현실에 한발, 미래를 향해 한발

Carnegie-" Fingerprint Cards Goldfinger "-지문센서 시장 끝판 보고서

작성일 작성자 진혼곡


Carnegie-" Fingerprint Cards Goldfinger "


We expect 71% revenue growth in 2015~2017 following a commercial breakthrough in 2015. Fingerprint Cards benefits from a first-mover advantage amid a global drive for integration of fingerprint sensors into smartphones. A broad product portfolio, long experience with OEM integration, strong IP and low costs have made Fingerprint Cards a winner in this new era of biometric security mass adoption.

We are initiating coverage of Fingerprint Cards with a BUY recommendation and target price of SEK350

Given fingerprint adoption of around ~26% in consumer electronics currentlywe calculate a total addressable consumer electronics market of USD1.67bn, growing to USD2.15bn by 2017e and 47% adoption driven by global initiatives in online banking, cloud computing and sensor technology allowing an acceptable mass-market priceEmerging verticals such as payment cards may provide additional potential.

We estimate that Fingerprint Cards has close to 100% market share in consumer electronics, excluding Apple and Samsung. With little competition initially, we estimate 71% average revenue growth and 84% EPS growth in 2015~2017eWe expect the scalable business model to offset ASP and gross margin pressure, and increasing penetration to mitigate market share pressure.

The company¿s first-mover advantage has created an entrenched position and a large scale which is hard to compete with. We argue that the massive earnings momentum warrants at least a P/E(16e) of 15x despite technological and competitive risks tied to a new market."



Investment case 

Fingerprint Cards (FPC) is looking to grow revenue by 960% in 2015e, enjoying its first-mover  advantage  in  a  global  drive  for  integration  of  fingerprint  sensors  into smartphones. A broad product portfolio, long experience with OEM integration, strong IP and low costs have made Fingerprint Cards a winner as we enter an era of biometric security mass adoption. With around ~25% fingerprint adoption currently, we calculate a total addressable consumer electronics market of USD1.67bn, growing to USD2.15bn by  2017e  driven  by  global  initiatives  in  online  banking,  cloud  computing  and  sensor technology allowing an acceptable mass-market price. We estimate Fingerprint Cards has close to 100% market share in consumer electronics, excluding Apple and Samsung. With little competition initially, we estimate 71% average revenue growth and 84% EPS growth  in  2017e.  We  believe  Fingerprint  Cards  has  ample  potential  and  we  initiate coverage  with  a  BUY  recommendation  and  a  target  price  of  SEK350,  reflecting  a P/E(16e) of 15x. 


Close to 100% market share in consumer electronics excluding Samsung & Apple 

FPC was the first and currently one of three companies delivering fingerprint sensors to the largest  smartphone/tablet  producers  in  the  world.  Its  market  share  excluding  Apple  and Samsung  is  close  to  100%  in  consumer  electronics.  Apple  uses  proprietary  technology  and Samsung uses Synaptics and Egistec. With a surge of smartphones entering the global market in 2015, most OEMs have picked FPC as provider of fingerprint scanner components, i.e. Huawei, Sony  and  HTC,  together  with  smaller  Asian  brands  such  as  Coolpad,  ZTE  and  Meizu. Consequently revenue guidance in 2015 is for SEK2.5bn, up from SEK234m in 2014, or 960% growth Y/Y driven by a stunning SEK1.25bn order inflow in Q2(15).

  

Secular trends are enforcing higher security and rendering PIN codes obsolete 

Trends in the global payment industry, moving towards online banking and digital points of sale, together with cloud computing and ‘the internet of things’ central for adoption of fingerprint sensors across consumer and industrial applications. These trends enforce demand for greater security through convenient solutions. Biometric security is the cornerstone of ID verification and is the only viable technology to date that is sufficiently low cost. 


USD2.15bn market in 2017e based on 47% adoption 

We calculate a current total addressable consumer electronics market for FPC of USD1.67bn in 2015e, based on a 25% penetration rate. As we have witnessed massive adoption of fingerprint scanners  in  most  high-end  phones  and  low-cost  smartphone  OEMs  in  Asia  due  to  mobile payment support, we see penetration of 37% in 2016e. We calculate an addressable consumer electronics market in 2017e of USD2.15bn based on 47% adoption. Additional opportunities may reveal themselves in traditional verticals such as time & attendance application and payment cards. Successfully driving adoption of payment cards may open a USD1.4bn–2.2bn market in few years.  


Entrenched position due to a first mover advantage 

We believe FPC has ample potential. The combination of several strategic moves has made the company the preferred sensor supplier for the most important market, smartphones and tablets. A success attributed to proven manufacturing process, broad product portfolio and integration precision on the back of long experience. FPC can deliver a high-quality sensor, at low cost, and strong performance comparing to peers. 


71% average revenue growth 2015-17e    

Target of SEK350 based on 15x EPS(16e) of SEK24 

We estimate a revenue CAGR(15-17e) of 71% with an average EBIT margin of 27% in our forecast  period.  Here,  we  have  accounted  for  significantly  increasing  competition  from  new vendors entering the market. We expect ASP and, consequently, gross margin pressure to be offset by a scalable business model. We expect falling market share to be offset by increasing penetration. We calculate fully diluted EPS(15e) of SEK7.0, growing to EPS(17e) of SEK23. Despite technological and competitive risks, we argue the massive earnings momentum warrants a PE(16e) of 15x. We initiate coverage of FPC with a BUY and target of SEK350


We see four main risks to our investment case: 

 • Competitive pressure may result in severe ASP and gross margin fall.

 • Fingerprint sensor penetration may develop slower than expected.  

 • Competing ultrasonic technology reaches mass market acceptable prices. 

 • Change  in  smartphone/tablet  design  trends.  Current  initiatives  are  pointing  to  sensor integration under the cover glass and we have yet to see a product from FPC.  


Estimates 

Prominent clients such as Huawei, HTC and Sony 

FPC is facing a fast-growing market for integrated fingerprint sensors smartphones and tablets. The order backlog ended Q2(15) on SEK1.34bn after SEK1.25bn order inflow. The volumes were mainly driven by formidable smartphone vendors such as Huawei, Sony and HTC. Other Asian  manufacturers  contributed  as  well. Namely:  Oppo, Newman,  LeTV,  Coolpad, Gionee, Meizu, Dazen, ZTE, oneplus and ZUK. 

As the company was able to increase its production capacity earlier than expected management lifted  revenue  guidance  to  SEK2,500m  for  2015e.  The  company  is  guiding  for  approx.. SEK860m in Q3(15e). We estimate revenue of SEK2,476m for 2015e reaching SEK8.25bn in 2017e. 


Revenue and gross margin 

We expect the main revenue driver to be increasing adoption of sensors in consumer electronics 

We expect revenue to grow by 130% in 2016e and 27% in 2017e driven by increasing adoption of fingerprint sensors in consumer electronics. We have built a top-down forecast model for FPC, largely based on forecast growth in smartphones, tablets and notebooks and penetration rates.  We  expect  31%  smartphone  penetration  in  2015e,  reaching  45%  in  2016e.  Tablet penetration is around 25% in 2015e. We expect to see sensors being integrated into notebooks from  2016e  and  significant  volumes  from  2017e.  In  sum,  we  estimate  total  non-Apple penetration to grow from 25% in 2015e to 37% in 2016e and 47% in 2017e

We  have  delimited  the  addressable  market  to  smartphone,  tablet  and  notebook  vendors excluding  Samsung  and  Apple.  Apple  has  been  using  a  proprietary  solution  since  buying Authentec, while Samsung uses Synaptics (Validity). We base our global volume estimates on forecasts by formidable industry experts: IDC and Gartner. For 2015–17e we have assumed a 20% market share for Samsung.  

Opportunity within payment cards 

There are opportunities in payment card integration that we expect to materialise in 2016 with volumes from 2017 and beyond, not included in our estimates. In addition, FPC is shipping a number of sensors used in the Chinese banking sector. Roughly 10% of H1(15) revenues were non-smartphone related, of which the majority for the Chinese banking sector. 10% of H1(15) revenues is ~SEK58m, or SEK116m annualized, and is considered ‘run-rate’.  

We account for increasing competition pressuring ASP, gross margin and market share 

Due to rapidly changing market dynamics we take a rather conservative stance. We base our revenue forecasts on a market share estimate and ASP assumption. With new market entrants we expect market shares to decline from ~96% in 2015e to 80% in 2016e and 70% in 2017e. With competitive pressure we also assume ASP falling from ~USD3.3 to USD3.0 in 2016e and USD2.5 in 2017e. The most formidable competitors is Goodix, Egistec and potentially Silead. 

Cautious view on gross margins 

With increasing competitive pressure we remain wary of FPC’s ability to maintain gross margins. FPC’s gross margin inherently scales upwards with falling sensor size as cost falls more than price currently. Hence the management expects the gross margin adjusted for depreciation of R&D expenses and shipment cost will be in the mid-40s. However, we take a cautious stance and expect competitive pricing pressure to offset margin gains. Unfortunately there are no peers against which to benchmark gross margins.  


