May 23, 2011MORGAN STANLEY BLUE PAPER                                                                                     ...
MORGAN STANLEY RESEARCHMay 23, 2011Cloud Computing Takes OffGlobal Technology and Telecommunications TeamContributors to t...
MORGAN STANLEY RESEARCHMay 23, 2011Cloud Computing Takes Off Table of ContentsMORGAN STANLEY BLUE PAPER Measuring Cloud Im...
MORGAN STANLEY RESEARCHMay 23, 2011Cloud Computing Takes OffMeasuring Cloud Impacts: Stronger than Investors AnticipateClo...
MORGAN STANLEY RESEARCHMay 23, 2011Cloud Computing Takes OffBreadth and depth of public cloud adoption expanding          ...
MORGAN STANLEY RESEARCHMay 23, 2011Cloud Computing Takes Off SUMMARY OF KEY TAKEAWAYS BY INDUSTRY      Hardware           ...
MORGAN STANLEY RESEARCHMay 23, 2011Cloud Computing Takes OffWhat We Did and WhyBlue Papers are collaborative reports focus...
MORGAN STANLEY RESEARCHMay 23, 2011Cloud Computing Takes Off   dedicated to a particular customer in managed hosting.     ...
MORGAN STANLEY RESEARCHMay 23, 2011Cloud Computing Takes Offstrategies should help EMC take share, no matter the location ...
MORGAN STANLEY RESEARCHMay 23, 2011Cloud Computing Takes OffExhibit 10                                                    ...
MORGAN STANLEY RESEARCHMay 23, 2011Cloud Computing Takes OffOur Cloud Analysis Drives                                     ...
MORGAN STANLEY RESEARCHMay 23, 2011Cloud Computing Takes OffMeasuring Cloud Impacts: The Coming Server Squeeze            ...
MORGAN STANLEY RESEARCHMay 23, 2011Cloud Computing Takes OffExhibit 14                                                    ...
MORGAN STANLEY RESEARCHMay 23, 2011Cloud Computing Takes OffExhibit 17                                                    ...
MORGAN STANLEY RESEARCHMay 23, 2011Cloud Computing Takes Offdemand. In our model we have assumed a 8.5:1                  ...
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Cloud Computing 23 may 2011
Upcoming SlideShare
Loading in...5
×

Cloud Computing 23 may 2011

2,001

Published on

Morgan Stanley Cloud Computing

Published in: Business, Technology
0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total Views
2,001
On Slideshare
0
From Embeds
0
Number of Embeds
0
Actions
Shares
0
Downloads
75
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

Cloud Computing 23 may 2011

  1. 1. May 23, 2011MORGAN STANLEY BLUE PAPER MORGAN ST ANLEY RESEARCH Global 1 Adam Holt 1 Keith Weiss, CFA 1 Katy Huberty, CFA 1 Ehud Gelblum, PhD 1 Simon Flannery 1 Sanjay Devgan 1 Atif Malik 1 Nathan Rozof, CFA 2 Adam Wood 2 Patrick Standaert 2 Francois Meunier 3 Jasmine Lu 4 Grace Chen 3 Bill Lu 5 Keon HanCloud Computing Takes Off Vipin Khare 6Market Set to Boom as Migration Accelerates 7 Masaharu Miyachi Global Technology andPublic cloud workloads may increase at a 50% CAGR in the next three years— Telecommunications Teamabout twice as fast as the market currently anticipates. Our AlphaWise evidence is *See page 2 for all contributors to this reportbased on the first global survey of IT managers on cloud migration in the US, Europe andAsia-Pacific. Among the 300 IT decision-makers we interviewed, the percentage using the 1 Morgan Stanley & Co. Incorporatedpublic cloud is expected to rise from 28% to 51% in three years, while the portion of their 2 Morgan Stanley & Co. International plc+workload running in the cloud likely will more than double, from 10% to 22%. 3 Morgan Stanley Asia Limited+ 4 Morgan Stanley Taiwan Limited+Server growth shifts to the cloud… We have revised our global technology team’s 5 Morgan Stanley & C. International plc, Seoulserver model to reflect these findings. We estimate that the number of servers shipping Branch+into public-cloud environments will grow at a 60% CAGR through 2013. This migration of 6 Morgan Stanley India Company Private Limitedworkloads from on-premise environments to the public cloud will continue to be a key 7 Morgan Stanley MUFG Securities Co., Ltd.+driver of technology spending. Morgan Stanley Blue Papers focus on critical..and on-premise server growth moves downward. Survey respondents cited server investment themes that require coordinatedhardware as the primary area of savings from the migration to cloud computing. They perspectives across industry sectors, regions,expect to reduce server spending by 8.6% over the next three years. Vendors that or asset classes.depend on on-premise server shipments will likely be hard hit.Capturing the cloud. In this report 17 analysts from seven industries discuss how shiftsin workload location, increased use of virtualization and grid computing, and denserworkload-to-server ratios affect various sectors in the IT universe—both positively andnegatively. We tend to favor cloud service providers that are consolidating demand ontotheir platforms and vendors that support the build-out of cloud environments. We identifyeight companies that are best positioned for the opportunities we see in cloud AlphaWise conducts evidence-based investmentcomputing—Accenture, Salesforce.com, Broadcom, EMC, Juniper Networks, Quanta, research to help validate key investment debatesRackspace, and VMware. A basket of all the names, identified as “Best Positioned for on behalf of Morgan Stanley Research analystsCloud Migration,” is available under the Bloomberg ticker MSMSBPCM. worldwide.Morgan Stanley does and seeks to do business with companies covered in Morgan Stanley Research. As a result, investors should be aware that the firm mayhave a conflict of interest that could affect the objectivity of Morgan Stanley Research. Investors should consider Morgan Stanley Research as only a single factorin making their investment decision.For analyst certification and other important disclosures, refer to the Disclosure Section, located at the end of this report.* = This Research Report has been partially prepared by analysts employed by non-U.S. affiliates of the member. Please see page 2 for the name of each non-U.S.affiliate contributing to this Research Report and the names of the analysts employed by each contributing affiliate.+= Analysts employed by non-U.S. affiliates are not registered with FINRA, may not be associated persons of the member and may not be subject to NASD/NYSErestrictions on communications with a subject company, public appearances and trading securities held by a research analyst account.
