrial,
Bras{lia-DF.
Adams, P., Fontana, R. and Malerba, F. (2013). The magnitude of innovation by demand in
a sectoral system: The role of industrial users in semiconductors. Research Policy, 42, 1:
1–14. doi: 10.1016/j.respol.2012.05.011.
Alves, A. S., Botelho, J. J. and Mendes, L. (2017). An exploratory assessment of the gaps for
health innovation in Brazil: Challenges and a proposed research agenda. Revista de
Administracao~ e Inovacao,
~ 14, 2: 98–108. doi:https://doi.org/10.1016/j.rai.2017.03.001.
Alves, A. D., Quelhas, O. L., da Silva, M. H. and Lameira, V. D. (2015). On the role of
university in the promotion of innovation: Exploratory evidencies from a university-
industry cooperation experience in Brazil. International Journal of Innovation and
Learning, 17, 1: 1–18. doi: 10.1504/IJIL.2015.066061.
Arend, R. (2013). The business model: Present and future   Beyond a Skeumorph. Strategic
Organization, 11, 4: 390–402. doi: 10.1177/1476127013499636.
Barros, A. R. (2016). Roots of Brazilian Relative Economic Backwardness. Academic Press,
Cambridge, Massachusetts.

1850031-22
Business Modeling and Public Policy in Brazil's Semiconductor Business

Botelho, A. J. (2012). Erawatch Country Reports 2011: Brazil. Report for European Union
ERAWATCH International Project, EC/DGRI and JRC-IPTS and ERAWATCH
Network. Available at https://rio.jrc.ec.europa.eu/en/library/erawatch-country-report.
Botelho, A. J. and Alves, A. S. (2015). A study of measures to encourage the translation of
public sector research for economic and social bene¯t in Brazil. Report for Securing
Australia's Future Project \Translating research for economic and social bene¯t: Country
comparisons" on behalf of the Australian Council of Learned Academies. Available at
acola.org.au/PDF/SAF09/2%20Brazil.pdf.
Botelho, A. J. and Bastos, G. M. (2010). Innovation as a driver for building an oil & gas
industrial in Rio de, From Agglomeration to Innovation: Upgrading Industrial Cluster in
Emerging Economies, Janeiro, Brazil. eds. M. Tsuji and A. Kuchiki. Palgrave Macmilan,
Chippenham, pp. 326–356.
Botelho, A. J. and Pimenta-Bueno, J. A. (2008). Financing university–industry relations:
Int. J. Innovation Technol. Management Downloaded from www.worldscientific.com

University booster or innovation driver. University and Development in Latin America:


Successful Experiences of Research Centers, ed. S. Schwartzman. Sense Publishers,
by UNIVERSITY OF SAO PAULO on 06/25/18. For personal use only.

Rotterdam, pp. 43–72.


Botelho, J. A. and Almeida, M. (2010). Overcoming institutional shortcomings for academic
spion-o® policies in Brazil. International Journal of Tecnology Management and Sus-
tainable Development, 9, 3: 175–193. doi: 10.1386/tmsd.9.3.175 1.
Brainard, L. and Martinez-Dias, L. (eds.) (2009). Brazil as an Economic Superpower? Un-
derstanding Brazil's Changing Role in the Global Economy. Brookings Institution Press,
Washington, DC.
Chesbrough, H. (2010). Business model innovation: Opportunities and barriers. Long Range
Planning, 43, 2–3: 354–363. doi: 10.1016/j.lrp.2009.07.010.
Christensen, C. M. (1997). The Innovator's Dilemma. Harvard Business School Press, Cam-
bridge, MA.
Chu, M.-T., Khosla, R. and Chai, K.-H. (2014). A cluster analysis of IC design industry.
International Journal of Innovation and Technology Management, 11, 2: 1–26. doi:
10.1142/S0219877014500035.
Cui, Y., Jiao, J. and Jiao, H. (2016). Technological innovation in Brazil, Russia, India, China,
and South Africa (BRICS): An organizational ecology perspective. Technological Fore-
casting & Social Change, 107: 28–36. doi: 10.1016/j.techfore.2016.02.001.
Del-Vecchio, R., Brito, J. and de Oliveira, B. F. (2014). Patterns of university–industry
interactions in Brazil: An exploratory analysis using the instrumental of graph theory.
Quality & Quantity: International Journal of Methodology, 48, 4: 1867–1892. doi: 10.1007/
s11135-013-9857-x.
Dutrenit, G. and Arza, V. (2015). Features of interactions between public research organi-
zations and industry in Latin America: The perspective of researchers and ¯rms. Devel-
oping National Systems of Innovation: University–Industry Interactions in the Global
South, eds. E. Albuquerque, W. Suzigan, G. Kruss and K. Lee. Edward Elgar Publishing,
Cheltenham. Availabel at http://idl-bnc.idrc.ca/dspace/bitstream/10625/53627/1/IDL-
53627.pdf.
EC. (2011). Knowledge Intensive Business Services in Europe. European Commission,
Brussels.
Edmondson, A. C. and McManus, S. E. (2007). Methodologial ¯t in management ¯eld
research. Academy of Management Review, 32, 4: 1155–1179. doi: 10.5465/AMR.2007.
26586086.
Eisenhardt, K. M. and Graebner, M. E. (2007). Theory building from cases: Opportunities and
challenges. Academy of Management Journal, 50, 1: 25–32. doi: 10.5465/AMJ.2007.
24160888.
Ernst, D. (2005). Complexity and internationalization of innovation: Why is chip design
moving to Asia. International Journal of Innovation Management, 9, 1: 47–73. doi:
10.1142/S1363919605001186.