Costs 

The  main  cost  is  silicon  for  producing  sensors.  For  2015e,  we  expect 77%  of  cost  to  be materials. To enable the huge growth ahead, we pencil in significant ramping of sales expenses, increasing  administrative  expenses  and  R&D  to  meet  future  demand  for  different  sensor applications and continuous improvement of existing technology.  FPC is a fast growing organisation which makes it hard to estimate how fast cost will scale. As our base case we estimate a 10% sequential Q/Q opex (excl. COGS) growth through 2015 and 2016e. For 2017e we pencil in 20% Y/Y growth in opex (excl. COGS).  

Significantly dilutive warrants programme 

Share count and EPS FPC  has  a  significant  number  of  warrants  outstanding  as  a  part  of  employee  compensation programmes. Some 11.33% of the share capital will be diluted over the next couple of years with 8.25% for exercise on 6 September–6 October 2015. Our EPS calculation assumes all warrants fully diluted. As the warrants are deep in the money we have assumed full dilution of shares and included SEK160m in our cash flow estimate. The current share count is 58.4m and 65.0m fully diluted

We calculate a fully diluted EPS(15e) of SEK7.0 growing 232% to EPS(16e) of SEK29.0. We expect  EPS(17e)  of  SEK27.3.  The  decline  in  EPS  from  2016–17  is  explained  by  FPC  fully utlising its deferred tax assets in 2016e. FPC had SEK415m in deferred tax assets at YE(14). 


Valuation 

Valuating high-growth technology stocks is a difficult task due to a high degree of uncertainty and cash flow well into the future together with limited historical figures. The fingerprint sensor sub-sector is completely new and we have yet to see how its ecosystem will ultimately develop. 

To evaluate FPC we have evaluated three approaches: theoretical, DCF and peer comparison.  

A theoretical approach suggests a fair P/E(16) of 28x. 

• A DCF suggests a more moderate view and reflects a fair P/E(16e) of 15x. 

• A peer-based approach suggests a fair multiple of 13–40x. 

We  conclude  a  target  of  SEK350,  reflecting  a  P/E(16e)  of  15x.  Our  target  reflects  a  15% premium to the Nordic Technology Hardware sector, trading at 13x P/E(16e). We believe a premium to the sector is warranted due to the high relative growth prospects. We believe the considerable uncertainty linked to how this new sub-sector will evolve will limit upside in the short term. If FPC can maintain its strong market share in the medium term there is a clear valuation upside as reflected by peer multiples ranging from 20–45x P/E.  

Theoretical approach 

For the theoretical approach we use a Gordon’s growth model. Our assumptions are outlined below. The model is driven by earnings growth. We have assumed a 29% average growth rate over a 5-year period 2015–20. With minimal investment the reinvestment ratio is 90%, which yields a ‘fair’ earnings multiple of 28x.  

DCF approach 

Our DCF assumptions are listed below, where we arrive at a fair value of SEK363

Peer comparison 

There are few direct peers for FPC. However, we compare it to Nordic technology hardware peers  and  international  high-growth  technology  companies.  With  superior  growth  of  an estimated 71% in 2015–17e we argue it should be valued among the top percentile of peers. Multiples range from 20–47x and above for 2016 earnings for companies growing above 20%.  

Another fingerprint sensor designer in the Nordics is Next Biometrics. However, the company has had little traction with OEMs as yet and therefore estimates are highly uncertain. Thus, the most  comparable  company  in  the  Nordics  is  Nordic  Semiconductor  (NOD).  NOD’s  main growth driver is Bluetooth SMART chipsets, the standard communication standard for devices related to the internet of things. Thus its partly mono product exposure and structural growth drivers  are  highly  comparable.  NOD  trades  at  19x  2016  earnings  and  has  traded  at  23x  on average for the past four years.  

It is also relevant to compare to other international high-growth technology companies. Our selection  below  is  technology  hardware  companies  listed  in  Europe  or  the  US.  The  median P/E(16e)  for  companies  growing  more  than  20%  in  2015–17e  is  22.5x.  The  median  for semiconductor companies only is 20.0x. The top 75th percentile trades at 46x and above, with median growth of 32%.   

As we expect FPC to grow 71% in 2015–17e we argue that the top percentile multiples are warranted. However, Fingerprint Cards is a mono-product company which carries considerable technological  risk.  Until  we  see  how  the  how  the  competitive  landscape  evolves  we  find  it unlikely  that  investors  will  pay  a  significant  premium  to  the  Nordic  sector  trading  at  13x P/E(16e).    

Key risks 

Below we have listed what we view as key risks for the investment case.  

Change in design trends – sensors below the cover glass 

There is a constant drive in smartphone and tablet designs for a minimalistic approach. Part of this development has sparked speculation about placement of next-generation sensors. Initially, Apple  placed  it  on  the  home  button  and  the  other  OEMs  followed.  Now  we  have  seen integration on the back and the side. However, many speculate about placement under the non-active  front  glass.  Fingerprint  sensors  today  have  proven  to  work  sufficiently  only  under ‘sapphire’  glass  on  the  home  button.  Thus,  sapphire  glass  is  used  on  the  home  button  and another  type  (i.e.  gorilla  glass)  on  the  rest  of  the  phone.  Many  consider  sapphire  glass  too expensive for mass market adoption. However, if sapphire glass is used for the whole screen cover, sensors could be placed under the glass. Technological development may enable sensors below conventional cover glass. However, this has not yet been proven. Idex’s presumable work with the glass company Corning may develop a viable product eventually. FPC says it is working on  sensor  under  cover  glass,  but  has  yet  to  show  a  product.  If  sensor  under  cover  glass  is broadly  adopted,  without  FPC  having  a  viable  product,  it  may  shave  FPC’s  market  share significantly.  

Competitive price pressure 

An increasing number of sensor developers are entering the fast growing market and several may be able to produce ‘good enough’ sensors in a short time. We account for competitive price pressure  in  our  estimates  with  both  ASP  and  gross  margin  falling  over  our  forecast  period. However,  this  competition  might  be  more  severe  than  our  base  case,  leading  to  shrinking margins. With increased competition and relatively little incremental innovation in fingerprint sensors,  FPC  faces  the  risk  of  being  commoditised.  In  such  a  scenario  we  could  see  EBIT margins of 5–10% longer term.  

Competing technologies 

Competing technologies will pose a risk for FPC, in addition to the above mentioned under-glass sensors. We believe FPC will be a strong player in the capacitive space. However, emerging technologies  such  as  ‘active  thermal’  sensors  and  ‘ultrasonic’  sensors  may  pose  a  risk.  We consider the risk posed by Qualcomm’s ultrasonic sensor as the most prominent. It has a strong relationship  with  OEMs  and  is  able  to  bundle  its  sensor  with  chipsets  used  in  phones.  If adopted, the technology is likely to be included in high-end smartphones at first and could lead to a higher drop in market share than is reflected in our estimates.  

Penetration rates 

Our revenue estimates are largely based on penetration rates for fingerprint sensors in consumer electronics. Adoption may prove to be slower or faster than expected.  


Potential catalysts 

Samsung design win 

Samsung  holds  ~20%  of  the  market  excluding  Apple.  Samsung  currently  uses  Synaptics (Validity)  sensors.  Samsung  is  seemingly  dissatisfied  with  the  sensor  performance  and  has brought  in  Egistec  to  improve  the  sensor.  However,  if  this  solution  continues  to  bring dissatisfied users, FPC stands a chance to win the design.   

M&A 

FPC is a mono-product company with a semiconductor technology that would fit nicely with a portfolio of the formidable semiconductor houses, e.g. Qualcomm, NXP or Intel. The EV/sales range  in  large  semiconductor  transactions  over  the  past  three  years  has  been  3.5–4.5x.  This translates to SEK307–394 per share


Company introduction 

Fingerprint Cards AB (FPC) is a fab-less company developing fingerprint sensors based on the capacitive (electrical signals) technology. It has built a global market-leading position in smaller touch fingerprint sensors used in most smartphones/tablets. It was the first and currently one of three companies delivering fingerprint sensors to the largest smartphone/tablet producers of the world, excluding Apple. Synaptics, through its subsidiary Validity, is the second, delivering touch fingerprint  sensors  to  Samsung’s  handsets.  With  a  surge  of  smartphones  entering  the  global market  in  2015,  many  have  picked  FPC  as  their  preferred  provider  of  fingerprint  scanner components. Consequently revenue in 2015 is guided at SEK2.5bn, up from SEK234m in 2014, or 970% growth Y/Y

History

FPC was founded in 1997 with its first-generation fingerprint touch sensors. After being listed in 1998, the company developed several generations of its area sensor, secured several IPs and expanded its distribution network into the US and EMEA regions. By 2007 it was the biggest supplier  of  touch  fingerprint  sensors  for  the  Chinese  banking  sector.  Demand  and  focus eventually  shifted  towards  swipe  sensors,  for  which  the  first  commercial  order  was  in  2011. After  Apple’s  launch  of  the  iPhone  5  with  a  touch  fingerprint  sensor,  focus  shifted  back  to touch fingerprint sensors. FPC got its first design win in the handset segment in 2012. By 2013, 21 mobiles/tablets carried FPC’s technology, and by 2014, 18 more smartphones/tablets were introduced with FPC technology.  