  2. 2. MORGAN STANLEY RESEARCHMay 23, 2011Cloud Computing Takes OffGlobal Technology and Telecommunications TeamContributors to this ReportUS SoftwareAdam Holt1 +1 (415) 576-2320 Adam.Holt@morganstanley.comKeith Weiss, CFA1 +1 (212) 761-4149 Keith.Weiss@morganstanley.comJennifer Swanson, CFA1 +1 (212) 761-3665 Jennifer.Swanson@morganstanley.comJonathan Parker1 +1 (212) 761-4223 Jonathan.Parker@morganstanley.comMelissa Gorham1 +1 (212) 761-3607 Melissa.A.Gorham@morganstanley.comUS HardwareKathryn Huberty, CFA1 +1 (212) 761-6249 Kathryn.Huberty@morganstanley.comScott Schmitz1 +1 (212) 761-0227 Scott.Schmitz@morganstanley.comUS Telecom ServicesSimon Flannery1 +1 (212) 761-6432 Simon.Flannery@morganstanley.comEdward Katz1 +1 (212) 761-3244 Edward.Katz@morganstanley.comUS Communications EquipmentEhud Gelblum, PhD1 +1 (212) 761-8564 Ehud.Gelblum@morganstanley.comKimberly Watkins, CFA1 +1 (415) 576-2060 Kimberly.Watkins@morganstanley.comMichael Kim1 +1 (212) 761-3247 Michael.K.Kim@morganstanley.comUS SemiconductorsSanjay Devgan1 +1 (415) 576-2382 Sanjay.Devgan@morganstanley.comAtif Malik1 +1 (415) 576-2607 Atif.Malik@morganstanley.comUS IT ServicesNathan Rozof1 +1 (212) 761-4682 Nathan.Rozof@morganstanley.comGlenn T. Fodor, CFA1 +1 (212) 761-0071 Glen.Fodor@morganstanley.comMatt Lipton, CFA1 +1 (212) 761-6980 Matthew.Lipton@morganstanley.comUS InternetScott Devitt1 +1 (212) 761-3365 Scott.Devitt@morganstanley.comEurope SoftwareAdam Wood2 +44 20 7425-4450 Adam.Wood@morganstanley.comAshish Sinha2 +44 20 7425-2363 Ashish.Sinha@morganstanley.comEurope Communications EquipmentPatrick Standaert2 +44 20 7425-9290 Patrick.Standaert@morganstanley.comEurope SemiconductorsFrancois Meunier2 +44 20 7425-6603 Francois.Meunier@morganstanley.comAsia TechnologyJasmine Lu3 +852 2239-1348 Jasmine.Lu@morganstanley.comGrace Chen5 +886 2 2730-2890 Grace.H.Chen@morganstanley.comKeon Han4 +82 2 399-4933 Keon.Han@morganstanley.comBill Lu3 +852 2848-5214 Bill.Lu@morganstanley.comVipin Khare6 +91 22 6118-2236 Vipin.Khare@morganstanley.comMasaharu Miyachi7 +81 3 5424-5321 Masaharu.Miyachi@morganstanleymufg.com1 Morgan Stanley & Co. Incorporated 4 Morgan Stanley & Co. International plc, Seoul Branch+ 7 Morgan Stanley MUFG Securities Co., Ltd.+2 Morgan Stanley & Co. International plc+ 5 Morgan Stanley Taiwan Limited+3 Morgan Stanley Asia Limited+ 6 Morgan Stanley India Company Private LimitedSee page 97 for recent Blue Paper reports. 2
  3. 3. MORGAN STANLEY RESEARCHMay 23, 2011Cloud Computing Takes Off Table of ContentsMORGAN STANLEY BLUE PAPER Measuring Cloud Impacts Stronger than Investors Anticipate ................................................................................................................................... 4 What We Did and Why..................................................................................................................................................... 7 The Coming Server Squeeze ........................................................................................................................................... 12 Cloud Compression Effects.............................................................................................................................................. 16 Industry Takeaways Hardware ......................................................................................................................................................................... 20 Software........................................................................................................................................................................... 27 Telco Services.................................................................................................................................................................. 37 Networking Equipment ..................................................................................................................................................... 39 Semiconductors ............................................................................................................................................................... 44 IT Services ....................................................................................................................................................................... 47 Cloud Computing Primer ......................................................................................................................................................... 54 The Spectrum of Public Cloud Solutions: SaaS, PaaS, IaaS .......................................................................................... 60 SaaS: Applications in the Cloud ...................................................................................................................................... 61 PaaS: Application Development and Deployment Platform in the Cloud ........................................................................ 67 IaaS: Infrastructure in the Cloud...................................................................................................................................... 70 Public Cloud Build-Out and Operating Cost Analysis....................................................................................................... 74 Private Clouds: Sharing on the Inside ............................................................................................................................. 78 Appendix I: Morgan Stanley Best Positioned for Cloud Migration Basket Constituents .......................................................... 86 Appendix II: Cloud Survey Results .......................................................................................................................................... 87 Appendix III: Quantifying the Cloud Impact on Server Growth ................................................................................................ 93 3
