1. Produced and issued by: ABN AMRO Bank NV+
Equity | Tech Hardware & Equip | United Kingdom
7 October 2009
Change of recommendation
ARM
Buy (from Hold) ARM-ed and dangerous
Target price
£1.80 (from £1.21)
We upgrade ARM to Buy with a target price of 180p for two reasons: 1) higher
EBIT margins for 2009-14F driven by growth in Smartphones and MCU and 2) re-
Price
£1.43 rating of the stock due to our forecast of market share gains in notebook PCs,
Short term (0-60 days) which could substantially disrupt the Intel/Microsoft model by 2012/13.
n/a
Sector relative to market Key forecasts
Underweight
FY07A FY08A FY09F FY10F FY11F
Revenue (£m) 259.2 299.0 297.5 342.4 382.7
Price performance EBITDA (£m) 108.2 124.2 111.3 136.7 % 164.8 %
Reported PTP (£m) 86.70 100.8 85.60 111.2 % 139.6 %
(1M) (3M) (12M)
Normalised PTP (£m) 86.70 100.8 85.60 111.2 % 139.6 %
Price (£) 1.27 1.19 0.90
Normalised EPS (p) 4.68 5.64 4.94 6.41 % 7.99 %
Absolute (%) 12.7 20.4 59.3
Rel market (%) 11.7 4.5 78.7 Dividend per share (p) 2.00 2.20 2.42 2.66 2.93
Rel sector (%) 11.7 8.4 65.9 Dividend yield (%) 1.40 1.54 1.69 1.86 2.05
Normalised PE (x) 30.50 25.30 28.90 22.30 17.90
Oct 06 Oct 07 Oct 08
1.6 EV/EBITDA (x) 16.50 14.20 15.60 12.40 9.98
1.4
EV/invested capital (x) 2.63 2.18 2.03 1.88 1.72
ROIC - WACC (%) 2.31 3.63 2.58 3.92 5.23
1.2
1.0 Use of %& indicates that the line item has changed by at least 5%. year to Dec, fully diluted
Accounting standard: US GAAP
0.8 Source: Company data, ABN AMRO forecasts
0.6
0.4 ARM could become a major competitor of Intel
ARM.L Europe Technology Over the next two to three years, we believe ARM could become a viable alternative to Intel
in the PC market. First, we believe ARM processors will match Intel’s performance while
Market capitalisation beating them on power consumption and possibly cost. Second, we expect PC
£1.84bn (€2.01bn)
manufacturers to switch from Intel/Microsoft OS-based platforms to ARM/Chrome OS-based
Average (12M) daily turnover
platforms beginning in 2H10 to reduce their dependence on Intel and improve margins. Third,
£8.03m (€9.18m)
with ARM-based PCs gaining traction with consumers, we believe ARM could receive
RIC: ARM.L, ARM LN
Priced at close of business 6 Oct 2009. support from Microsoft and port Windows to the ARM architecture. We estimate ARM could
Source: Bloomberg
capture 30% of the notebook PC processor market by 2014, creating a major disruption to
the Intel-Microsoft domination of the PC market.
Strategic value of ARM could trigger bid speculation, in our view
We do not expect PC market share gains to give a major boost to ARM’s EPS before 2014,
but we believe it may lead to a re-rating of the stock as the company could become an
Analysts acquisition candidate. If ARM were to be acquired by a major player, it would likely cost
Didier Scemama hundreds of millions of dollars to unify their systems and it would disrupt the long-term
+44 20 7678 0772 roadmap of virtually all handset and consumer electronics OEMs, an unacceptable risk for
didier.scemama@rbs.com
the entire electronics industry, in our view. So, we believe a consortium of electronics
Alexandre Faure companies could pre-emptively bid for all or part of ARM in an effort to guarantee its
+44 20 7678 7231
alexandre.faure@rbs.com independence.
Marketing analyst Upgrading ARM to Buy from Hold with a new target price of 180p
Paraag Amin, CFA
We raise ARM’s medium-term growth and margin forecasts on growth in smartphones and
+44 20 7678 7513 MCU. We also raise our DCF-based target price from 121p to 180p. Our target price implies
paraag.amin@rbs.com mid-term EBIT margin of 43% (37.5% previously). At our target, the stock would trade on an
250 Bishopsgate, London, EC2M 4AA, FY10F P/E of 28x, near the top-end of its historical P/E range of 15-29x.
United Kingdom
Important disclosures can be found in the Disclosures Appendix.
+
http://www.abnamroresearch.com ABN AMRO group companies are subsidiary undertakings of The Royal Bank of Scotland Group plc.
2. The basics
Versus consensus Catalysts for share price performance
! Stronger-than-expected shipments of smartphones, as they carry an average 4x ARM-based
EPS (p) ABN Cons % diff
AMRO chips than an average electronics product;
2009F 4.94 4.80 +3.0%
! Continued shift in the mobile phone market from mid-range phones to smartphones: we
2010F 6.41 6.00 +6.9%
expect smartphones to account for 16% of the overall mobile phone market this year and 20%
2011F 7.99 7.00 +14.2%
by 2011;
Source: Bloomberg, ABN AMRO forecasts
! Further market share gains in the Home and Enterprise segment at major chipmakers such as
STMicroelectronics, Broadcom and LSI;
! Faster adoption of ARM-based MCU in the embedded market;
! Major product announcement from ARM licensees indicating their entry into the PC processor
market;
! Microsoft announcing it would port its Windows operating system to the ARM architecture;
! Google’s Chrome OS receiving support from major PC OEMs such as Dell, HP, Acer and
Forced ranking* Toshiba; and,
Company Rec Upside / ! Increased subsidies from mobile operators for netbooks and notebooks based on ARM
Downside
processors.
Nokia Buy +27%
ARM Buy +26% Earnings momentum
ASML Buy +23%
ARM, like all semiconductor companies, should benefit from improving trends in the chip sector
Infineon Buy +18%
and from the secular growth in the smartphone market. We expect consensus forecasts to
Ericsson Hold -8%
Philips Hold -13%
increase post 3Q09 results.