Business model and value chain 

FPC has one of the broadest product portfolios in the industry with 10+ sensor sizes and types. These include sensors for the home button, backside and side button integration that are ready for mass production. FPC recently announced the 1200 series, which will begin with sensors that can read through zirconia glass being ready for mass production in Q3(15), followed by product releases in the near future. We should expect more product launches in this category and we are aware that the company is working on a cover-glass solution.  

The lion’s share of revenue stems from the smartphone/tablet segment. However, FPC is also present  in  other  vertical  markets  such  as  corporate  access  in  the  Chinese  banking  sector. Fingerprint scanners are generally deployed in a number of vertical markets and segments, but smartphone/tablet markets make up the largest addressable market.  

Sales are mainly conducted through distributors either as component sales or as project sales in cooperation with the client. The sales are largely conducted on own account or in cooperation with distributors. China is FPC’s primary market through its smartphone and banking market. However, Korea, Japan, Taiwan, North America and India are also important.  

The clients are mainly OEMs, product developers and system integrators. Building relationships with the OEMs is key for increasing adoption of FPC’s sensors. Sales efforts are also focused more towards module manufacturers (sub-contractors to the OEMs) as the dominant share of sales is components (the sensor area only, i.e. not including the module). This is important to reach a broader market and for the OEMs to capitalise on competition and de-risk production. FPC has entered into cooperation with several module manufacturers, e.g. Crucialtek, O-film and Dreamtech

As FPC is a fab-less company, all designs are done in-house while production is outsourced. FPC  buys  cut  silicon  wafers  from  SMIC  which  are  then  converted  into  sensor  units.  The algorithm is partly proprietary and partly sourced from Precise Biometrics of Sweden. Assembly is done through one of the modules mentioned above.  


Following  the  massive  increase  in  demand  and  design  wins,  the  company  has  expanded  its workforce to 196, of which 115 are full-time employees and the rest are consultants. More than 20  people  do  marketing  and  sales  and  during  2014  the  R&D  department  comprised  80 employees.  The  R&D  personnel,  of  which  30%  has  a  PhD,  largely  stems  from  the  telecom industry.  The company currently conducts R&D in four locations across Sweden. The idea is to keep the IP close to home. To further integrate development, FPC acquired Anactum Design AB which specialises in ASIC development.  

A mixed bag of advantages 

We believe FPC’s success is attributable to several competitive advantages. The combination of these  has  made  the  company  a  preferred  sensor  supplier  for  the  most  important  market, smartphones  and  tablets.  As  FPC  is  one  of  several  fab-less  fingerprint  scanner  producers utilising  the  capacitive  principle,  we  believe  the  main  reason  for  success  is  its  proven manufacturing process and experience with client hardware adaption. FPC can deliver a high quality sensor, performing well comparing to peers, at a low price and with integration precision on the back of long experience.  

The purpose of this section is to delimit and discuss competition. As demand for swipe sensors is nearly non-existent in today’s application, we focus on touch fingerprint sensors.  

Does size matter? 

Traditionally, law enforcement and governing bodies have relied on easily detectable, rigid and unique fingerprint traits known as minutiae points. Today, high-security fingerprint sensors are too  expensive  for  integration  into  mass-market  devices.  This  has  forced  developers  to  scale down sensor size to make it affordable. Intuitively, smaller sensor size reduces security as one scan reads fewer unique fingerprint traits (minutiae points). However, FPC claims that its high-resolution fingerprint scanner can maintain security levels by detecting micro traits (e.g. sweat pores).  

While other sensor producers claim that micro pores are not a sufficient measure of identity, micro traits are de facto universally approved by consumer electronics OEMs and the payment industry as a sufficient measure of identity today; e.g. Alipay (Alibaba’s mobile wallet solution) is standard with most smartphones in Asia and has approved FPC’s sensor for payment (<100mm sq).  

We believe the consumer will neglect whether a sensor is 50mm sq or 144mm sq. The demand for increased security will be driven by the payment industry or other stakeholders of fraudulent transaction and identity thefts. Hence, a 0.001–0.002% false acceptance rate (FAR) is accepted, micro traits are accepted, and small sensors are in demand (more on this discussion on page 32). 

Our  thesis  is  that  smaller  sensors  will  continue  to  be  favoured  in  high-volume  consumer electronic applications due to their low-price design flexibility given that security levels remain firm.  The  more  risk-averse  online  banking/payment  vendor  can  easily  apply  two-step verification without causing meaningful consumer frustration, i.e. fingerprint plus a 4-digit PIN. We therefore believe designers of small sensors, such as FPC, will be favoured by the lion’s share of high-volume smartphone/tablet segments. However, we also believe larger sensors will be applied in niche segments with smartphones and tablets for high security application. For stand-alone purposes, we believe larger size sensors will be favoured. 

We  have  split  fingerprint  sensors  into  large  and  tiny  formats.  Tiny  sensors  are  focused  on convenience  and  customised  unit  deployment,  whereas  large  sensors  focus  on  the  highest possible security (minimum FAR) while maintaining the high convenience level (minimum false rejection  rate).  Convenience  formats  range  from  small  50mm  sq  sensors  featured  in  Apple’s phones to 100mm sq sensors provided by FPC and Synaptics (Validity). Security focused sensor formats range from 144–230mm sq.  


Sensor developers need to reduce sensor sizes to meet format restrictions posed by smartphone and  tablet  producers.  However,  our  research  concludes  that  tiny  formats  cannot  securely identify a person without the use of micro traits.  

Good enough? Lower size equals lower price 

If the sensor size is a subjective matter, as proven de facto, then competition should revolve around price versus performance within segments. As mentioned, FPC is strong in smartphones and  tablets,  while  other  competitors  such  as  Next  Biometrics  are  strong  in  higher  security applications or stand-alone applications.  

In  this  section  we  focus  our  cost  analysis  on  the  active  thermal  principle  applied  by  Next Biometrics and capacitive sensors used by companies such as FPC, Goodix, Synaptics (Validity) and Egistec. We view these as the only viable technologies for mass market to date. Qualcomm has released a new ultrasonic sensor, but it is still in development and we consider it unlikely to be competitive on price. Idex expects to launch a capacitive sensor during H2(15e) (more on page 36). 

In brief, we conclude that the polysilicon solution applied by Next Biometrics is and will be the cost  leader  for  the  foreseeable  future  when  comparing  like-for-like  sensor  size.  However, capacitive sensor designers can reach similar cost when scaling down  the sensor. Hence, the capacitive sensors using silicon as base material will not compete in the same segment as larger polysilicon sensors.  

Cost breakdown 

Most of the cost for a capacitive sensor producer is silicon. Silicon sensors are mostly produced using 8in silicon wafers.  The approximate cost of one 8in wafer is ~USD500–600 according to TSMC, but can vary depending on volume and customer relations. An 8in wafer will fit about 450–176 sensors that measure 62–148mm sq. This makes the silicon unit cost USD0.6–1.7. The polysilicon unit cost per sensor is ~USD0.6–1.0 for sensor sizes 144–201mm sq.  

Technological advancements over the years have allowed for production of 11.8in wafers, which lowers  the  unit  cost  for  silicon  producers.  Industry  expert  IC  Insight  estimates  that  11.8in wafers  will  represent  70%  of  wafer  capacity  by  2017.  A  17.7in  wafer  will  fit  759  sensors measuring 148mm sq. However, the productivity gain from scaling up from 8in to 11.8in and to 17.7in is only 15–20%, according to ICE Corporation. However, unit cost for the silicon sensor element is not likely to be lower than polysilicon.  

Illustrated below is an indicative unit cost comparison between the polysilicon sensor and the silicon sensor manufactured at leading silicon foundries such as TSMC and UMC. 

The  assumptions  used  in  the  calculations  above  rely  on  ICE  Corporation  statements  about productivity gains from increasing wafer size and Next Biometrics’ statement about unit price. 

Silicon  sensor  producer  assembly  and  ASIC  costs  are  based  on  Next  Biometrics’  industry insight. The assumptions are: 

• 20% productivity gains from increasing wafer size.  

• USD1.0 polysilicon unit price. The most important cost is the actual raw material base component, the silicon. Assembly and ASIC  costs  can  fluctuate  among  producers.  

Our  cost  analysis  shows  that  silicon  148mm  sq sensor costs will reach only USD2.8 compared to a USD0.6 for 144mm sq polysilicon sensors. Therefore it is unlikely that silicon producers can compete with the polysilicon sensors on price in the foreseeable future if producing the same sizes.  

However,  smaller  silicon  sensors  enable  producers  to  reach  commodity  unit  costs  as  low  as USD0.6 when reducing the size to 32mm sq. FPC delivers sensors from 92mm sq with its old 1020 model to as low as 32–62mm sq. As silicon is the main cost component, COGS effectively scales proportionally with sensor size.  