  4. 4. MORGAN STANLEY RESEARCHMay 23, 2011Cloud Computing Takes OffMeasuring Cloud Impacts: Stronger than Investors AnticipateCloud Growth Drivers Over the past five years, cloud computing has become one of the defining secular trends within technology, and we believePublic cloud workloads may increase at a 50% CAGR in the next the effects are just beginning to be felt across the industry.three years—about twice as fast as the market currently anticipates. This Blue Paper is the Morgan Stanley global technologyOur AlphaWise global survey of more than 300 IT decision-makers team’s most comprehensive effort yet to gather what we havesuggests workloads running in public cloud environments will grow at a learned thus far on what is driving the movement of workloads50% CAGR over the next three years, with 25% growth in managed to cloud computing environments and to measure andhosting workloads. forecast the effects of that movement.Cloud service providers growth will likely have a longer tail thanconsensus reflects. We expect to see improving breadth and depth of The bottom line: The migration of workloads to cloudadoption as the percentage of respondents running workload in the public environments will be one of the primary drivers of technologycloud grows from 28% today to 51% in three years and the percentage of spending over the next three years, producing significantworkloads running in the public cloud grows from 10.0% to 21.5%. growth opportunities for those companies positioned toLooking for best-positioned companies in the cloud arms dealers. provide cloud services or build out cloud infrastructures.We see explosive growth in public cloud servers (63% three-year CAGR) Conversely, cloud migration will present significant challengesand solid growth in managed hosting servers (more than 12%). Vendors for vendors tied to on-premise server environments.best positioned to benefit from the clouds build-out are likely to do well.Key Beneficiaries Migration to the public cloud reaching an inflection point. Our work suggests a 50% CAGR in the growth of public Accenture. Increasing assessments and IT strategy work are good for consultants, and Accenture is the best positioned here. cloud-based workloads over the next three years. Our analysis suggests robust growth not just for the newer, less- Broadcom. Broadcoms Trident family of data center switches will developed markets in the public cloud like Platform as a benefit the company as infrastructure builds increase due to the Service (PaaS) and Infrastructure as a Service (IaaS), but migration to cloud-based server environments. also the more mature Software as a Service (SaaS) segment Salesforce.com. SaaS workloads expected to sustain a 50% CAGR — where we forecast workloads growing at a similar 50% through 2014 are well ahead of the 11% new billings growth we CAGR. Within on-premise environments, the provisioning of model. With new apps ramping and a strong PaaS offering, workloads into private cloud or virtualized environment should Salesforce will be a key consolidator of cloud demand. see rapid expansion as well, growing from 32% of workloads EMC. Increased storage consumption in managed/cloud data today to 52% in three years. centers, an expanding partner ecosystem and product set, and balanced software-hardware mix make EMC a top pick. Exhibit 1 Our Survey Suggests a 50% CAGR in Public Cloud Juniper Networks. With one of the only truly flat data center Workloads over the Next Three Years… architectures today, QFabric arms Juniper with a first-to-market advantage for large-scale data center and cloud build-outs. Number of workloads (mil.) Quanta. Usage of lower-cost Asian original design manufacturers 10,000 (ODMs) among the large public cloud vendors is ramping up; Public Cloud Quanta is well positioned to gain share during the shift to the cloud. 7,500 +50% CAGR Rackspace Hosting. One of the only remaining pure plays in the +25 IaaS space, Rackspace appears firmly established among 5,000 Enterprise and small/medium businesses as a primary option for +8% CAGR cloud deployment. 2,500 VMware. Just 32% penetrated into x86 workloads, further 0 penetration of core server virtualization could sustain near-20% 2010 2013e growth in the core, while management, virtual desktop infrastructure (VDI), and public cloud drive above-consensus revenues. On-Premise Managed Hosting IaaS PaaS SaaS e=Morgan Stanley Research estimates. SM Source: AlphaWise , Morgan Stanley Research 4
  5. 5. MORGAN STANLEY RESEARCHMay 23, 2011Cloud Computing Takes OffBreadth and depth of public cloud adoption expanding Secular Cloud Challengesrapidly. In our AlphaWise survey, 28% of respondents Server spending the most negatively affected by the move to thereported that they run workloads in a public cloud today; in cloud. In our AlphaWise survey, 54% of our respondents cited serverthree years, that percentage is expected to expand by more hardware as a top-three area of cost savings from the move to cloudthan 80% to 51%. Accompanying this is an expansion in the computing. The average 8.6% expected reduction in server spendingdepth of penetration, with the percentage of overall workloads over the next three years due to the move to the cloud, dwarfs the 1.0%running in the public cloud growing at a 29% CAGR to 22% and 0.4% expected savings for storage and networking, respectively.from 10% today. On-premise server growth goes negative. A shift of workloads to more-efficient cloud environments and increased utilization of currentExhibit 2 server resources in private cloud environments push on-premise new…as Both the Breadth and Depth of Usage Expands server shipments to a -1% CAGR over the next three years in our model. Public cloud usage (%) Challenges for vendors tied to the growth of on-premise data 60 centers. Growth drivers are shifting as vendors try to incorporate public cloud strategies; those slow to move will see significant headwinds to growth. 40 Models in Flux  Brocade. Brocade offers a competitive fabric-based strategy, but its 20 ability to execute and penetrate large accounts remains a concern.  Cisco. Lacking a flat architecture for large-scale cloud deployments 0 in its portfolio, we believe Cisco remains in a defensive position. Current One year Three years  Capgemini. With about 50% of its outsourcing business (20% of Workloads Respondents utilizing total) coming from information technology outsourcing (ITO), we seeSource: AlphaWiseSM, Morgan Stanley Research potential for short-term headwinds.  