STMicro Hold -21%
Logitech Sell -22% Valuation and target price
Wolfson Sell -25%
At our target price, the stock would trade on FY10/11F P/Es of 28.1x and 22.5x, within its
Alcatel-Lu. Sell -51%
historical forward P/E range of 15-29x. Given our view that ARM is rapidly becoming a strategic
* by difference to target price as at time of
publication. Recommendations may lie
asset for the electronics industry, we believe it will continue to trade towards the high end of its
outside the structure outlined in the forward P/E, particularly if the market were to price in a potential full or partial bid for ARM shares.
disclosure page.
Source: ABN AMRO forecasts
We forecast a 21% EPS CAGR from 2009 (4.94p) to 2014 (13.05p). Discounting this P/E (21x) by
five years would imply a target P/E of 13.5x on 2014F EPS, which also gives us a fair value of
177p, similar to our DCF-based analysis. The stock currently trades on FY09/10/11F P/Es of
28.9x, 22.3x and 17.9x, on our forecasts.
How we differ from consensus
We are substantially more positive than Bloomberg consensus for FY10/11F EPS (6.9% and
14.2% above), as we have a more positive view than the market on the company’s revenue
Key events growth, particularly in terms of royalties, which drive higher gross and EBIT margins.
Date Event Risks to central scenario
27/10/09 3Q09 results
Downside risks to our investment case and target price include: 1) substantially weaker end-
Source: Company
demand than we model, particularly in the smartphone market; 2) a less successful uptake of
Google’s operating system (OS) in the PC market, hindering ARM’s progress; 3) inventory build-
up in the supply chain leading to much weaker-than-expected 2010 semiconductor industry
growth (our estimate +14.5%), and 4) materially higher GBP/USD rates than our assumption of
1.65x.
ARM | The Basics | 7 October 2009 2
4. Contents
ARM could become a threat to Intel 5
We upgrade ARM from Hold to Buy, as we expect the company to become a real 5
threat to Intel in PCs beginning 2012. We have raised our mid-term EBIT margin
forecast from 37.5% to 43%. As a key strategic asset, we believe ARM could
generate acquisition interest.
Why ARM processors are likely to become successful in PCs 10
Over the coming two to three years, we believe ARM could become a serious 10
contender in the PC processor market, disrupting Intel’s domination.
Introduction 10
ARM processors have caught up with Intel on performance while requiring substantially less 10
power
Quantifying the possible benefit to ARM’s bottom line 14
ARM could become an acquisition candidate, in our view 16
Revisiting our medium-term forecasts; 40% EBIT margin by 2012/13F 19
We have revised our financial forecasts to reflect three elements: the faster and 19
broader adoption of smartphones, our higher market-share assumptions for ARM in
the embedded segment, and initial take-off of ARM’s architecture in notebook PCs
from 2011F.
Financial statements 24
ARM | Table of Contents | 7 October 2009 4
5. ARM could become a threat to Intel
We upgrade ARM from Hold to Buy, as we expect the company to become a real threat to
Intel in PCs beginning 2012. We have raised our mid-term EBIT margin forecast from 37.5%
to 43%. As a key strategic asset, we believe ARM could generate acquisition interest.
Why ARM processors are likely to become successful in PCs
The ‘ARM’ world and the ‘Intel’ ARM processors have so far dominated the portable electronics market (handsets, MP3 players
world are on a collision course, and the like), and we believe they are on the verge of a substantial take-off in the PC market. So
in our view
far, all notebook and desktop PCs have been run on an x86 architecture (coming either from Intel
or AMD), and netbooks, a new class of entry-level portable computers with smaller screens and
size, have seen the first head-to-head battle between ARM and Intel’s architecture.
We believe that, although the market is slowly realising that ARM is becoming a viable alternative
to Intel in the netbook market, it has not understood how much ARM is becoming an alternative to
Intel (and AMD) in the notebook and desktop PC markets.
So how exactly could ARM become successful on Intel’s home turf?
! ARM processors have caught up with Intel on performance while requiring
substantially less power. In the past, ARM processors have lagged those of Intel, clocking at
600MHz in, for instance, the iPhone 3G S, vs 2GHz for Intel’s mobile Atom processors. ARM
recently announced a new processor based on two Cortex-A9 cores capable of delivering
2GHz of processing speed. Beyond that, we believe ARM processors already outperform
Intel’s on power consumption and size. ARM-based microprocessors combine on a single chip
an applications processor, memory controller and graphics processor, while Intel needs two or
three chips to do the same. On footprint, the Cortex-A9 is a third the size of Intel’s current
Atom chip manufactured on the 45nm process node. A single-chip approach is not only less
costly but also requires substantially less power. A slimmer and lighter form factor enables
entry into new markets and more end-products.
! ARM-based processors for PCs likely to be announced soon. Several ARM licensees
(Qualcomm, Samsung, Nvidia and Marvell in particular) have already developed system-on-
chips (SoCs) around ARM processors clocking at 1Ghz. So far these chips have been aimed
at the high-end smartphone and the netbook PC markets. With the announcement of a 2Ghz
dual ARM core, we believe these same companies will likely soon launch SoCs that integrate
this new ARM processor with other functionalities, such as graphics processors (GPU), on a
single chip, allowing them to enter new markets such as higher-end notebooks.
! David against Golliath? While it might appear unlikely that a small company based in
Cambridge, England, can compete effectively with Intel in the PC market, we believe this is
not the most helpful way to look at a possible confrontation. The sum of the revenues over the
last 12 months of the ARM licensees most likely to enter the PC processor market (namely
Qualcomm – chipset division only – Marvell, Nvidia, Broadcom, Texas Instruments and
Freescale, all of which have already announced processors running at 600Mhz or higher)
comes to US$30.2bn, only 10% lower than Intel’s total. The combined R&D budget of these
ARM licensees is US$7.6bn, 1.4x higher than that of Intel. The key element of differentiation in
the next three to five years in the portable PC market is likely to become similar to the key
success factors in the smartphone chip market, namely the ability to integrate different
technologies on a single chip. Supplying a PC processor will not be enough, in our view, to
win in the portable PC market in three to five years; supplying a 3G platform and connectivity
as well as the system software than runs the platform will likely differentiate the winners from
the losers. We believe ARM is well positioned to deliver on this trend.