FPC states that it has an ASP of USD3.0–3.5 which translates into a unit cost of USD1.8–2.1, which  coincides  well  with  the  calculations  above.  Keep  in  mind  that  FPC  largely  sells  the fingerprint  component  only,  which  enables  reduced  ’assembly  and  ASIC  cost’,  as  displayed above. 

Sensor producers have made statements about “lower” production costs, although no specific production  costs  or  technological  improvement  have  been  mentioned.  But  Idex’s  off-chip solution looks promising and will start sampling in H2(15). This sensor claims to use a smaller amount of silicon. We view all sensors sizes below USD4 as mass market applicable. FPC has an ASP of about USD3.0–3.5 (~50% module and 50% sensor head only). 

Competition among small capacitive sensors 

Capacitive and active thermal technologies are the only ones eligible for mass market currently. The only designer using the active thermal technology is Next Biometrics and its active thermal principle.  Other  technologies  are  currently  inapplicable  for  mass  market  (see  page  36).  Next Biometrics focuses only on larger sensor sizes, which has rendered them less attractive for size sensitive  application  such  as  smartphones  and  tablets.  Competition  between  the  two technologies  is  likely  in  notebooks.  However,  fingerprint  scanner  adoption  in  notebooks  is lagging for reasons discussed on page 30. 

Competition  among  capacitive  sensor  designers  is  more  diffuse.  Here  we  focus  only  on competition in the small touch fingerprint sensors.  Several designers are able to produce ‘good enough’ sensors. Today there are at least five significant designers, namely Fingerprint Cards (Sweden), Goodix (China), Egistec (Taiwan), Synaptics (US) and Silead (China). However, other designers are entering the market, but with no material traction yet, such as Idex, Qualcomm, Vkansee, BYD, FocalTech and MicroArray trying to grab a share in the fast growing market.  

We believe these factors have determined FPC success to date, rated by importance: 

• First mover advantage - proven production and technology track record. 

• Attractive price versus performance. 

• Good design flexibility due to broad portfolio and small size together with shape and colour optionality. 

• Accumulated integration experience which reduces time to market for OEMs.  

New players and reading through glass 

Other  capacitive  sensor  designers  are  planning  to  enter  the  market  and  several  are  sampling sensor  chips,  notably  Idex  of  Norway.  Another  factor  to  consider  is  the  ability  to  integrate sensors under the cover glass of smartphones and tablets.  

Goodix is sampling this solution in collaboration with panel producer TPK. Idex has showcased this technology and aims to start sampling by Q4(15). Recently, Idex announced collaboration with  a  Fortune  500  distributor,  which  we  read  as  being  the  glass  manufacturing  company Corning. If so, it seems Idex has been chosen as the viable sensor technology for in-glass sensor solution. Corning is the manufacturer of the popular ‘gorilla glass’ used in many smartphones today. FPC  states  it  is  possible  to  read  through  cover  glass  with  its  chips;  however,  no  applicable solution has been announced to date. Management has also stated that there is demand for such a solution, but nothing has materialised mainly due to performance issues. 

FPC recently released a  new  sensor  series  able  to  read  through  zirconia  glass  and  we  should  expect  more announcements in this product series

Qualcomm’s ultrasonic sensor also claims to read through both cover glass and other material. We believe the standard for sensor placement is set by smartphone ‘trend-setters’, most notably Apple. Hence, a move to a functioning in-glass solution could shave FPC’s market share if the company cannot provide a good solution.  


Market overview 

The prospective size of the total addressable market is undoubtedly big. FPC is strong in the smartphone and tablet segment. Opportunity persists in the notebook segment and prospective application  in  payment  cards  is  in  development. In  traditional  markets  where  security  is paramount and format applicability is less prioritised we believe companies with inherent lower production  costs,  i.e.  Next  Biometrics,  stand  stronger.  In  this  section  we  will  focus  on  the consumer electronics market.  

Consumer electronics

We  calculate  the  current  total  consumer  electronics  total  addressable  market  for  FPC  is USD1.67bn in 2015e. Underpinned by trends such as the convergence of physical and digital point-of-sale systems, online banking, cloud computing and ’the internet of things’, we believe the  total  addressable  consumer  electronics  market  in  2017  will  be  USD2.15bn (consumer electronics  market  only).  We  view  the  fingerprint  sensor  market  as  two  principal  segments: consumer  electronics  and  the  traditional  market  (e.g.  access  control,  time  and  attendance, government, and law enforcement).  


Consumer electronics is the growth market for fingerprint technology. The addressable market value will depend on how fast the market adopts the fingerprint reading standard. To assess the addressable market for FPC, we have built a forecast model based largely on industry experts and  excluded  devices  running  on  Apple’s  OS.  We  have  also  applied  several  scenarios  of adoption and ASP development described below.  

Payment cards 

Opportunity may reveal itself in integration of fingerprint sensors in credit cards and debit cards. However, there is little traction as yet. The challenge has been integrating a silicon sensor into flexible plastic card while maintaining durability of the sensor and of the card itself. FPC expects development in this area to materialise in 2016 at the earliest. The company has an ongoing project with payment card or smart card designer of Norway on behalf of Mastercard. Zwipe announced  on  24  June  that  it  had  added  10  channel  partners  and  is  experiencing  increasing demand for its solutions.  

Idex  of  Norway  has  announced  an  agreement  with  a  “global  payments  company”  for  the development  of  biometric  payment  cards.  It  has  also  successfully  demonstrated  its  flexible sensor  in  an  ISO-compliant  flexible  card.  About  14.4bn  payment  cards  were  issued  in  2012 (Nielsen Report 2013). This number is expected to grow to 20.6bn by 2017e. In addition, the average replacement rate of payment cards is ~2–3 years. If penetration reaches (say) 10% by 2017e with a sensor ASP (sufficiently low at ~USD100–150 per sensor (sensor head only), this represents  an  addressable  market  of  USD1.4bn–2.2bn.  We  have  not  added  payment  card volumes to our estimates.  

Traditional verticals 

The traditional verticals are hard to quantify, but sources lead us to believe this is a multi-billion dollar market. Industry experts estimate the potential for fingerprint solutions in the banking sector alone to be ~USD1.0bn–1.2bn. NEXT estimates the traditional market to be 9m–12m sensors sold per year in 2014. As cost per unit has been reduced significantly in recent years, additional growth will come from replacement of conventional security measures such as PIN codes and OTP generators. Due to FPC’s focus on consumer electronics and, in our view, less competitive  position  in  traditional  segments,  we  will  focus  on  consumer  electronics  in  this report.   

Consumer electronics market 

We have divided the consumer electronics market into three categories: smartphones, tablets and  notebooks  (including accessories).  For  the  purpose  of  assessing  market  potential  for fingerprint sensors, we will provide an outline below before assessing growth potential.  

When  estimating  FPC’s  addressable  market  we  start  with  forecasting  growth  in  personal electronics such as smartphones, tablets and notebooks. Based on statements by manufacturers and current trends, we firmly believe that fingerprint reader integration will become the new standard. 

However, this applies only to high- to mid-end and niche corporate models in the short term.  Here we are primarily seeking to: 

• Assess the market trend for smartphones, tablets and notebooks. 

• Delimit the market to segments outside Apple OS devices, as Apple Inc. uses proprietary sensor solutions (Authentec). 

• Assess penetration of biometric security solutions. 

Total addressable market (TAM) 

We have built a forecast model based on estimates by leading industry experts IDC and Gartner. We have excluded products from Apple. We expect smartphone tablet and PC shipments to grow from 1.35bn in 2014 to 1.83bn by 2017e and 2.10bn by 2020, a CAGR(15–20e) of 5.0%

Not all devices will have fingerprinting technology. Therefore we have outlined three scenarios for fingerprint technology adoption. The adoption rate will depend on several factors including unit  price,  economic  development,  and  growth  contribution  by  cloud  computing,  online banking and the internet of things. In our three penetration scenarios we outline three adoption rate trajectories. 

The market value will depend on how fast it adopts fingerprint reading. Our base case assumes 41% blended penetration in 2016e. In the low and high penetration scenario we pencil in total adoption of 33% and 51% respectively.  


Average selling prices 

We  do  not  believe  the  USD3–4  unit  price  for  small  sensors  used  today  is  a  limiting  factor. Assessing the bill of materials (BOM), leading industry sources state that sensor integration is viable only if the additional cost is below USD4. Illustrated by the table below, the implied cost increase for fingerprint integration is below USD4 if we assume the cost of other sensors is in the area of ~USD1.0, below the USD4 threshold. 

However, we are seeing handsets with integrated fingerprint sensors selling as low as USD120 which would imply a total sensor cost of ~USD1.5 using the same ratio. Today FPC sells for ~USD3.5–4.0 (full module/LGA/BGA).  


ASP is inversely correlated with volume, competition and technological advancement. As such we believe sensor price will continue to be pressured toward USD2 and below for the coming generations of affordable low-end smartphones out of Asia. We believe smartphone prices will continue to drop as smartphone penetration increases in low-cost countries. Companies such as Telenor  point  to  an  inflection  point  for  mass  market  adoption  in  these  countries  around USD50–100 and below. This means the total BOM should be less than USD45–80.  