Hewlett-Packard. Hewlett-Packard lacks a clear strategy to attackServer spending sees the biggest negative impact from cloud data centers with traditional server and networking products.this shift in the near-term. Given the combination of However, converged portfolio is taking share in on-premise dataworkloads shifting to public cloud environments and the centers.increasing utilization seen in on-premise server environments  Microsoft. Microsoft’s dominant share in server operating systemsfrom virtualization and private cloud technologies, our revised is almost solely in on-premise environments. However, its publicserver model now forecasts a -1% CAGR in on-premise cloud offerings polled the strongest of any vendor in our survey.server shipments over the next three years. This trend will  Red Hat. While well positioned for the cloud build out, Red Hat’sresult in near-term beneficiaries among vendors supplying the current subscription base is largely tied to on-premise deployments,build-out of public and private cloud environments and longer- and its virtualization, PaaS, and IaaS offerings are nascent.term challenges for vendors tied strongly to on-premise  SAP AG. A ramp in the BBD reseller network is likely to drive highercorporate server demand. top-line growth and meaningful revenue contribution for the group. We estimate business-by-design (BBD) revenues at €83 million inExhibit 3 2012e (less than 1% of group SQL server reporting services),…While On-Premise Server Shipments Stagnate reaching about €900 million in 2015e, about 10% of group SSRS. Server shipment growth — by environment YoY growth (%) Potentially Secularly Challenged 100  Atos Origin. With about 40% of its sales derived from ITO, we see 80 the company facing material booking/revenue headwinds in the near 60 term. 40  Dell and QLogic. The demand shift to cloud service providers 20 purchasing from Asian ODMs puts server growth at risk. 0  Symantec. About 50% of company revenues are tied to on-premise -20 servers and storage, and SaaS-based businesses represent just 2010 2011e 2012e 2013e 2014e 11% of revenues. On Premise Managed Hosting Public Cloud Totale=Morgan Stanley Research estimates.Source: Morgan Stanley Research 5
  6. 6. MORGAN STANLEY RESEARCHMay 23, 2011Cloud Computing Takes Off SUMMARY OF KEY TAKEAWAYS BY INDUSTRY Hardware As companies shift from on-premise to managed or public cloud environments, x86 server share leaders are at risk of market share loss to Asian ODMs.  Hewlett-Packard and Dell, x86 market share leaders, are most at risk from a depressed on-premise server demand environment.  Asian ODM’s like Quanta and Wistron that ship directly to the largest cloud service vendors will outgrow premium vendors and continue to gain market share.  Savings reaped from virtualization and the cloud will be spent on technologies that can improve growth and/or profitability; data analytics is likely to do well. Storage Storage spend will continue to benefit from increased consumption in managed/cloud IT data centers, with only a mild offset as companies decrease on-premise storage spend.  We prefer EMC for its recent broadening of the product suite, partner ecosystem expansion, and more balanced software/hardware mix (which can reduce gross margin volatility) as compared with peers. Semiconductors Cloud build-outs drive the need for greater data center switching, security, and processing power.  Broadcom is our preferred play on growth in the cloud-based data center switch market.  The compute resources needed to meet security requirements from cloud-build outs increase by greater than 10 times; Cavium Networks benefits most from the higher compute needs. Software Public cloud growth will be stronger and longer lasting than investors expect; at the same time, a stagnant on-premise server market means shifting growth drivers for many infrastructure vendors.  A 50% CAGR in public cloud workloads should drive the upside to our current models for cloud service providers; vendors able to consolidate demand like Salesforce.com and SuccessFactors will be the key beneficiaries, in our view.  A 20% CAGR in virtualized servers sustains VMware’s core business, while investments in management, virtual desktop, and public cloud infrastructures become larger growth drivers.  Growth drivers are shifting for Red Hat, as server growth shifts significantly towards cloud environments.Network Equipment Adoption of cloud architectures should drive the off-premise data center switching market to a 21% CAGR through 2015, leading the total data center market to 10% growth over the same period.  As cloud adoption grows, a share shift toward higher-scale public cloud and managed hosting data center switching equipment should drive off-premise switching to grow at a 21% CAGR, to $3.5 billion, through 2015.  We forecast the much larger on-premise data center switching market, which in 2010 stood at 25% of the total Ethernet market, to grow at a far more subdued 5% rate over the same period.  Juniper Networks is our preferred play on the cloud build out, as QFabric gives the company a first-to- market advantage for large-scale data center and enterprise cloud build-outs. Telecom Services The large cap telco sector has prioritized cloud computing as a way to enhance wireline growth.  Telco cloud offerings continue to lag larger IT integrators and cloud specialists for mindshare among IT decision makers.  We see the recent spate of acquisition activity as a means to accelerate cloud growth and broaden the telco solution set to more fully encompass cloud-based offerings.  Rackspace appears firmly established among enterprise and small/medium businesses as a primary option for cloud deployment. IT Services The emergence of cloud computing forces CIOs to redesign their fundamental IT strategies, creating opportunities for firms that can assist in the transition.  Increasing assessments and IT strategy work is good for consultants; Accenture is the best-positioned company here.  Cognizant Technology Solutions’ on-site investments give the company a horse in the race and a better position for outsourcing work.  CSC has new “as a service” offerings and incorporates the cloud into all of its new ITO pursuits, which is a positive and will be critical to the company’s long-term success. 6
  7. 7. MORGAN STANLEY RESEARCHMay 23, 2011Cloud Computing Takes OffWhat We Did and WhyBlue Papers are collaborative reports focusing on key secular enterprise. High network and compute demands ofthemes transcending both sectors and geographies, where service providers should benefit semiconductor providersMorgan Stanley looks to define the key debates and give that focus on high signal integrity at high data speedsinvestors a clearer understanding of what will define the with low power consumption. Finally, transitions in ITcompanies most likely to benefit from or be challenged by architectures are a positive for consultants, andthose trends. Within the technology industry, cloud computing Accenture is best positioned to benefit here.represents a key secular theme, and one that has significantimpacts across all sectors. Admittedly, cloud computing is a 4) Cloud computing primer. For investors coming up totopic on which much has been written. What differentiates this speed on the secular trend or looking for a refresher, ourproject is its focus on analyzing and quantifying the effects of primer offers the most comprehensive report yet on thethe migration of