! Google’s Chrome OS could open up the PC market to ARM. After releasing the Linux-
based Android OS for smartphones, Google is about to launch the Chrome OS, also based on
Linux but targeting the PC market. Although Linux-based OSs (like Ubuntu) have had little
success so far in the netbook market (consumers prefer to stick with the familiar name of
Windows), we believe this could change if consumers see the Google brand on netbooks.
ARM is a major partner in Google’s Chrome launch. We believe ARM/Chrome will initially be
ARM | Executive Summary | 7 October 2009 5
6. successful in the netbook market and, as ARM-based 2Ghz processors hit the market, we
believe higher-specification notebooks and perhaps even desktop PCs could be based on
ARM/Chrome OS by 2011/12.
! Support from Microsoft would be the cherry on the cake. Clearly, as Windows is only
available on the x86 processor (ie, those coming from Intel or AMD), PC makers must work
with Intel or AMD to make Windows-based PCs. While still highly hypothetical, we believe the
threat from Google’s Chrome OS might push Microsoft into supporting ARM-based processors
for Windows in the future. If that were to happen, we believe it would further cement ARM as a
credible alternative in notebook/desktop PCs for consumer and corporate customers (50% of
the notebook market in units). Clearly, for ARM, winning in the corporate segment would take
several years as CIOs are typically slow to adopt new technologies for fear of massive
disruption to their IT systems; however, we believe these companies could embrace ARM-
based processors once the architecture has been field-proven by consumers.
! How about AMD? We can not ignore AMD, which still commands 15-20% of the PC
processor market. While ARM won’t be in direct competition with AMD for some time (AMD is
not really exposed to the netbook PC market), it is inevitable that ARM and AMD will be in
competition at some point, if our assumption of ARM taking 30% market share in notebook
PCs within five years is correct. However we think the trends are better illustrated by focusing
our analysis on the relationship between ARM and Intel.
! Where could we be wrong? Clearly, for ARM to be successful in netbooks and notebook
PCs, Google’s Chrome OS will have to be reasonably successful with consumers. So far,
Linux-based OSs (such as Google Chrome) have had limited success in the market place
because consumers are not familiar with them. It is currently unclear how well designed and
easy to use Google Chrome will be, because it will only be available from 2H10. Secondly, for
ARM and Google to win a reasonable market share in the PC market, they will have to win
some business away from the Intel-Microsoft duo, which has been an impossible task for other
companies in the last 15 years.
Quantifying the benefit to ARM’s bottom line
We provide three scenarios to estimate how much ARM might benefit if it is successful in the PC
market. Overall, we estimate that even reasonable success for ARM in the notebook/netbook PC
market (30% share in notebooks and 35% share in netbooks by 2014F) will not have a material
impact on EPS.
In our scenarios, we have only adjusted the incremental PD units related to the sale of an ARM-
based processor running a netbook or notebook, but have maintained our assumptions regarding
other ARM-based chips (such as a 3G baseband, WiFi, bluetooth or GPS).
Chart 1 : EPS sensitivity Chart 2 : ARM-based PC units Chart 3 : % of portable PC with 3G
14 120 45%
13 40%
100
12
35%
11
80 30%
10
EPS (p)
25%
m of units
9 60
20%
8
40 15%
7
6 10%
20
5 5%
4 0 0%
2009F 2010F 2011F 2012F 2013F 2014F 2009F 2010F 2011F 2012F 2013F 2014F 2009F 2010F 2011F 2012F 2013F 2014F
Base case Bear case Bull case Base case Bear case Bull case Base case Bear case Bull case
Source: ABN AMRO forecasts Source: ABN AMRO forecasts Source: ABN AMRO forecasts
Bottom line, we estimate ARM’s 2014 EPS will range between 12.8p under our bearish scenario
and 13.4p under our bullish scenario. Under our most bullish scenario, only 6% of ARM’s
Processor division (PD) royalty revenues come from portable PCs (PC processor and/or 3G
baseband), demonstrating that royalties remain substantially more sensitive to growth in the
smartphone market.
ARM | Executive Summary | 7 October 2009 6
7. ARM could become an acquisition candidate, in our view
What would be Intel’s response if ARM succeeds in the PC market?
! What could Intel do if ARM were to become a threat in the PC market? The most obvious
option would be to cut prices to hurt ARM-based processor vendors like Qualcomm or
Samsung. While this may hurt ARM licensees, we believe it will only delay the success of
ARM in the PC market, as PC OEMs will most likely jump at any opportunity to diversify away
from Intel. This was probably best demonstrated when Hon Hai, the largest electronics
contract manufacturer, commented publicly on 16 July 2009 that it would make US$200
netbooks based on ARM processors to meet demand from telecom operators. Also, on
September 2009, the Financial Times reported that Dell would soon make netbooks based on
ARM processors. Intel could update product to compete better with ARM-based competitors.
However, the key issue for Intel will be power consumption and die size. It is generally easier
to improve speed/power consumption, as ARM is doing, than to scale down die size, as Intel
would need to do. Third, Intel could capitulate in wireless and refocus on the high-end wired
world, which seems unlikely to us as Intel’s management views wireless computing as the
company’s big growth area based on public comments made by CEO Otellini at the Intel
Developers Conference in September 2009. Intel also recently (29 June 2009) licensed
Nokia’s 3G baseband IP, indicating that it intended to be a major player in the wireless chip
market. Lastly, given the potential damage ARM’s market share gains in the PC processor
market would have on Intel, we cannot rule out that Intel may bid for ARM in order to eliminate
its nascent competitor.
! Why ARM is so valuable to the global electronics value chain? If ARM were to disappear,
it would cost each of its licensees in the semi market US$50m-100m pa (at least) to replace
the ARM processor and develop their own, based on our estimates (a typical leading edge
chip design costs US$40m-50m for instance). Electronics OEMs (such as Nokia, Apple or
Samsung) have hundreds of engineers writing software for designs based on the ARM
architecture. To change these designs to a new architecture would not only be extremely
costly but would also greatly hurt product innovation and time to market (it would take years
before all software code is ported to this new processor architecture with continuing support
for ARM devices already in the market). Finally, foundries (such as TSMC, GlobalFoundries
and UMC) rely not only on ARM IP for their process technology but, more importantly, need
ARM-based SoCs from companies like Qualcomm and TI to load their fabs. Even if a new
architecture emerged from a hypothetical Intel takeover of ARM, it would take months before
these designs could be qualified on major foundry companies’ processes.