1850031-23
A. da S. Alves, A. J. J. Botelho & V. Duarte

Ernst, D. (2013). Why does China Still Play a Limited Role in Semiconductor Innovation?
East-West Center, Honolulu.
Ernst, D. (2015). From Catching Up to Forging Ahead: China's Policies for Semiconductors.
East-West Center, Honolulu.
Ernst, D. and Kim, L. (2002). Global production networks, knowledge di®usion and local
capability formation. Research Policy, 31, 8–9: 1417–1429. doi: 10.1016/S0048-7333(02)
00072-0.
Fuller, D. (2010). Networks and nations: The interplay of transnational networks and
domestic institutions in China's chip design industry. International Journal of Technology
Management, 51, 2–4: 239–257. doi: 10.1504/IJTM.2010.033804.
Fuller, D. (2014). Chip design in China and India: Multinationals, industry structure and
development outcomes in the integrated circuit industry. Technological Forecasting &
Social Change, 81: 1–10. doi: 10.1016/j.techfore.2012.10.025.
Gutierrez, R. M. and Leal, C. F. (2004). Complexo eletrônico: Estrategias para uma indústria
Int. J. Innovation Technol. Management Downloaded from www.worldscientific.com

de circuitos integrados no Brasil [The Brazilian Electronics Complex: Strategies for a IC


by UNIVERSITY OF SAO PAULO on 06/25/18. For personal use only.

industry in Brazil]. BNDES Setorial, 19: 3–22.