For  fingerprint  sensor  integration  in  such  low-end  smartphones  we  would  need to  see ASP down  to  USD1.5–2.0  before  being  viable.  This  should  be  possible  through  economies  of extreme scale, pressured margins across the value chain through competition, and technological advancement in wafer production and more efficient component integration. 

Please  note  that  we  use  full  module  cost  when  referring  to  ASP.  We  are  aware  that  some manufacturers are selling only the sensor component for less than USD3.0 at the moment while module and assembly are sourced through third parties, e.g. FPC. Hence, the general ASP used in this calculation is slightly higher for 2015 compared to the USD3.3 used for FPC estimates

 

 

Smartphone volumes 

Fingerprint  integration  in  smartphones  was  started  by  Apple  and  Samsung.  The  smartphone market is high volume with annual sales of 1,300m units in 2014, according to IDC estimates. The market is characterised by long lead times and focus on format applicability. Lead times are likely to be 6–9 months according to several sources

The  biggest  challenge  for  phone  manufacturers  is  how  the  phone  should  look  to  be  a commercial success. Technological progress has made the fingerprint sensors small enough to fit smartphones. Higher requirements are driving the smartphone market towards more secure solutions such as small area and away from swipe sensors.  

Tier  1  smartphone  makers  are  most  likely  to  be  the  first  to  adopt  a  fingerprint  sensor.  We consider Tier 1 the top five (top six including Apple) as Tier 1. Some 60% of the market in 2014 for smartphones was Tier 1 (Apple, Samsung, Huawei, Lenovo, LG and Xiaomi). Among the top  six,  Apple,  Samsung  and  Huawei  have  integrated  an  fingerprint  sensor  in  the  flagship models. This represents 50% of total volume. If fingerprint sensors are the new standard for these suppliers it implies ~640m sensors sold per year or ~580m excluding iOS. Lenovo and Sony released new products with a fingerprint sensor in August 2015. Other global players such as  LG,  Microsoft/Nokia  and  Xioami  are  still  sitting  on  the  fence  (Xiaomi  is  rumoured  to include the sensor in its new flagship model).  

At USD4.0 per unit, Tier 1 implies a USD2.3bn market alone. In addition, brands such as LG,  Microsoft/Nokia  and  Xiaomi  are  likely  to  adopt  the  fingerprint  sensor  due  to  its  massive popularity  and  requirement  for  mobile  wallet  initiatives  such  as  Alipay  in  Asia  (Xiaomi  is rumoured to be including a fingerprint sensor in its next flagship model). 

The global smartphone market is expected to grow by a CAGR of 6.1% in 2015–17e according to data compiled by Gartner and IDC. We expect 1,932m units to be shipped in 2020, up from 1,300m  units  in  2014.  Industry  experts  estimate  2018e  to  be  the  inflection  point  of  mass adoption due to affordable smartphones entering the market. 

According  to  IDC,  Apple  iOS  had  ~15%  of  the  total  smartphone  market  in  2014,  and  is expected to maintain 15% by 2020. Apple’s market share in the US far outweighs the market share in China, by 36% to 14%. As most growth is likely to come from the Asia Pacific region, we expect the growth rate in android-based devices to be faster than iOS devices. 

The total number of non-iOS smartphones shipped in 2014 was 1,107m. By extrapolating the market share we estimate that 1,642m non-iOS devices will be shipped in 2020; CAGR 6.5%.  

Tablet volumes 

When  fingerprint  sensors  are  made  sufficiently  cheap  and  become  standard  in  smartphones, they is also likely to become standard in larger tablets. We believe that one sensor will be fitted to an entire ecosystem to increase uniformity. For example, Samsung is likely to have the same sensor in the phone and tablet.  

In 2014, a total of 230m tablets were shipped. According to Gartner, tablet shipments will reach 259m units in 2020, a CAGR of 2%. Estimates from industry experts have been cut severely in recent years. The popularity of the tablet was reduced by the increasing size of smartphones, or ‘phablets’, and the increasing portability of notebook computers.  

IOS’s market share in 2014 was 28%. IDC expects a decline in Apple’s market share to 23% in 2020, as Android devices have a stronger foothold in Asia. If we assume that Apple’s market share will fall proportionately over the forecast period, 260m non-iOS tablets will ship in 2020.  

Notebooks  

In June 2013, Microsoft demonstrated a touch fingerprint sensor together with a beta version of Windows 8.1 which will be standard for authentication, enhanced security, and to improve ease of use. Our understanding is that it will push the entire Windows ecosystem towards biometric identification. This entails a second growth opportunity in addition to the traditional market. Today, there is no significant adoption of fingerprint sensor in notebooks. Our channel checks report that notebook integration is less appealing due to the innate support for SPI from sensor modules. This is a standard commonly used in smartphones and light applications. Notebooks use USB. Hence, by integrating fingerprint sensor modules today, OEMs have to use a USB converter which by itself adds another USD1–2 to the BOM. However, we expect notebook adoption  to  follow  increasing  consumer  adoption  of  fingerprint  sensors  through  other consumer electronics.  

There are about 1.5bn users of the Windows OS today. Even though the market is stable, there is significant turnover of stock. According to Microsoft, 100m new licences for Windows 8 were sold during the first six months after the launch. Furthermore, when Windows establishes a new standard  it  has  a  tendency  to  spread  to  other  ecosystems.  If  we  include  all  devices  with  the potential to be fitted with a fingerprint sensor in the broad Windows ecosystem, it is 1.4bn–2.0bn devices. This includes notebooks, desktop computers, gaming consoles and accessories. 

PC and notebook sales have suffered from substitution to tablets. Sales fell on average 7% in 2011–13. However, notebooks will contribute with significant demand in the coming years as more  than  300m  units  are  sold  annually.  Global  PC  shipments  totalled  316m  units  in  2013, versus 351m 2012, and are expected to increase slightly from this level.  

Apple holds a relatively small market share in the PC market. For our forecast we have assumed a stable market share for iOS, or Apple, based systems.  

Fingerprint technology 

We have divided the fingerprint sensor into two segments: convenience-focused and security-focused.  Convenience-focused  formats  are  small  and  provide  a  simple  alternative  to username/password combinations used for accessing personal devices. Security-focused formats are larger and provide a range of credible references as they are used by corporations for access control, time and attendance, and point of sales purposes. 

There are currently three primary uses of biometric data for high security purposes: 

• Domestic  security  and  law  enforcement:  i.e.  automated  fingerprint  identification  systems (AFIS), capable of rapidly determining an individual’s identity, previously used identities, and past activities.  

• Enterprise and e-government services: administration of people, processes and technologies. 

• Personal information and business transactions: business plans that meet customer demands for services at any time, from any location and through many communication devices. 

The convenience segment has focused primarily on replacing username/password combinations where  very  high  security  is  not  necessary.  To  our  knowledge,  these  are  not  accepted  by  the payment industry as stand-alone user verification.   

Security versus convenience – does size matter? 

Sensors come in different sizes and shapes. According to the Federal Information Processing Standards  (FIPS)  publication  201,  a  sensor  must  measure  at  least  211.2mm  sq,  produce  500 pixels per inch (PPI), and 8-bit per pixel is to be approved for government use. This effectively disqualifies  most  fingerprint  sensors.  However,  compromises  can  be  made  for  a  one-to-one matching purpose unlike one-to-many, such as the AFIS (automated fingerprint identification system) used by the US government. 

To achieve a certain level of security a sensor must read enough unique traits on one finger to avoid false acceptance.  

Next Biometrics, one of FPC’s competitors, firmly states that lower sensor size equals lower security.  This  is  based  on  the  assumption  that  a  smaller  sensor  can  detect  fewer  unique fingerprint traits (minutiae points) in one scan. However, this is counteracted by claims from several other manufacturers that the use of micro traits (i.e. minor unique traits such as sweat pores) enables equal security. Evidently, micro traits are changing with time and wear. Thus, some  argue  it  should  not  serve  as  a  reliable  measure  of  identity.  If  these  micro  traits  are changing, the false rejection rate should increase and potentially build consumer frustration. We can see this as a problem in stand-alone fingerprint scanner solutions. However, for applications such as smartphones, the consumer can easily enrol a new set of fingerprints.   

The facts tell us that micro traits are universally approved by consumer electronics OEMs and by the payment industry as a sufficient measure of identity; e.g. Alipay (Alibaba’s mobile wallet solution)  is  standard  with  most  smartphones  in  Asia  and  has  approved  FPC’s  sensor  for payment (>100mm sq). The FIDO alliance, a member organisation of fingerprint designers and counterparts,  is  developing  a  standard  for  sensors.  However,  given  that  the  stakeholders  are several designers, we doubt it will be ground-breaking or exclusive. 

The  development  in  FPC’s  revenue  mix  tells  the  same  story.  Sales  of  smaller  sensors  are increasing as FPC manages to maintain security levels (using micro traits) while scaling down the sensor. Smaller sensors are cheaper and yield greater design flexibility and are thus favoured by OEMs  in  consumer  electronics.  FPC’s  management  points  to  0.001–0.002%  FAR  as  the accepted security level. This means one falsely approved fingerprint in 100,000 or one in 50,000.   