! Why ARM could be bid for by a consortium of electronics companies? Given the
importance of ARM for the electronics ecosystem, we believe that companies in various parts
of that system could organise and bid for all or part of ARM. This consortium could then
license out the ARM technology and preserve the product relationship with each consortium
member. With the company’s EV of £1.7bn and scattered shareholding, we believe a takeover
of ARM would be relatively easy to organise since its free float is 100%. Spread over 10 or
more companies that are major users of ARM technology or have a vested interested in
seeing ARM succeed (like Nokia, Apple, Samsung, LG, Sony, Google, Qualcomm, Marvell,
Nvidia, Broadcom, Texas Instruments, TSMC, GlobalFoundries and many others) it would
cost them a limited amount.
Revisiting our medium-term forecasts; 40% EBIT margin by 2012/13F
We revise our forecasts to reflect: 1) the faster and broader adoption of smartphones, 2) higher
market-share assumptions in the embedded segment (mostly microcontrollers and SIM cards),
and 3) initial take-off of the ARM architecture in notebook PCs from 2011F onwards. Overall, we
raise our FY09/10/11F EPS estimates from 4.77p/5.90p/6.10p to 4.94p/6.41p/7.99p.
ARM | Executive Summary | 7 October 2009 7
8. Table 2 : Summary of estimate changes
2009F 2010F 2011F
Old New Change Old New Change Old New Change
Revenues (£m) 294.5 297.5 1.0% 327.3 342.4 4.6% 348.7 382.7 9.8%
Gross margin 90.9% 91.0% +0.1pts 91.2% 90.7% -0.5pts 91.3% 91.3% -
EBIT margin 27.6% 28.3% +0.7pts 30.8% 32.0% +1.2pts 33.1% 36.0% +2.9pts
Tax rate (%) 25.6% 25.6% - 26.5% 26.5% - 28.0% 27.0% -1pt
EPS (p) 4.77 4.94 3.7% 5.90 6.41 8.6% 6.10 7.99 31.0%
Source: ABN AMRO forecasts
The major driver of our forecast changes is the growing proportion of royalties in the mix, driven
primarily by the increased adoption of ARM-based chips in smartphones and embedded markets
and, to a lesser degree, ARM’s nascent success in netbooks and notebooks (see Charts 4 to 6).
Chart 4 : PD royalty unit shipments Chart 5 : PD royalty revenues Chart 6 : PD ASPs per segment
9,000 500 0.25
8,000 450
7,000 400 0.20
6,000 350
m of units
0.15
US cent
5,000 300
US$m
4,000 250
0.10
3,000 200
2,000 150 0.05
1,000 100
0 50 0.00
2008 2010F 2012F 2014F 0 2008 2010F 2012F 2014F
2008 2010F 2012F 2014F
Traditional phones Converged devices Traditional phones Converged devices Traditional phones Converged devices
Netbooks Notebooks Netbooks Notebooks Netbooks Notebooks
Home Entreprise Home Entreprise Home Entreprise
Embedded Embedded Embedded
Source: Company data, ABN AMRO forecasts Source: Company data, ABN AMRO forecasts Source: Company data, ABN AMRO forecasts
We expect royalties to account for 62% of total revenue by 2014 vs 49% this year (Chart 7). Given
the near-100% gross margin on royalties, the growing proportion of royalties in the mix ought to
mechanically lift gross and EBIT margins. We raise our mid-term (2011-17) EBIT margin forecast
from 37.5% to 43% (see Chart 8 for our near-term EBIT margin assumptions).
Chart 7 : ARM revenue split 2008-14F Chart 8 : ARM revenue growth vs EBIT margin (%)
100% 900 50%
90% 800 45%
80% 700 40%
35%
70% 600
EBIT margin
30%
60% 500
US$m
25%
50% 400
20%
40% 300
15%
30% 200 10%
20% 100 5%
10% 0 0%
0% 2008 2009F 2010F 2011F 2012F 2013F 2014F
2008 2009F 2010F 2011F 2012F 2013F 2014F
Licensing Royalties Dev Sys
Licensing Royalties Dev Sys Services Services EBIT margin
Source: Company data, ABN AMRO forecasts Source: Company data, ABN AMRO forecasts
We raise our target price to 180p (from 121p) and upgrade ARM to Buy
Based on our base case assumptions, we upgrade our recommendation to Buy (from Hold) with a
new DCF-based target price of 180p. Our DCF valuation is based on a WACC of 9%, which
assumes a risk-free rate of 5%, beta of 1.0x, market risk premium of 4% and unlevered balance
sheet. Note that if we were to use the current risk-free rate of 3.6% and five-year equity beta
average for ARM of 0.9x, WACC would be materially lower at 7.2%, which we estimate would
ARM | Executive Summary | 7 October 2009 8
9. boost ARM’s fair value to 227p. We have decided to use more conservative assumptions for beta
and risk-free rate to be prudent. That said, ARM has unusual characteristics not only for a
technology company (high visibility, recurring revenues, high margins and high free cash flow
generation) but for any company (it has limited competition and we forecast its end-markets will
grow in excess of 15% for the next five years). Given the nature of ARM’s business (‘annuity-like’
in a way), one can understand why such a low WACC would be warranted.
At our target price the stock would trade on FY10/11F P/Es of 28.1x and 22.5x, within its historical
forward P/E range of 15-29x. Given the increasing potential strategic value of the company, we
believe it will continue to trade towards the high-end of its forward P/E, particularly if the market
were to price in a potential full or partial bid for ARM shares.
We forecast a 21% EPS CAGR from 2009 (4.94p) to 2014 (13.05p). Discounting this P/E (21x) by
five years would imply a target P/E of 13.5x on 2014F EPS, which also gives us a fair value of
177p, similar to our DCF-based analysis. The stock currently trades on FY09/10/11F P/Es of
28.9x, 22.3x and 17.9x, on our forecasts.