application workloads to cloud environments. markets, players, and emerging trends in cloudThis measurement enables us to estimate better the computing.magnitude and longevity of cloud impacts and to understand Overview of Key Conceptsthe derivative impacts on existing technologies. There are fourmain components to this project:  Workloads versus servers. Workloads represent units of application functionality that run on a server, most often1) Global survey of cloud computing usage intentions. complete applications. Historically, in x86 server The basis for our analysis is a global AlphaWise survey environments companies would run just one application of 309 IT decisions makers on where application workload per server. The advent of server virtualization workloads are currently run, where they expect workloads software for x86 CPUs and more powerful CPUs designed to run going forward, and what the spending impacts from to better enable virtualization allowed companies to run the migration of workloads to cloud environments might multiple workloads per server effectively. In virtualized be. environments, a workload is often equated to a virtual machine (VM). Public cloud vendors make use of grid and2) Server model revised to a workload-based multi-tenant architectures, in addition to virtualization, to framework. Servers are the base unit of data center enable multiple workloads to run on a single server. build-outs in this exercise. Our revised server model incorporates the effects of higher workload densities in  Workload density. This indicates the average number of virtualized and grid environments and explicitly forecasts workloads run on a server. IDC estimates an average of 6.6 server shipments by on-premise, managed hosting, and workloads being run per server in virtualized environments public cloud environments to understand better the today. Industrywide, we estimate that close to 20% of effects these technologies and the shifting deployment servers are running multiple workloads today, carrying mix has on overall server demand. Our three-year unit close to 50% of total workloads. However, given that 80% CAGR forecast drops from 6% to 3.4%. of servers still run one workload, the industrywide workload density is closer to 1.5.3) Impacts on ancillary technologies and stocks. Analyst  On-premise versus private cloud. On-premise refers to from all technology sectors globally took the conclusions the ownership and location of a company’s data center or of the global survey and our analysis of the server servers, owned and managed by the customer and located impacts and reflected how these trends would affect the either on its own premises or in a colocation facility. Private companies in their sectors. What appears to be a clear cloud refers to a set of technologies that enable the pooling negative for x86 server market share leaders like Dell of compute resources used in on-premise environments. and Hewlett-Packard is likely a strong market opportunity We estimate that only 17% of on-premise servers utilize for Asian ODMs like Quanta and Wistron. VMware has any private cloud technologies. plenty of headroom in its core server virtualization space, and as investments in management, virtual desktop, and  Public cloud versus managed hosting. Both are third- the public cloud begin to ramp up, Juniper has a party hosted compute environments. However, in public significant first-mover advantage in aligning its switching cloud environments compute resources are shared product strategy to the needs of public cloud service dynamically among all customers, where as infrastructure is providers over Cisco, which continues to focus on the 7
  8. 8. MORGAN STANLEY RESEARCHMay 23, 2011Cloud Computing Takes Off dedicated to a particular customer in managed hosting. Growth in workloads a clear positive for cloud service providers. A 50% growth rate for workloads run in public cloud environments is a clear positive for cloud serviceKey Macro Conclusion providers and significantly above market expectations —Public cloud to show robust growth over the next three particularly in more mature segments like SaaS, where ouryears. Currently, only 28% of IT managers globally report analysis indicates workload growth can sustain a 50% CAGRrunning workloads in any type of public cloud environment. through 2013. This type of growth compares favorably to theSaaS deployments are most common (18%), with PaaS and 11% CAGR in new billings growth we currently forecast forIaaS both being used by just under 14% of respondents. Over SaaS market leader Salesforce.com and would suggestthe next three years, both the breadth and depth of public considerable upside in the model. The strongest growth iscloud usage is expected to expand greatly, with a 23% CAGR expected from the IaaS market at a 54% CAGR. Amazon,in the number of respondents who expect to utilize the public with its broadening portfolio of compute services and largecloud for some percentage of their application workloads and partner ecosystem, is likely to continue benefiting here. Whilea 29% CAGR in the percentage of workloads expected to be public cloud represents only 14% of Rackspace revenuesrun in public cloud environments. This 29% CAGR in the today, success with this offering is more likely to affect itspercentage of workloads run in public cloud environments, stock price than similar success would affect Amazon’s. Weplus the 16% CAGR we forecast for workloads overall, yields currently model a 46% CAGR in Rackspace’s cloud business,50% growth in public cloud workloads. a forecast well supported by our analysis. By comparison, industry analyst groups Gartner, IDC, and 451 forecasts aExhibit 4 25% three-year CAGR in the total public cloud market.Global Survey Reveals a Robust Migration ofWorkloads to Public Cloud Environments… Exhibit 6 All Public Cloud Options Grow Well: Durability of Workloads (%) 3-Yr CAGR SaaS Growth Perhaps the Biggest Surprise 100 33% Workloads (%) 29% 80 26% 25 8% 29% CAGR 60 20 33% 40 -7% 15 CAGR 20 29% 10 CAGR 0 Today One year Three years 5 26% CAGR On-Premise Managed Hosting PaaS SaaS IaaS 0Source: AlphaWiseSM, Morgan Stanley Research Today 1-year 3-years PaaS SaaS IaaSExhibit 5 Source: AlphaWiseSM, Morgan Stanley Research…and Broadening Usage, with Over 50% Expectingto Use Public Clouds within Three Years Vendors supporting the build-out of public cloud Respondents using the public cloud (%) environments are key beneficiaries. While our analysis 60 focuses on servers, all the elements associated with the build- +23% CAGR out of public cloud environments are likely to benefit. Within 40 +36% servers, public cloud vendors tend to favor Asian ODMs like +18% +29% Quanta and Wistron over tier-1 vendors like Hewlett-Packard +25% +29% +56% and Dell, but within networking and storage we found a more +50% 20 robust opportunity for tier-1 vendors. Juniper has a clear time- to-market advantage with its QFabric solutions geared towards service providers and large-scale data center build- 0 outs; we believe QFabric will help Juniper to sustain nearly Current 1-year 3-years SaaS PaaS IaaS Any public cloud 20% growth in infrastructure products through 2013.Source: AlphaWise SM , Morgan Stanley Research Continuous growth in data volumes and clearly defined cloud 8