We  believe  the  consumer  will  disregard  whether  a  sensor  is  50mm  sq  or  144mm  sq.  The demand for increased security will be driven by the payment industry or other stakeholders of fraudulent transaction and identity thefts. Hence, 0.001–0.002% FAR is accepted, micro traits are accepted, and small sensors are in demand.   

The question if this will change is highly relevant. Increasing government, corporate and media attention is focusing on online security, most notably the data breaches at Home Depot, Kmart, Staples  and  JPMorgan.  Hence  as  transaction  volumes  increase  and  identity  theft  and  frauds increase, we might see demand shift towards larger sensors.  

Our  thesis  is  that  smaller  sensors  will  continue  to  be  favoured  in  high-volume  consumer applications due to their low-price design flexibility given that security levels remain firm. The more risk-averse online banking/payment vendor can easily apply two-step verification without amassing huge consumer frustration, i.e. fingerprint plus a 4-digit PIN. We therefore believe designers of small sensors, such as FPC, will be favoured with the lion’s share of high-volume smartphone/tablet segments for the time to come. However, we also believe larger sensors will be applied in niche segments with smartphones and tablets for high security applications.  

Can algorithms compensate for size? 

Various sensor algorithms have been around for many years. Minutiae-based systems have been the dominant solution in commercial systems and quality critical open mass-market applications. Pattern-based algorithms have primarily been applied to small format sensors. 

Algorithms have improved gradually over the past decade, but they cannot replicate the security that  a  solid  picture  can.  No  algorithm  can  compensate  for  the  fact  that  a  small  sensor  will capture  a  smaller  portion  of  a  finger.  However,  algorithms  can  improve  the  scanner effectiveness and speed when evaluating the unique minutiae or micro traits. 

Why fingerprint security? 

The  most  common  approaches  to  identification  to  date  have  relied  on  access  cards  and username/password combinations. These methods are insecure for users and organisations. 

There are three levels of authentication:  

• Something you have (token – a key, card or badge). 

• Something you know (password, PIN, a memory). 

• Something you are (biometric trait – fingerprints, iris, voice etc.).   


Passwords are the most common identification method today and are universally recognised as a liability as the phrases are fairly easy to guess using software, or to steal. In most cases a person has several passwords which for organisational matters and ease of remembering are simplified. Single-sign-on services that gather all passwords under one have gained popularity, making it possible for criminals to acquire all passwords by acquiring one. Passwords’ apparent lack of security  was  recently  highlighted  when  the  ‘heartbleed’  bug  was  discovered.  The  bug  was  a scripting error discovered in the OpenSSL cryptographic software. This software is used by 66% of  all  web  servers  and  allowed  others  to  retrieve  information  such  as  passwords  and  other sensitive information untraced.  

Since the beginning of 2004, reports of card fraud have exploded and banks are looking for ways to protect the online banking channel. Different approaches have been implemented to cope with various fraud schemes. 

These methods include: 

• PIN codes. • one-Time-Password (OTP) sheets and tokens. 

• Transaction specific OTPs. 

• OTP by SMS. 

• Smart cards and USB tokens. 

A study by Datagenics and McAfee found that 10% of all PIN codes can be guessed in one attempt and that 17% of people in Europe have been victims of credit card fraud. The average victim lost USD1,800. The total amount of fraud committed through credit card related crimes was USD650m in the UK alone and USD11bn (+14.6% Y/Y) worldwide. The fraud rate (as a percentage of total transaction volume) is slightly above 5bps. There is an apparent need for better solutions. 

The majority of frauds relate to card-not-present (CNP) frauds. CNP transactions happen when a purchase is made through a physical or online merchant while the cardholder is not present. Often the merchant has no good way of identifying the person or transaction.  

OTP generators (token, sheet, SMS or calculator) and smart cards are the predominant way of authenticating transactions. The smart card is used together with a pin code and requires a smart card reader. Both methods are commonly viewed as the best fraud prevention methods to date.  

The hardware and software required to facilitate use of the smart card or OTP generator are either too low in terms of security or offer too little in the way of convenience. In the UK, an average of 33% of banking clients feel that the security measures in place are “a lot of hassle” or “a little hassle”. Hence there is demand for a more convenient solution.  


To achieve the highest level of security one would have to provide authentication level no. 1 through 3, e.g. something you have, know and are. Biometric traits such as retina, face, voice and hand geometry are the alternative to fingerprints. Technology reading these traits has been unsuccessful due to a high false acceptance rate (FAR) and high costs. A reader of these traits can  cost  USD10–500  and  usually  requires  integration  costs.  These  solutions  have  been implemented  only  at  especially  secure  locations  where  authentication  by  fingerprint  is  not enough.  

Thus the only cost-efficient biometric reader today is the fingerprint reader. A fingerprint reader approved  for  authentication,  issued  to  every  online  and  mobile-bank  user  seems  viable.  In addition, customers are turning away from alternative banking services such as payday loans, check-cashing services, or a pay-roll card and towards online solutions. In our view, this is the major driver for fingerprint sensors

Fingerprint sensor technologies 

The  analysis  of  fingerprints  for  matching  purposes  requires  comparison  of  print  pattern features. They are aggregates of ridges and minutiae points. The three basic ones are arch, loop and whorl. The major minutiae are ridge endings, bifurcation and short ridges. A fingerprint sensor captures a digital image of the pattern which is processed to create a biometric template. 

Several fingerprint reading technologies have been developed over the past decade. Below we will explain the technologies in a mass market context. 

Optical 

Optical fingerprint  imaging  involves  capturing  a  digital  image  using  a  small  lens  and  light-emitting phosphor layer that illuminates the surface of the finger. A scratched or dirty touch surface can cause a bad image of the fingerprint and it is relatively easily fooled by a ‘fake finger’. In addition, the nature of this technology does not allow it to be integrated into small devices. Hence it is not viable for mass market production. However, unlike capacitive sensors, optical sensors  are  not  susceptible  to  electrostatic  discharge  damage  (ESD),  which  is  why  this technology is found in larger applications and static installations.  

Ultrasonic sensors 

Ultrasonic sensors use high-frequency sounds to penetrate the outer layer of the skin where the reflected wave measurements form a digital image of the finger. This provides verification of a live finger. This makes the FAR and false rejection rate (FRR) low. Traditionally, this technology has been slow, expensive, bulky, and too data intensive for most access control applications. However, Qualcomm has introduced a new ultrasonic sensor, labelled SenseID, to be integrated into  its  Snapdragon  chipset  used  by  many  phone  vendors  today.  This  technology  looks promising as it can read through most materials and with higher security than common tiny touch fingerprint sensors as it produces a 3D image of the finger (versus 2D from capacitive sensors). Qualcomm expects snapdragon chipsets with support for this technology to be on the market by H2(15).  

Capacitive sensors 

These sensors use capacitor plates to image fingerprints. Skin is conductive enough to produce a capacitive coupling with an individual element on the array. Ridges being closer to the detector have higher capacitance and valleys have lower. They can be sensitive to ESD but insensitive to ambient lighting and are more resistant to contamination than some optical designs. 

We differentiate between passive and active capacitance: active capacitance uses a charging cycle to  apply  voltage  to  the  skin  before  measurement takes  place,  passive  sensors  do  not.  Active capacitance sensors measure the ridge patterns of the dermal layer like the ultrasonic method. Again, this provides low FAR and FRR. 

RF and pressure sensors 

RF sensors use a low frequency signal that is applied to the finger and then read by the detector array, with each pixel operating like a tiny antenna. This technology has the same benefits as an ultrasonic sensor. RF technology is often deployed together with the capacitive principle. 

Pressure sensors can be very thin. There are two types, conductive film detectors and micro-electro-mechanical devices (MEMS). Conductive film detectors use a double-layer electrode on flexible film. MEMS is a newer technology that uses tiny silicon switches on a silicon chip. At the moment pressure sensors are difficult to coat and have a specialised manufacturing process making it too expensive for mass market viability.  

Passive thermal sensors 

Thermal  sensors  can  convert  changes  in  temperature  into  a  specific  voltage.  When  a  finger touches  the  sensor,  it  reads  the  temperature  in  the  ridges.  An  image  is  created  by  the  skin-temperature ridges and the ambient temperature measure for valleys. A  drawback  is  that  the  temperature  change  is  dynamic  and  it  only  takes  about  a  tenth  of  a second for the sensor to reach the finger’s temperature. The sensor requires heating to create a temperature difference of at least 1 degree centigrade.  

Active thermal sensor 

The active thermal principle has solved the problem related to temperature difference in the conventional thermal sensor. A low power heat pulse is applied to each sensor pixel over a short period  of  time  and  a  response  is  measured.  This  is  quite  large  and  different  for  pixels  in proximity  to  a  finger’s  ridge  or  valleys.  The  proprietary  ASIC  reads,  processes  and communicates the signals into a high resolution image to the host system.  

Mass market production – what are the determinants? 