ARM | Executive Summary | 7 October 2009 9
10. Why ARM processors are likely to become
successful in PCs
Over the coming two to three years, we believe ARM could become a serious contender in
the PC processor market, disrupting Intel’s domination.
Introduction
ARM will likely challenge Intel in ARM processors have so far dominated the portable electronics market, particularly in handsets
the PC processor market, in our where they have over a 96% share. With the launch of the Cortex-A9 processor, ARM vendors
view
can now not only compete in the netbook and notebook markets but, in our view, are positioned to
take off in the PC market. Notebook and desktop PCs have consistently run on the x86
architecture (ie, coming from either Intel or AMD), but with netbooks, a new class of entry-level
portable computers with smaller screens and size, we see a possible battle between ARM and
Intel’s architecture.
ARM’s new Cortex-A9 processor (which can operate in excess of 2GHz) will allow chipmakers to
immediately integrate this dual-core processor into their system on chip designs for low-power,
high-performance applications such as netbooks and notebooks, and later scale up to the wired
world of desktops and servers.
So how exactly could ARM become successful on Intel’s home turf?
ARM processors have caught up with Intel on performance
while requiring substantially less power
ARM processors can beat Intel ARM has its roots in low-power embedded processing, which quickly found a home in mobile
on a number of performance devices due to its extremely low power consumption. These first-generation ARM processors
factors
previously operated at lower speeds but are now able to compete with Intel in the notebook
market, in our view.
For instance, the Cortex-A8 ARM core in the iPhone 3GS runs at 600MHz whereas the Intel
mobile Atom processor runs at 2GHz. ARM recently announced its new processor, Osprey, based
on two Cortex-A9 cores, which is capable of delivering 2GHz of processing speed. The Cortex-A9
will be fabbed in TSMC’s 40nm manufacturing process.
The table below shows how the Cortex-A9 measures up against the current Intel solution for
netbooks, the Pine Trail Platform, and the more comparable Medfield chip. The Pine Trail will be
available to netbook manufacturers in early 2010 followed by the Medfield SoC in mid 2011. This
is roughly the same timing as the Cortex-A9
Table 3 : ARM Cortex-A9 versus Intel’s Atom
ARM SoC Intel Chipset Intel SoC
Cortex-A9 (2011) Pine Trail (2010) Medfield (2011)
# of chips 2:
1 1
Atom chip & I/O hub chip
SoC Cortex-A9, mem ctrl, I/O hub No Atom, mem ctrl, I/O hub
# of cores 2 1 1
Speed ~2GHz ~1.6GHz ~2.0GHz
Technology 45nm 45nm 32nm
Price US$25-30 US$20-25 Unknown
Footprint Cortex-A9 1/3 size of the 45nm Atom Unknown
Power 2.5W for Atom
500mW Unknown
2.5W for I/O hub
Source: Company data
Few details are available about Medfield in terms of footprint, power consumption and price. Table
3 details the quantum leap Intel has to make to transition from Pine Trail to Medfield to match
ARM | Executive Summary | 7 October 2009 10
11. Cortex-A9 in these critical areas, which are necessary for mass adoption in netbooks and
smartphones, given Pine Trail consumes 5x more power than the equivalent ARM core.
! Chipset to SoC: Intel has consistently maintained a two chipset solution in its legacy
architectures but now has to integrate the equally power-hungry southbridge I/O controller hub
into the Atom processor. This is new territory for Intel and provides a serious challenge to
keep the die size at a low enough footprint to compete with ARM. At 45nm the Cortex-A9 is
already a third the size of the Atom, giving SoC vendors a lower-cost per chip or bandwidth to
integrate more functionality for the same price.
! Power: From Intel and ARM press releases, the Cortex-A9 SoC (500mW) dissipates a fifth of
the current Atom chip at 45nm. Although Medfeld fabbed at 32nm should automatically
consume less power (30-50%), we believe technology shrink alone cannot account for a 2W
reduction in power consumption between the Atom and the Cortex-A9, without even
considering the power impact of the integrated I/O hub chip. In the mobile sector, ARM has
raised the bar to 500mW, which Intel will have to meet or exceed.
! Price: On raw price Intel looks like it could have an edge over ARM when Medfield hits the
market in 2011. But semiconductor vendors such as Broadcom, Qualcomm, Marvell,
Samsung, Freescale and Nvidia will likely integrate more functionality into the Cortex-A9 SoC,
so it is difficult to draw any conclusions for Medfield from price alone, which will probably be at
about US$20. Even if Intel somehow delivers on power and footprint, integrates 20 years of
legacy code and achieves an adequate form factor for around A$20, it will likely loose market
share because ARM now offers a competitive product in the smartbook/netbook/notebook
space.
David against Goliath?
It's not just ARM vs Intel While it might appear unlikely that a small company based in Cambridge, England, is able to
compete effectively with Intel in the PC market, we believe this is not the most helpful way to look
at a possible confrontation. The sum of the revenues over the last 12 months of ARM’s licensees
most likely to enter the PC processor market (namely Qualcomm - chipset division only, Marvell,
Nvidia, Broadcom, Texas Instruments and Freescale, all of which have already designed
processors based on ARM running at speed of 600Mhz or above and are therefore the most likely
to launch products based on ARM Cortex-A9) comes to US$30.2bn, only 10% lower than Intel’s
total. The combined R&D budget of these ARM licensees adds up to US$7.6bn, 1.4x higher than
that of Intel.