  9. 9. MORGAN STANLEY RESEARCHMay 23, 2011Cloud Computing Takes Offstrategies should help EMC take share, no matter the location Exhibit 8(cloud or on-premise). VMware has invested heavily in …Driving Higher Utilization Rates in On-Premisebuilding out its suite of cloud solutions, from the full PaaS and Colocated Serverssuite in Cloud Foundations to actual SaaS applications with Workloads (%)Zimbra and SlideRocket; these solutions should give the 65company strong legs for growth in the public cloud. 63On-premise environments getting more efficient. At the 60 59same time that workloads are being deployed more often fromhighly efficient public cloud environments, the efficiency of on-premise server environments is improving as well, due to the 55increased usage of virtualization and other private cloud 53technologies. Survey respondents report 32% of their on-premise workloads are running in virtualized or private cloud 50environments today and they expect that percentage to Today One year Three yearsexpand to 52% in three years, driving a 33% CAGR in Source: AlphaWiseSM, Morgan Stanley Researchvirtualized workloads. Non-virtualized workloads remainessentially flat during this same period. The increased usage Servers expected to be the greatest source of costof private cloud technologies is driving server utilization rates savings. In our survey questions about expected net costsignificantly higher as well. While the absolute self-reported changes in spending on key data center components andutilization rates here seem high as compared with the levels about work areas where respondents expect to see thereported by the industry experts we have consulted, the 17% largest cost savings, server hardware was shown clearly to beimprovement in expected utilization reported by IT managers the technology that will receive the biggest blows in the moveis directionally very significant. to cloud computing. Respondents moving workloads to theExhibit 7 public cloud expect on average an 8.6% decline in serverIncreasing Penetration of Virtualization and Private spending over the next three years, well ahead of the 1.0%Cloud Technologies into x86 Workloads… average decline in storage spending and 0.4% decline in Workloads (%) networking spend. This result was corroborated by more than 60 50% of respondents who cited server hardware as a top three 52 area of expected cost savings due to the move to cloud computing. 43 40 Exhibit 9 32 Three-Year Decline in Server Spending Because of Cloud Migration… 20 Net impact of spending (%) 0 -0.4 0 -1.0 -2 Today One year Three yearsSource: AlphaWiseSM, Morgan Stanley Research -4 -6 -8 -8.6 -10 Servers Storage Networking Source: AlphaWiseSM, Morgan Stanley Research 9
  10. 10. MORGAN STANLEY RESEARCHMay 23, 2011Cloud Computing Takes OffExhibit 10 Exhibit 11…as More Than 50% of Respondents Cite Server Multiple Compressive Forces Weighing on ServerHardware as a Top-Three Area for Cost Savings Shipment Growth 0 5 10 15 20 25 30 35 40 45 x86 Server shipment growth (%) Y/Y growth (%) Data security 25 Cost – uncertain savings 15 Shift Virtualization to the Multi-tenant Loss of Control (upgrades, & Grid cloud architecture timing of backups, etc) 5 Regulatory or Compliance -5 Reliability (SLA requirements) -15 Data portability / -25 ownership 2008 2009 2010 2011e 2012e 2013e 2014e Source: Morgan Stanley Research Software compatibility Exhibit 12 Performance On-Premise Server Growth Expected to be Stagnant, Lock-in (ability to change Largest barrier but Cloud Sees Extensive Build-out Second largest barrier providers) Third largest barrier YoY Server Shipment Growth — By Environment Other Growth (%) 100Source: AlphaWiseSM, Morgan Stanley Research 80 60Several forces compressing on-premise server spending. 40After seeing a strong rebound in the economic recovery, we 20believe a combination of compressive forces acting upon the 0server market will keep overall server shipment growth below -20historical levels. Our revised server model now forecasts a 2010 2011e 2012e 2013e 2014e3% CAGR in total server shipments through 2014, down from On Premise Managed Hosting Public Cloud Total e=Morgan Stanley Research estimates.6% previously. This reflects a -1% CAGR in on-premise Source: Morgan Stanley Researchservers, offset partially by very strong growth in public cloudenvironments (54% CAGR) and decent growth in managed Weak server growth a significant headwind for vendorshosting environments (12% CAGR). still heavily leveraged to on-premise environments. The tier-1 server vendors like Dell and Hewlett-Packard are theWe estimate that on-premise servers make up about 87% of most obvious names likely to face this challenge. However,total server shipments today, and we expect that on-premise across sectors there are vendors still significantly tied to theservers will still account for 74% of overall server shipments in growth of on-premise data centers. Host bus adapters (HBAs)2014, even with the shift in workloads toward public cloud and represent over 70% of QLogic’s revenue, tying them closely tomanaged hosting. Thus, we anticipate that the build-out of server shipments in non-cloud environments. While buildingpublic cloud infrastructures will have a relatively small impact out their public and private cloud offerings, we would estimateon overall server growth. that Red Hat still derives about 70% of revenues from server operating systems in on-premise environments. With an estimated 80% data center share today and no flat architecture for large-scale cloud deployments, we believe Cisco remains in a defensive position. 10