Kasahara, Y. and Botelho, A. J. (2016). Catching up and falling behind: An appraisal
of Brazilian industrial policy in the twenty-¯rst century. European Review of Latin
American & Caribbean Studies, 101: 97–109.
Kirk, K. and Cotton, C. (2012). The Cambridge Phenomenon. Third Millenium Publishing,
London.
Lange, K., Muller-Seitz, G. and Windeler, A. (2013). Financing innovations in uncertain
networks: Filling in roadmap gaps in the semiconductor industry. Research Policy, 42, 3:
647–661. doi: 10.1016/j.respol.2012.12.001.
Lehoux, P., Daudelin, G., Williams-Jones, B., Denis, J. L. and Longo, C. (2014). How do
business model and health technology design in°uence each other? Insights from a lon-
gitudinal case study of three academic spin-o®s. Research Policy, 43, 6: 1025–1038. doi:
10.1016/j.respol.2014.02.001.
Lojek, B. (2007). History of Semiconductor Engineering. Springer-Verlag, Berlin, Heidelberg,
New York.
Mays, S. K. (2013). Rapid advance: High technology in China in the global electronic age.
Doctoral dissertation, Graduate School of Arts and Sciences, Columbia University, 462
pages.
McKinsey. (2013). McKinsey on Semiconductors. McKinsey & Company. https://www.
mckinsey.com/client service/semiconductors//media/32ae520663114c6eb1250bbdb92673c2.
ashx
Ministry for Science Technology and Innovation. (2015). Estrategia Nacional de Ci^ encia,
Tecnologia e Inovacao~ 2016–2019 [National Strategy for Science, Technology and Inno-
vation 2016–2019 ]. MCTI, Bras{lia-DF.
Newth, F. (2012). Business Models and Strategic Management: A New Approach (New York
edn.). Business Expert Press, Techlink, Singapore.
Pinto, K. E. and Feldmann, P. R. (2016). Why Brazil doesn't innovate: A comparison among
~ 13, 1: 63–82. doi: 10.1016/j.rai.2016.04.002.
nations. Revista Brasileira de Inovacao,
Probert, J., Connel, D. and Mina, A. (2013). R&D service ¯rms: The hidden engine of the high-
tech economy. Research Policy, 42, 6–7: 1274–1285. doi: 10.1016/j.respol.2013.03.004.
Rodriguez, M. and Machado, J. (2008). Are KIBS more than intermediate inputs? An
examination into their R&D di®user role in Europe. International Journal of Services
Technololgy and Management, 10, 2–4: 254–272. doi: 10.1504/IJSTM.2008.022122.
Teece, D. J. (2010). Business models, business strategy and innovation. Long Range Planning,
43, 2–3: 172–194. doi: 10.1016/j.lrp.2009.07.003.
Thun, E. (2007). The globalization of production. Global Political Economy, ed. J. Ravenhill.
2nd edn. Oxford University Press, Oxford, pp. 346–371.

1850031-24
Business Modeling and Public Policy in Brazil's Semiconductor Business

Tidd, J. and Bessant, J. (2013). Managing Innovation: Integrating Technological, Market and
Organizational Change, 5th edn. John Wiley & Sons, Hoboken, New Jersey.
Tongur, S. and Engwall, M. (2014). The business model dillema of technology shifts.
Technovation, 34, 9: 525–535. doi: 10.1016/j.technovation.2014.02.006.
Tsuji, M. and Kuchiki, A. (2010). From Agglomeration to Innovation: Upgrading Industrial
Clusters in Emerging Economies. Palgrave Macmillan, London.
van Marion, M. (2014). International Trade Policy and European Industry: The Case of the
Electronics Business. Springer, London.
Villa, A. and Antonelli, D. (2009). A Roadmap to the Development of European SMEs.
Springer-Verlag, London.
Yin, R. K. (2003). Case Study Research: Design and Methods. 3rd edn. Sage Publications,
London.
Zott, C. and Amit, R. (2013). The business model: A theoretically anchored robust construct
Int. J. Innovation Technol. Management Downloaded from www.worldscientific.com

for strategic analysis. Strategic Organization, 11, 4: 403–411. doi: 10.1177/147612701


3510466.
by UNIVERSITY OF SAO PAULO on 06/25/18. For personal use only.

Biography
Alex da Silva Alves is an Associate Professor of Finance and Entrepreneurship at
the Department of Economics, Management and Sociology, ESALQ/USP, where he
coordinates the \Esalq Management" Research Unit. He has over ten years of aca-
demic and consulting experience in science, technology and innovation policies and
technology valuation. His research interests are innovation ¯nancing strategies and
technopreneurship policies.

Antonio Jos e Junqueira Botelho is Full Professor at IUPERJ and a founding


partner of Innovastrat Consultoria Ltda., an international consulting ¯rm in science,
technology and innovation policy, digital society, corporate strategies and entre-
preneurship policy and ¯nance. As a pioneering researcher and practitioner in early-
stage venture capital in Brazil, he is an angel investor, co-founder, former president
and member of the board of directors of Gávea Angels, the ¯rst angel investors club
in Latin America. He has done extensive research and published widely on the
semiconductor and computer hardware and software industries and policies. His
current research interests are in university and research internationalization, inno-
vation and technopreneurship policy, early-stage innovation ¯nancing, renewable
energy strategies and development supply networks in the global oil and gas in-
dustry.

Virg³nia Duarte manages the SOFTEX Intelligence Unit in Brazilian Software


and coordinates the SOFTEX Observatory, carrying research and developing pro-
jects in the areas of smart cities, software and IT entrepreneurship promotion pol-
icies. She has published dozens of papers, books and book chapters in these areas.

1850031-25

You might also like