Table 4 : Comparing Intel’s skill set to ARM’s leading licensees within the PC and the Mobile & Wireless segments
Total ARM
PC chip skill set INTC QCOM TXN MRVL FSL NVDA BRCM licensees CSR ATHR
Processor design Strong Strong Strong Strong Strong Medium Strong Weak Weak
Leading edge litho design Strong Medium Medium Weak Weak Strong Medium Weak Weak
Low power processor design Weak Strong Strong Strong Medium Strong Strong Strong Medium
PC OEM/ODM relationships Strong Weak Weak Weak Weak Strong Strong Medium Strong
Mobile and Wireless chip skill set
SoC integration Medium Strong Medium Strong Weak Weak Strong Strong Medium
3G baseband Nokia IP Strong Weak Weak Weak Weak Medium Weak Weak
Mixed signal design Weak Strong Strong Strong Medium Weak Strong Medium Medium
RF CMOS Weak Strong Medium Strong Weak Weak Strong Strong Strong
Cellular RF Weak Strong Weak Weak Medium Weak Medium Weak Weak
Bluetooth Weak Medium Medium Weak Weak Weak Strong Strong Weak
WiFi Medium Weak Medium Strong Weak Weak Strong Medium Strong
GPS Weak Strong Medium Weak Weak Weak Strong Strong Medium
Wireless system IP Weak Strong Medium Medium Medium Weak Medium Medium Weak
Handset OEM/ODM relationships Weak Strong Strong Medium Weak Weak Medium Medium Weak
LTM sales (US$m) 33,612 6,197 10,421 2,466 4,013 2,820 4,318 30,235 539 437
% of INTC 100% 18% 31% 7% 12% 8% 13% 90% 2% 1%
LTM R&D budget (US$m) 5,407 2,163 1,693 774 1,024 806 1,144 7,604 142 121
% of INTC 100% 40% 31% 14% 19% 15% 21% 141% 3% 2%
Source: Company data, ABN AMRO estimates
ARM | Executive Summary | 7 October 2009 11
12. Beyond the size argument, we believe several of ARM’s licensees have mobile and wireless
system IP (including 3G platform design and wireless connectivity, as shown in the table above)
that will increasingly differentiate ARM in the portable PC space in the medium term, where Intel
is, in several cases, not even in the market.
We have also added CSR and Atheros to the table comparing Intel’s chip design skills and ARM’s
licensees. Even though we doubt either CSR or Atheros would enter the PC processor market, we
do believe that both companies have interesting and complementary skills that would be valuable
to Intel. We made this point in our last CSR note (SiRFin’ China, published 30 July 2009), but we
believe it is worth re-iterating.
Portable PCs likely converge The key element of differentiation in the next three to five years in the portable PC market is likely
with smartphones; wireless chip to become similar to the key success factors in the smartphone chip market, namely the ability to
vendors likely to be more
integrate different technologies on a single chip and on a leading-edge CMOS process. Supplying
competitive than Intel, in our
view
a PC processor will not be enough, in our view, to win in the portable PC market in three to five
years; supplying (preferably through an integrated chip with the apps processor) a 3G platform
and connectivity (bluetooth, FM radio, WiFi and GPS) as well as the system software than runs
the platform will likely differentiate the winners from the losers. We believe ARM is well positioned
to deliver on this trend.
Therefore, we believe the playing field for ARM/Intel competition is far more level than is initially evident.
ARM-based processors for PCs likely to be announced soon
Product announcements from SoC vendors (Qualcomm, Samsung and Marvell in particular) will now be able to take ARM’s
ARM licensees are likely to come Cortex-A9 processor and integrate a graphics processor, memory interface, and baseband all on
soon
the same piece of silicon. The result is lower chip ASPs and subsequently lower PC prices. For
example, NVIDIA currently buys the unbundled Intel Atom for around US$45 to use in its Ion
netbook platform. However, with Cortex-A9, Nvidia will now have another supplier choice for the
same level of performance as the Intel Atom. By integrating Cortex-A9 into its Tegra platform
(which currently uses an ARM11 + ARM9 pictured in Chart 9) Nvidia could quickly get to market
with a netbook platform at a low cost, which would provide real competition for Intel, most likely
resulting in Intel dramatically reducing its ASP.
RF and mixed signal chip design
Chart 9 : Nvidia Tegra processor
skills will become more
important in PCs
Source: Nvidia
RF and mixed signal chip design Last week, Broadcom announced the licensing of the Cortex-A9 dual core processor from ARM
skills will become more for use in mobile and wireless applications. Neither ARM nor Broadcom were specific about the
important in PCs exact application, but we think it could be for a netbook platform that would integrate with
Broadcom’s wireless technology (3G baseband, power management, RF transceiver, WiFi,
bluetooth, FM radio, GPS and touch screen IC). This integration factor favours ARM, in our view:
its leading licensees (Qualcomm, Marvell, Broadcom for instance) all have more to offer to PC
vendors than Intel. They already have all or most of the technology necessary to make a 3G
phone and it will be easy for them to re-use this IP in netbooks and notebook PCs. Intel, in
contrast, does not have any RF capability (apart from some WiFi capabilities) or mixed signal
design capabilities (used for power management and touchscreen ICs for instance).
ARM | Executive Summary | 7 October 2009 12
13. The handset market is at the leading edge of ARM integration with multiple chip vendors (eg
Qualcomm, Samsung and TI) already producing Cortex-A8 devices. They will likely integrate Cortex-
A9 to capitalise on the additional performance boost, which would allow them to target new markets.
In our view, the OS is the only aspect of the netbook configuration that is missing for seamless
application to notebooks or desktop PCs. The best chip in the world is no good unless it supports
an OS such as Google Chrome OS or Microsoft Windows OS.
Google’s Chrome OS could open the PC market to ARM
Google wants a piece of the PC Google’s Chrome OS would be the first major OS to support ARM’s architecture and is set to
operating system market launch in 1H10. Although netbooks running on flavours of the Linux OS (like Ubuntu) are already
available in the market (and are running on ARM processors), they have had only modest success
so far, as consumers are not familiar with the Linux OS. Chrome is Google’s streamlined Linux-
based OS that is designed for the web and will run on ARM’s low-power architecture cores. It is
also included in the netbooks, a move that is directly aimed at Microsoft. Given the broad appeal
of the Google brand to consumers, the ARM/Chrome alliance could prove to be much more
successful in the PC market, in our view. Initially, we believed that ARM/Chrome OS would be
successful in the netbook market. However, as ARM-based processors running at 2Ghz hit the
market, we believe that higher-specification notebooks and perhaps even desktop PCs aimed at
the consumer segment could be based on ARM/Chrome OS by 2011/12.