We have identified four characteristics that enable fingerprinting for mass market application. By  mass  market  we  mean  the  possibility  of  integration  into  devices  using  or  applying authentication at some level including notebooks, laptops, phones, tokens and corporate access. 

Security 

The  sensor  must  be  able  to  authenticate  a  person  with  a  sufficiently  low  FAR  rate.  Going forward,  all  signs  point  toward  a  need  for  securely  identifying  individuals.  Currently,  the payment industry accepts the use of 0.001–0.002% FAR on sensors in consumer electronics. However, we believe fingerprint application on a stand-alone basis will pull towards the same requirements as law enforcements. i.e. 8–12 minutiae points.  

Convenience 

The convenience level of a sensor is measured by the false rejection rate (FRR). It measures the probability of rejecting the correct fingerprint. For mass market applicability it should be able to read a fingerprint correctly in over 98–99% of times according to the industry. High FRR will naturally entail frustration and low consumer adoption. To obtain a low FRR, a sensor must be able to read the fingerprint when wet, dry, damaged and cut. It must also be scratch and shock resistant as well as being durable regarding ESD.  

Format applicability 

Today’s mass market applications for fingerprint sensors require sensor formats that will fit their designs, i.e. smartphones, USB tokens, key fobs. The sensors must be sufficiently customisable to meet requirements for colouring, size, flexibility and power consumption.   

Cost 

High  volume  markets  for  fingerprint  sensors  will  not  have  margins  to  sustain  high  prices. According to Microsoft, fingerprint sensors must cost less than USD4 for implementation in all devices.  

Competing technologies 

We  conclude capacitive  and  thermal  sensors  are  the  only  viable  technology  for  mass  market application  today.  The  ultrasonic  principle  from  Qualcomm  looks  promising.  However,  we believe the production cost will be higher than small capacitive and thermal sensors due to the complex nature of the technology currently

Passive thermal and capacitive swipe sensor: Users of this technology apply different pressure, angle and speed to the swipes. Complex algorithms have so far failed to compensate sufficiently for inconsistent swipes. Hence, these sensors have failed to produce low FRR in a high security environment. Manufacturers of both notebooks and smartphones are moving away from swipe towards touch fingerprint sensors. 

Optical  sensors:  these  sensors  are  big,  which  makes  them  unable  to  fit  into  today’s  smaller devices.  A company  called Vkansee  has  developed a  small  optical fingerprint  sensor,  but  we have yet to see it sampling or mass production.  

Pressure  and  traditional  ultrasonic  sensors  have  too  complex  production  methods  to  be commercially viable. However, Qualcomm recently launched a new ultrasonic prototype sensor which could prove to be viable. At this point we do not know performance or price, but we believe it is unlikely that it could compete on cost. That is, unless it is bundled with the chipset from Qualcomm and poses no real addition to the BOM. However, large scale adoption is likely to happen over time (1–3 years).  

Appendix 

Management 

Jörgen Lantto – CEO – stepped up as CEO in May 2015 after being Executive VP CTO since 2013. He is a technical college engineer and has experience as CTO at ST-Ericsson for three years before moving to Fingerprint Cards. He worked 22 years at Ericsson. He took over the position  from  the  previous  CEO  Johan  Carlström  who  stepped  down  to  focus  on  business development within the company. Johan  Wilsby  –  CFO  –  became  CFO  of  Fingerprint  Cards  in  2015.  He  has  previous  CFO experience from Transmode, HP Nordic region, and for Microsoft region Western Europe since 2004. 

Board of directors 

Urban Fagerstedt – Chairman – currently works as VP of R&D with Huawei technologies in Sweden and has an MSc in electronics engineering. Over the past five years he has experience from Ericsson’s Radio Networks design unit. He is a major shareholder in FPC. 

Katarina Bonde – board member since 2015. She currently sits on the board of Propellerhead Software AB, Micro Systemation AB, Image Systems AB, Mycronic AB, Avega Group, Nordax Bank AB and Birger Jarl Fondkommission AB. She has had other directorships in Sweden and the US as. i.e. Chairman of DIBS Payment Services Sverige and Chairman of Netreflector, Inc. She holds no shares. 

Alexander Kotsinas – board member – is partner in the venture cap fund Nexttobe AB. 

Lars Söderfjell – board member – currently is head of strategy & case research at Ålandsbanken and chairman of the Board of Sunfloro AB. He has long-standing experience of senior executive positions  in  the  fields  of  equity  research  and  asset  management  from  companies  including Myrberg Fondkommission, Danske Bank, Handelsbanken and ABG Sundal Collier. He holds shares in FPC. 

Carl-Johan von Plomgren – board member – is currently employed as sales director Northern Europe  &  chief  compliance  officer,  Villeroy  &  Boch  Gustavsberg.  He  is  board  member  of Villeroy & Boch Gustavsberg AB, Villeroy & Boch Gustavsberg AS (Norway), Villeroy & Boch Gustavsberg  AS  (Denmark)  and  Villeroy  &  Boch  Gustavsberg  OY.  He  has  considerable professional experience from senior positions in the IT industry during 1987–2001, gained from such  IT  companies  as  Dell,  Compaq  and  WM-data,  and  subsequently,  as  of  2001,  from industrial  and  finance  companies,  such  as  General  Electric,  Havells  Sylvania  and  Villeroy  & Boch Gustavsberg. He holds shares in FPC. J

an Wäreby – board member – is currently employed as Senior VP and head of Group Function Sales & Marketing at Ericsson. He is chairman of the Board of LM Ericsson International AB and during 2001–11 was a Board member at Sony Ericsson Mobile Communications AB. He does not hold share in FPC.  

Shareholder structure

Historical equity issues 

FPC cards has issued 58.4m shares in total, 22.5m of which through exercised warrants. In total, it has issued shares for SEK520m, of which 87m through warrants. The company states that it does not need further capital injections.  


Share data & key ratios

Disclosures and disclaimers


원본 출처 및 관련 링크

http://theleadmagnet.com/CarnegieGoldfinger.pdf


FPC-Carnegie Goldfinger_150908.pdf


http://www.carnegie.se/privatbank/Aktuellt/stor-potential-i-fingerprint-cards/



p.s

자세한 내용은 첨부한 46쪽 보고서를 참조하세요. 

투자은행 Carnegie 보고서는 지금까지 본 지문센서 시장 보고서중 최고의 보고서입니다.


다양한 Factor에 대해 다루고 있는데 기술, 시장, FPC 성장 전망과 스마트폰뿐만 아니라 태블릿, 노특북 시장에 대한 자료도 있어

참조할 만한 부분이 많아 여러번 읽어도 충분한 가치가 있는 보고서로 판단됩니다.


자료가 충실해서 최대한 원문을 그대로 옮기는데 주력하였습니다.


개인적인 350 kr는 x15는 어떻게 주었나 했더니 노르딕 하드웨어 섹터 평균치 x13에 15% 프리미엄 준 것. 2016년 실적 예상으로는 x9.

개인적으로 성장성에 비해 너무 적게 준 것 아닌가에 의문을 가졌는데 중기적으로 지속적으로 높은 점유율을 유지한다면 x20~40배도 언급.

이론상 x28배를 언급했는데 이는 모건스탠리 대만이 중국 O-film의 목표주가로 산정한 수치이기도 함.

범위를 x13~x40로 주었는데 모건스탠리는 더 넓은 범위를 주었고 Base를 x40로 언급한 것으로 기억.

Carnegie는 20%이상 성장하는 기업에 대해서 x20~x47배 범위를 언급하기도 함


경쟁자 목록에서 제가 주목하는 회사는 미국 Synaptics, 중국 Goodix보다 중국 Egis Tech과 BYD Microelectronics 정도이며

당분간 FPC 독주가 예상되나 2018년~2020년 또는 2020년 이후 현재 MEMS/CMOS 센서와 초음파 기술이 통합되는 형태로 제조사와 시장의 요구가 나올 것으로 예상. 

Goodix의 강점은 대만 TPK와 공동 개발한 Invisible Fingerprint Sensor로 대형 업체에서 관심이 없자 중국 Blackview를 통해 상용화하여 시장의 관심을 끌려고 하고 있음. 시장 동향에서 Goodix는 Meizu MX4 Pro이후 이렇다 할 대형 제조업체를 확보하지 못한 상태.


현재 확인되지 않은 ZTE Nubia Z9, ZTE V3, Lenovo VIBE P1이 FPC일 확율 높은 NDA 상태일 수도 있고 개인적으로 Egis Tech 또는 BYD Microelectronics일 수도 있는 상황인데 현재까지 공급사가 알려지지 않은 상태.(3개 모델 계속 공급사 모니터링 중)


Carnegie가 삼성, 애플을 제외하고 시장동향을 FPC에 의한 100%점유율은 사실상 시장동향상 정확한 자료상태.