  11. 11. MORGAN STANLEY RESEARCHMay 23, 2011Cloud Computing Takes OffOur Cloud Analysis Drives Evidencethe Following Rating ChangesPlease see the reports published concurrently with this Core Questions for Evidence ResearchBlue Paper for details on our changes in  What is the current penetration of cloud computing and howrecommendation for the following names. In addition, is it expected to change in the next one to three years?Teradata has been added to the Morgan Stanley Best  What is the potential impact on on-premise servers andIdeas list. server-related markets?  Who are the likely beneficiaries of greater adoption of cloud Ticker Rating computing? CRM EW  OW TDC EW  OW  What are the key barriers to the adoption of cloud environments? HPQ OW  EW QLGC EW  UW The Evidence Majority of respondents are already using some sort of cloud and expect more than one third of their workloads to be in the cloud or managed hosting in three years  Public cloud is expected to increase its share of workloads from today’s 10% to 22% within the next three years.  Managed hosting’s share is likely to increase from today’s 11% to 14% of all workloads during the same period. Servers are likely to be negatively affected by migration to cloud, but storage and networking are less so.  Reduced use of on-premise servers by 18% over the next three years could drive cuts in server spending by 9%. Microsoft is likely to gain the most from a broader adoption of public cloud environment.  Of respondents, 68% expecting to move workloads to cloud mention Microsoft as the vendor of choice.  IBM is the most mentioned among those moving workloads to managed hosting. Security concerns are the dominant barrier to adoption for 43% of respondents.  Costs/uncertain savings is the second most-often mentioned reason (38% of respondents) for not migrating to a cloud environment.  Fewer respondents believe that software compatibility (15%) or cloud’s performance (13%) are areas of concern. What Gives Us Confidence  Breadth of the survey. We conducted 304 phone interviews with decision-makers of IT strategy and policy (US, 103; Europe, 101; and Asia-Pacific, 100), representing 106 companies have 1 to 500 employees and 198 have more than 500 employees.  Timeliness of the data. Fielding conducted March– April 2011. 11
  12. 12. MORGAN STANLEY RESEARCHMay 23, 2011Cloud Computing Takes OffMeasuring Cloud Impacts: The Coming Server Squeeze Exhibit 13Key Points Mounting Pressures from the Shift to Cloud Increasing utilization rates industrywide create significant headwinds Computing Will Weigh on Server Growth for server shipment growth. We now forecast servers growing at just (%) Millions 20 10 a 3.4% three-year CAGR, down from our 6% prior forecast. 15 These headwinds are exacerbated by the movement of workloads to 10 8 cloud environments. Our survey suggests the percentage of 5 workloads running in on-premise environments falls from 79% today 0 6 to 62% in four years. -5 4 Offsetting stagnant server shipment growth into on-premise -10 environments is a 12% three-year CAGR for server shipments into -15 2 managed hosting environments and a robust 54% CAGR for servers -20 shipping into public cloud environments. -25 0 2005 2006 2007 2008 2009 2010 2011e 2012e 2013e 2014e Total Servers Shipped Y/Y Growth %Our global survey of over 300 IT managers suggests a steady e=Morgan Stanley Research estimates. Source: IDC, Morgan Stanley Researchshift in workloads moving out of company owned, on-premisecompute environments to managed hosting and public cloud Our revised server model uses application workloads as theenvironments. We estimate that more than 20% of workloads basis for our estimates. Analyzing the deploymenthave already moved off premise during the last six years, environment and number of workloads per server (density)reaching a key inflection point for accelerated adoption. (Past provides unique insights on the shifting dynamics of servertechnology cycles have accelerated after reaching a 20% units growth, in our opinion. We base our new server modeladoption rate.) By 2014, we expect only 62% of workloads will on four key variables:be run on on-premise. Our work suggests that servers inmanaged hosting and public cloud environments run at higher 1) Workload installed base growth. As serverutilization rates than typically found in on-premise environments become increasingly virtualized andenvironments. Thus, the movement of workloads to the cloud, workloads shift to high utilization cloud environments, thealong with the rising utilization rates seen in on-premise traditional drivers of server growth (application growthdeployments from the adoption of virtualization and private and refresh) become more disaggregated from actualcloud technologies, creates significant headwinds for server server shipments. By changing the basis of our forecastunit growth — particularly in on-premise environments. to workloads, we believe we can better understand how these shifts affect server shipments. In 2010, we estimateMoving to a workload based server model. We have re- there were 56.9 million workloads running worldwide,engineered our server model to forecast server growth based on the sum of the last four-and-a-half years ofdiscretely in: 1) on-premise, 2) managed hosting, and 3) workload shipments, as defined by IDC. Conversationspublic cloud environments — as based on the shift of with industry analysts suggest annual workload growthworkloads into these environments. Additionally, we forecast has held relatively consistent at 14-15%. Improvementsthe effects of increasing rates of virtualization within these in application development and deployment tools inserver bases. Bottomline, over the next three years we expect recent years is lowering the time to market for newthat on-premise server shipments will decline at a 1% CAGR application workloads. At the same time, the availabilitywhile shipments into managed hosting environment grow at a of public cloud infrastructure lowers the upfront capital12% CAGR and shipments into public cloud environments at expense of deploying new application workloads. As wea 60% CAGR. Overall, our server unit growth forecast for have seen in other cycles, new enabling technologies can2011 drops from 6% to 4.1%, below the five-year average of spur growth of use cases to take advantage of the new5% and well below the strong rebound to 16% growth in 2010. capabilities (e.g., higher bandwidth internet connectionsOver the next three years we expect overall server shipments leads to VoiP). Thus, we believe workload growth canto grow at a 3.4% CAGR as headwinds from the shift to the actually accelerate over the next several years.cloud continue to mount. 12