Support from Microsoft would be the cherry on the cake
Difficult to predict if or when The final blow to Intel could come from Microsoft. Microsoft and Intel (Wintel) have dominated the
PC value chain for the last 20 years, polarising profits on their respective products (MPU/chipset
for Intel and operating system for Microsoft) and leaving very little profit for the PC makers.
Clearly, the fact that Windows is only available on the x86 processor (ie, those coming from Intel
or AMD) means PC makers use either Intel or AMD to make Windows-based PCs. Although ARM
has some of its people sitting at Microsoft, we have no evidence that Microsoft could evolve to
support ARM-based processors for Windows. However, we believe ARM’s relationship with
Google’s Chrome OS is likely to tip the balance in its favour, because:
! A free Chrome OS will likely cause a major disruption to Microsoft’s Windows OS software
revenue, which accounted for 26.7% of Microsoft revenue in FY09 with a significant 72%
operating margin (as reported in their 2008 annual report).
! By comparison Windows’ Server division operates in a much more competitive market with
IBM, Unix and Linux, so attaining only a 35% operating margin. Microsoft’s Windows operating
profit could fall if Google’s Chrome OS provides serious competition.
If Microsoft were to move toward supporting ARM-based processors, we believe it would further
cement ARM as a credible alternative in notebook and desktop PCs, not only for the consumer
segment but also for the corporate segment. If Microsoft extends support for ARM in the latter half
2010, we estimate ARM/Microsoft netbooks could hit the shelves in the latter half of 2011.
Chart 10 : Hypothetical timeline for ARM architecture deployment in portable PCs
140m Proliferation of low cost
mobile internet devices
120m
Microsoft announces
Mobile computing unit forecast
First Microsoft A9
100m ARM support?
netbooks in stores
Google Chrome
80m OS release First A9 Chrome
netbooks in stores
60m Broadcom working on Release of ARM A9
A9 device platforms to OEMS
40m
20m ARM announces
Cortex-A9
m
3Q09 4Q09 1Q10 2Q10 3Q10 4Q10 1Q11 2Q11 3Q11 4Q11 1Q12 2Q12
Source: ABN AMRO
ARM | Executive Summary | 7 October 2009 13
14. In our view, ARM could fairly easily take share in the consumer notebook market, but the
enterprise notebook market will be more challenging. Today, the notebook market is split 50/50
between corporate and consumer in terms of units. For ARM to win in the corporate segment
would take several years, as CIOs are typically slow to embrace new technologies, preferring
instead to wait for initial problems to be ironed out and avoid any disruption to IT systems and
legacy support. Also as power consumption becomes more of a pressing issue for CIOs, ARM’s
offering should become more competitive. We believe enterprises could embrace ARM-based
processors once the architecture has been field-proven by consumers.
Quantifying the possible benefit to ARM’s bottom line
Unlikely to have a significant We provide three scenarios to estimate how much ARM could benefit if it were successful in the
impact for the next five years, in PC market. Overall, we estimate that even reasonable success for ARM in the notebook/netbook
our view
PC market (30-40% share by 2014) will not have a material impact on EPS.
In our scenarios, we have only taken into account the incremental PD units related to the sale (or
not) of an ARM-based processor running a netbook or notebook. We do not take into account
other ARM-based chips (such as 3G baseband, WiFi, bluetooth or GPS) that are already reflected
in our model.
Chart 11 : EPS sensitivity Chart 12 : ARM-based PC units Chart 13 : % of portable PC with 3G
14 120 45%
13 40%
100
12
35%
11
80 30%
10
m of units
EPS (p)
25%
9 60
20%
8
40 15%
7
6 10%
20
5 5%
4 0
0%
2009F 2010F 2011F 2012F 2013F 2014F 2009F 2010F 2011F 2012F 2013F 2014F
2009F 2010F 2011F 2012F 2013F 2014F
Base case Bear case Bull case Base case Bear case Bull case Base case Bear case Bull case
Source: ABN AMRO forecasts Source: ABN AMRO forecasts Source: ABN AMRO forecasts
As we explain below, we estimate that ARM’s 2014 EPS would range between 12.8p under our
bearish scenario and 13.4p under our bullish scenario. Under our most bullish scenario only 6% of
ARM’s PD royalty revenues would come from portable PCs (PC processor and/or 3G baseband),
indicating that ARM’s royalties will remain substantially more sensitive to growth in the
smartphone market.