다만, 20여개 이상의 중소기업의 30여개 이상의 터치센서 탑재 모델에 대한 언급은 빠져있는 상태

-->구체적인 1개 이상의 터치센서를 탑재한 모델을 내놓은 중소기업은 Elephone, Cubot, Ulefone, Oukitel, Timmy, Mlais, Turing(미국), iOcean, Blueboo, Mstar, ZeaPlus, Mijue, Blackview(Goodix 센서 채택 예상), Qi Cheng, Doogee, xiaolajiao(Small Red Pepper, Egis Tech 센서 채택), Macoox, Smartrisan(Hammer), Pyle, Malata, KooBee, Leagoo, Zopo, Green Orange 약 24개 회사로 모니터링 되며 FPC 센서 채택율이 높을 것으로 예상되며 모듈 공급사는 FPC 제휴 파트너인 대만 Primax, 대만 LiteOn등으로 예상.


연도별 지문센서 시장 예상은 저보다 보수적으로 본 듯. 

수치는 삼성을 고객으로 확보여부에 따라 많은 차이가 날 수 있는 것으로 판단. 

개인적으로 2016년은 FPC 센서를 삼성에 공급하는 해가 될 것임. 빠르면 갤럭시 S7일 수도 있으나 모듈은 삼성 내재화 지속으로 예상.

삼성에 시냅틱스 뒤에 실제로는 Egis Tech이 있는 것을 아는 증권사, 전문가도 드문데 이를 정확히 언급한 것도 정확성을 더함.

Egis Tech(Synaptics)은 삼성의 다양한 Form Factor와 장기적인 지문센서 개발 능력이 FPC보다는 떨어지고 무엇보다 삼성이 요구하는 센서가격을 FPC는 맞추어 줄 수 있음. A8, Tab S2, Note5에서도 기대했는데 확인할 수 없었음.

다만, 이 보고서 Potential 아래 Samsung design win부분에서 "Samsung  is  seemingly  dissatisfied  with  the  sensor  performance  and  has brought  in  Egistec  to  improve  the  sensor"를 언급하고 있는데 Design Win은 이미 작년에 삼성향이 있었고 언제든지 기회를 제공할 것으로 판단.


2015년은 SD810의 발열문제와 안드로이드 5.0.1의 메모리 누수, 배터리 누수 문제로 안드로이드 업데이트에 문제가 발생하였고 발열문제로 프리미엄 모델 발표가 늦춰지거나 무기한 연기되거나 소비자의 외면을 받은 측면도 있음. 개인적으로 2015년 스마트폰, 지문센서 시장의 최대 악재는 센서시장 자체보다는 AP칩 발열문제 였다고 여기고 있음. 올해 12월 이후 또는 내년 1Q 이것이 해결되는 것을 최우선 관심사항으로 두고 있을 정도임.


영업이익율 부문운 크루셜텍 원가와도 상관있는데 제가 아는 것보다 약간 높게 잡은 것도 눈에 띈다.

현재 중국 스마트폰 시장에서 보급형에서 이어 699위안, 799위안 저가 스마트폰에서도 지문센서가 탑재되는 상황은 이미 충분한 가격하락이 있다고 보여 현재 3$ 센서 시장공급가는 유지될 것으로 예상. FPC는 현재 경쟁자 다운 경쟁자가 없는 시장상황에서 가격정책을 펼칠 이유가 없는 상태.


초음파 기술을 발표한 Sonavation U2F 악세사리 시장에 진출한 상태로 모니터링 되어 위협이 되지 않고 퀄컴이 올해 12월~내년 1Q 제조업체에서 30여개 이상의 모델을 내놓을 것을 언급한 바 이는 막혀있던 프리미엄 모델이 발표가 급격히 증가시킬 것.

또한, SD620은 SD808과 SD810보다 GPU성능이 뛰어난 AP칩으로 시장출하시 보급형 모델 공급도 늘어나 소비자들은 다양한 모델과 성능을 경험해 볼 수 있을 것. 퀄컴 Sense ID 기술에 대해서는 누차 언급하여 생략함.


보고서를 읽으면서 2014년 Hauwei Mate7에서 FPC 1020이 성공을 많은 사람들이 회자하지만 Meizu MX4 Pro와 Oppo N3에서 메인공급자였던 크루셜텍이 IDEX 센서문제로 메탈 케이스 노이즈를 처리하지 못해 Meizu MX4 Pro에서 Goodix로 Oppo N3에서는 FPC 1021이 교체하면서 런칭되고 제조사의 다양한 Form Factor요구에 맞춰 1155, 1140, 1145, 1022, 1035등을 빠르게 시험, 개발하고 Under Glass FPC 1200 개발에 크루셜텍과 FPC의 긴밀한 협력도 큰 의미가 있다고 판단.  노르웨이 IDEX는 2014년 큰 기회를 잃어서 2015년을 힘겹게 보내고 있는 반면, FPC 큰 성공을 이루고 있는 결과를 낳음. IDEX는 최근 포춘 500대 기업에 In-Glass 솔루션을 내년에 공급할 것으로 발표한 바 있는데 이 보고서에서 코닝사를 언급했는데 매우 흥미롭네요. 코닝사는 스마트폰 커버 글라스로 사용되는 Gorilla 시리즈를 공급하는 회사로 대부분의 스마트폰이 현재 Gorilla 3,4 를 사용하고 있을 정도인데 커버 글라스 업체와 현재 개발중이라면 매우 의미있는 것.


웨이퍼 칩생산, 웨이퍼 가공, 모듈 조립-가공 단가 자료가 기존 중국 증권사, 모건스탠리 대만 자료와 차이가 남.

SW 알고리즘은 제조업체와 계약은 대당 라이센스로 알려졌는데 Carnegie 보고서에서 대당 0.2$로 산정됨.


FPC는 TSMC가 아닌 SMIC에서 웨이퍼 레벨 칩 생산을 해서 다소 차이가 날 수 있음.


센서 가격은 웨이퍼 레벨 생산단계에서 웨이퍼 크기에 따라 생산되는 센서 갯수가 20%차이가 날 수 있음.

대량주문으로 파운드리 업체 SMIC에서 생산할 때 웨이퍼 크기가 큰 것으로 생산하면 생산단가는 줄어듬을 알 수 있으며 또한 가장 큰 1020과 가장 작은 1140, 1022/1035와는 가격 차이가 약 0.5$ 난다는 것도 알 수 있고 Carnegie 보고서는 3$~3.5$로 언급하고 있음.


Carnegie는 팹리스 업체 BEP를 1.8~2.1$로 보아 2.5~2.8$을 향후 2~3년내 마지노선(FPC 몫 0.7$)으로 판단.


FPC가 중국시장 공략을 위해 중국업체인 O-film을 생태계(Ecosytems)로 끌어들인 것도 시기 적절하고 중요하고 옳은 결정이었다고 판단. 국내 파트론(삼성 카메라 모듈 최대업체)은 Ecosystem에서 빠졌고 드림텍(각종 컨넥터 업체)은 존재감이 없으나 이 보고서는 드림텍을 언급했으며 대만 Primax와 LiteOn은 주력 사업인 카메라 모듈사업에 치중하고 있는 것으로 예상. 다른 센서업체 모듈 파트너에 대한 정보는 지문인식 시장 의견 게시판에 공급망 분석 글을 참조하길 바람.


크루셜텍 월 700~1000만대, O-film 월 400만대로 시작했지만 이제 크루셜텍 1200만대 이상, O-film 월 800만대(내년 월 1000만개~1200만개 증설 예정) 이상 CAPA를 늘렸거나 늘리는 중으로 판단되며 다른 센서업체 대비 시장 공급능력도 충분한 상태.

-------------------------------------------------

3) Carnegie 보고서는 가격,  경제적인 개발같은 직접적인 이유말고도 클라우드 컴퓨팅, 온라인 뱅킹, 사물인터넷(IoT) 성장 파급력도 채택율에 영향을 줄 것으로 언급.  

애플 제외한 FPC 센서 침투율 예상: 2015년 25%, 2016:37%, 2017:47%

애플, 삼성 제외한 시장 점유율 예상: 2015년 ~96%, 2016년 80%, 2017e:70%


2015년 4Q 매출 예상액-1032M SEK(3Q FPC 예상 매출액은 860M SEK)

FPC 2015년 센서 판매 예상 갯수: 8500만개, 2016년 2.27억개, 2017년 3.46억개

FPC 2015년 센서 ASP: 3.3$, 2016년 3$, 2017년 2.5$

애플, 삼성을 제외한 스마트폰, 태블릿, 노트북 시장 댓수 2015:13.17억대, 2016:13.40억대, 2017:14.66억대
애플을 제외한 지문센서 시장 2015:4.18억대, 2016년:6.19억대, 2017년:8.61억대, 2018:12.33억, 2019년:13.54억대, 2020년:14.79억대
글로벌 지문센서 ASP: 2015년 4$, 2016년:3$, 2017년:2.5$, 2018년:2.3$, 2019년:2.1$, 2020년:2$

매출 예상액과 영업 이익율-2015:2476M SEK(40%), 2016:5701M SEK(39%), 2017:7248M SEK(37%)
크루셜텍 영업 이익율은 3Q되야 의미 있는 수치가 나올 것으로 판단되며 7%~8%이상을 기대해 봅니다.




맨위로
통합 검색어 입력폼