ARM | Executive Summary | 7 October 2009 14
15. Table 5 : Bull/Base/Bear scenario analysis
Base case 2009F 2010F 2011F 2012F 2013F 2014F
Netbook shipments (m) 25 33 45 51 59 67
3G attach rate (%) 3% 8% 15% 20% 25% 30%
ARM share % 3% 15% 20% 25% 30% 35%
Notebooks shipments (m) 168 180 192 206 220 236
3G attach rate (%) 3% 8% 15% 20% 25% 30%
ARM share % 0% 1% 3% 5% 10% 15%
ARM-based PCs (m) 1 7 15 23 40 59
3G baseband units in PCs (m) 6 17 36 51 70 91
Attach rate for 3G in PCs (%) 3% 8% 15% 20% 25% 30%
PD royalty units (m) 3,476 4,479 5,424 6,355 7,325 8,449
Incremental units due to PCs 0.2% 0.5% 0.9% 1.2% 1.5% 1.8%
Blended ASP (c ) per netbook/notebook 0.00 0.02 0.03 0.04 0.06 0.08
PD royalty revenues (US$m) 198 262 312 356 398 455
Incremental royalties due to PCs 0 0 2 3 6 11
% of total 0% 0% 0% 1% 2% 3%
EPS (p) 4.94 6.41 7.99 9.52 11.06 13.05
Bear case 2009F 2010F 2011F 2012F 2013F 2014F
Netbook shipments (m) 25 33 45 51 59 67
3G attach rate (%) 3% 5% 8% 12% 15% 17%
ARM share % 3% 5% 8% 10% 10% 10%
Notebooks shipments (m) 168 180 192 206 220 236
3G attach rate (%) 3% 5% 8% 12% 15% 17%
ARM share % 0% 0% 0% 1% 2% 3%
ARM-based PCs (m) 1 2 4 7 10 14
3G baseband units in PCs (m) 6 11 19 31 42 51
Attach rate for 3G in PCs (%) 3% 5% 8% 12% 15% 17%
PD royalty units (m) 3,476 4,468 5,397 6,319 7,267 8,365
Incremental units due to PCs 0.2% 0.3% 0.4% 0.6% 0.7% 0.8%
Blended ASP (c ) per netbook/notebook 0.00 0.01 0.01 0.02 0.02 0.03
PD royalty revenues (US$m) 198 261 310 353 393 445
Incremental royalties due to PCs 0 0 0 1 1 2
% of total 0% 0% 0% 0% 0% 0%
EPS (p) 4.94 6.40 7.94 9.46 10.92 12.76
Bull case 2009F 2010F 2011F 2012F 2013F 2014F
Netbook shipments (m) 25 33 45 51 59 67
3G attach rate (%) 3% 8% 15% 25% 33% 42%
ARM share % 3% 15% 23% 30% 35% 40%
Notebooks shipments (m) 168 180 192 206 220 236
3G attach rate (%) 3% 8% 14% 23% 32% 40%
ARM share % 0% 3% 8% 20% 25% 30%
ARM-based PCs (m) 1 10 26 56 76 97
3G baseband units in PCs (m) 6 17 34 60 90 122
Attach rate for 3G in PCs (%) 3% 8% 14% 23% 32% 40%
PD royalty units (m) 3,476 4,483 5,433 6,397 7,381 8,520
Incremental units due to PCs 0.2% 0.6% 1.1% 1.8% 2.2% 2.6%
Blended ASP (c ) per netbook/notebook 0.00 0.02 0.04 0.08 0.10 0.12
PD royalty revenues (US$m) 198 262 313 362 409 470
Incremental revenues due to PCs 0 1 2 9 16 26
% of total 0% 0% 1% 3% 4% 6%
EPS (p) 4.94 6.41 8.00 9.68 11.33 13.42
Source: ABN AMRO estimates
ARM | Executive Summary | 7 October 2009 15
16. ARM could become an acquisition candidate, in our view
What if ARM hurts Intel? What would be Intel’s response if ARM succeeds in the PC market?
How might Intel respond to ARM in the PC market?
If we are proven correct and ARM becomes a real threat to Intel in the netbook and notebook PC
markets, Intel could respond in a number of ways:
! The most obvious response is price cuts. Intel is well positioned in this respect as it
controls both the fabrication and distribution of its chips. Price cuts would hurt ARM-based
processor vendors like Qualcomm, Samsung or Marvell. While this may hurt ARM’s licensees,
we believe it would only delay ARM’s success in the PC market. PC OEMs, who until now
have had Intel as sole supplier, would most likely jump at the opportunity to diversify away
from Intel or, at the very least, create a dual supply strategy. This was probably best
demonstrated when Hon Hai, the large electronics contract manufacturer, said on 16 July
2009 that it would make US$200 netbooks based on ARM processors to meet demand from
telecom operators. Also in September 2009, the Financial Times reported that Dell would soon
make netbooks based on ARM processors. In effect, what this means is a transfer of pricing
power away from Intel and towards the PC OEMs.
! Intel could improve its product and edge out ARM-based competitors on performance.
Intel is not standing still and continues to progress on its System on Chip (SoC) development
path. SoC is a new area for Intel, which has traditionally maintained a chipset approach
fabbed in Intel foundries (eg, Atom processor chip, graphics chip, I/O controller hub chip).
ARM is climbing the performance ladder with its SoC architecture and Intel has changed its
approach and has now begun to manufacture Atom SoC chips at TSMC (Intel/TSMC press
release dated 1 March 2009). As technology moves mainstream to the 32nm node, it would
not be surprising if Intel fabs all future Atom chips at TSMC. This could be a double-edged
sword for Intel, as explained below:
! Upside: Atom chips have to be low cost (around US$20) to compete with ARM-
based SoCs. Offloading Atom processing to TSMC would free up capacity at Intel
fabs for higher-margin processors targeting desktops, workstations and servers. If
Intel manufactures Atom-based SoCs in Intel fabs, it most likely would not be able
to achieve the same cost per wafer as TSMC due to TSMC’s significant scale in
SoC production. This would put Intel at a notable disadvantage in any price war vs
ARM-based SoCs fabbed by TSMC.
! Downside: The ongoing story in fabrication is that fabs need to function at
maximum capacity to achieve the lowest cost per wafer. By outsourcing Atom
production to TSMC, Intel would reduce the utilisation of its fabs and most likely
increase the cost of its other processors.
Intel’s Pine Trail (which has an integrated Atom/Graphics chip) is to be fabbed at
TSMC in early 2010, followed by Medfield (an integrated Atom/Graphics/IO
controller on one chip) in 2011. However, the key issue for Intel will be power
consumption and die size. Considering the 20 years of legacy code that Intel has to
support and the power consumption of its existing chips (Atom chip 2.5W, I/O
controller chip 2.5W), it could struggle to compete with ARM (Cortex-A9 500mW)
on size. It is always easier to scale up in terms of speed/power consumption like
ARM is doing, than to scale down in size as Intel urgently needs to.
! Intel could license ARM technology for the portable segment (netbooks, notebooks)
and focus its own x86 architecture on the high-end wired-world processors (servers,
workstations), which are its strength and produce the highest margins. ARM has a long road
ahead before it can scale up and compete in the wired world with incumbent Intel. If Intel were
to license ARM technology for the netbook/notebook space, it would have time to refocus on
improving performance and power consumption in the wired high-end computing space. We
believe that Intel is unlikely to abandon its own x86 core in portable PCs, as it would be an
admission of failure for Intel. Plus, Intel’s management sees the wireless/portable space as
the growth area for the company as indicated by CEO Otellini at Intel Developers Conference
in September 2009. In addition, Intel also recently (29 June 2009) licensed Nokia 3G
baseband IP, indicating that it intended to be a major player in the wireless chip market and
3G-capable portable PCs. However, if Intel were to license ARM technology, it would
ARM | Executive Summary | 7 October 2009 16