  13. 13. MORGAN STANLEY RESEARCHMay 23, 2011Cloud Computing Takes OffExhibit 14 3) Percentage of servers running virtualization. WeBetter Enabling Technologies (Cloud/Frameworks) estimate on-premise data centers use the least amount ofCan Spur Higher Growth in Workloads virtualization software as many mission-critical Workloads (mil.) applications and workloads remain on dedicated servers 120 18 within a company’s own data center. We estimate the use 16 100 of virtualization or shared compute technologies 14 80 12 increases at managed hosting sites and is 100% in public 10 cloud environments. 60 8 40 6 Exhibit 16 4 Increasing Virtualization Adoption at All Data 20 2 Center Locations 0 0 Servers with virtualization software (%) 2007 2008 2009 2010 2011e 2012e 2013e 2014e 100 Total workload installed base Y/Y growth %e= Morgan Stanley Research estimates. 80Source: IDC, Morgan Stanley Research 602) Location of workloads: On-premise, managed hosting, or public cloud. Location is important because 40 of the variations in workload density seen in different environments and the contrasts in buying behavior 20 between enterprise customers and service providers. In general, our work suggests that public clouds run more 0 workloads per server and use more shared compute 2005 2006 2007 2008 2009 2010 2011e 2012e 2013e 2014e technologies (e.g., virtualization, grid, and multi-tenant On Premise Managed Hosting Public Cloud architectures), leading to high rates of unit consolidation. e=Morgan Stanley Research estimates. Source: IDC, Morgan Stanley Research On the other hand, on-premise data centers run more applications on standalone, dedicated servers (one-to- 4) VM workload-to-server ratio. Whether a shared one relationships), which do not compress server units. compute environment (like a multi-tenant architecture) or Service providers are more likely to use white box actual VMs running on servers, the consolidation effect servers or build their own servers, and enterprises are on servers is similar. Our conversation with public cloud more likely to look for a full-service vendor for on-premise providers and industry experts suggests that the server deployments. consolidation rates in the public cloud tend to be higher than in managed hosting or on-premise environments.Exhibit 15Workloads are Moving Away from On-Premise There are various reasons for this: 1) within SaaSEnvironments environments the applications are optimized to run on Workloads (%) their particular infrastructure; 2) many cloud applications100 tend to be more lightweight than applications running in on-premise environments; and 3) the management skill 80 sets within cloud providers tend to be higher than those in commercial on-premise environments. 60 40 20 0 2005 2006 2007 2008 2009 2010 2011e 2012e 2013e 2014e On Premise Managed Hosting Public Cloude=Morgan Stanley Research estimates.Source: IDC, Morgan Stanley Research 13
  14. 14. MORGAN STANLEY RESEARCHMay 23, 2011Cloud Computing Takes OffExhibit 17 Exhibit 19Public Cloud Data Centers Achieve the Highest Base Case Implies 3.4% Server Shipment CAGR,Levels of Workload Consolidation 8.9% in Bull Case, -3.9% in Bear Case VMs per Server Server shipments (mil.) 10 12 10 8 8 6 6 4 4 2 2 0 2010 2011e 2012e 2013e 2014e 0 2005 2006 2007 2008 2009 2010 2011e 2012e 2013e 2014e Bear Case Base Case Bull Case On Premise Managed Hosting Public Cloud 10-14 CAGR: 10-14 CAGR:e=Morgan Stanley Research estimates. 10-14 CAGR: -3.9% 3.4% 8.9%Source: IDC, Morgan Stanley Research e=Morgan Stanley Research estimates. Source: Morgan Stanley ResearchScenario analysis and sensitivities. Our base case servermodel rests firmly on the expected workload deployment Exhibit 20characteristics reported in our survey work for today, one year Server Growth Could Continue in the 7-12% Rangefrom now, and three years from now. Some of the benefits of in our Bull Case, -3% to -5% in Bear Caseour new workload-based server model is to be able to: 1) test (%)the sensitivities of server growth to the pace of some of these 16key technology trends we see in the industry, and 2) be able 12to develop bull and bear case scenarios for server growthbased on potential changes in these key variables. 8Exhibit 18 4Key Assumptions behind our Server Model 0 2009 2010 2011e 2012e 2013e 2014e -4 Workload Growth (%) Bull 15.7 17.0 18.5 20.4 20.5 -8 Base 15.7 16.0 16.2 16.7 16.8 2010 2011e 2012e 2013e 2014e Bear 15.7 15.0 14.0 12.5 11.8 Bull Case Base Case Bear Case VM Workload per Server e=Morgan Stanley Research estimates. Bull 5.9 6.2 6.5 6.8 7.0 7.3 Source: Morgan Stanley Research Base 5.9 6.2 6.5 6.8 7.0 7.3 Bear 5.9 6.2 6.7 7.2 7.6 8.4 As we look through the key sensitivities in the model, the two On-Premise Workload (%) most important assumptions to server growth relate to the Bull 83 79 76 72 68 64 percentage of workloads deployed in on-premise Base 83 79 75 69 64 59 Bear 83 79 74 68 62 57 environments and the mix of virtualized (hosting VMs) versusSource: Morgan Stanley Research non-virtualized (non-hosting VMs) servers. All else being equal, a lower percentage of workloads being deployed on- premise lowers overall server shipment growth. This is because public cloud providers like Salesforce.com are able to compress multiple workloads onto each of their servers (in the case of Salesforce.com, up to 30:1); any migration from on-premise environments will result in lowered server 14
  15. 15. MORGAN STANLEY RESEARCHMay 23, 2011Cloud Computing Takes Offdemand. In our model we have assumed a 8.5:1 Exhibit 22compression ratio in 2011 for public cloud environments, Higher Percentage of Servers Used in Virtualizedversus an average of 2.0 workloads per server in on-premise Environments Reduces Overall Server Growthenvironments (virtualized servers run 6.5 workloads, and non- 2011 2012evirtualized run 1.0). Therefore, where 8.5 workloads would Serverg Serverrequire 4.25 servers in an on-premise environment, those rowth growthsame workloads only require one server now in the public Physical server 85.7 14.9 83.8 13.5 shipments as % 84.7 12.0 82.8 10.8cloud. of total server 83.7 9.2 81.8 8.2Exhibit 21 shipments 82.7 6.6 80.8 5.8Increased Shift from On-premise Environments 81.7 4.1 79.8 3.5Reduces Overall Server Growth 80.7 1.8 78.8 1.4 2011 2012e 79.7 -0.5 77.8 -0.7 Server Server 78.7 -2.6 76.8 -2.6 growth growth 77.7 -4.6 75.8 -4.5 e=Morgan Stanley Research estimates. % of Workloads 76.8 10.6 71.4 10.2 Source: Morgan Stanley Research on-premise 76.3 9.0 70.9 8.5 75.8 7.4 70.4 6.9 Finally, as we look at the VM-per-host compression rate of 75.3 5.7 69.9 5.2 6.5:1 mentioned above, we see that it too is highly relevant, 74.8 4.1 69.4 3.5 though it would have to have a more significant movement to 74.3 2.5 68.9 1.9 make as much of an impact. For instance, an adjustment of 73.8 0.8 68.4 0.2 roughly 6% to the VM-per-host ratio—from 6.5:1 to 6.9:1— 73.3 -0.8 67.9 -1.4 only changes the server growth rate by about 3%. However, 72.8 -2.5 67.4 -3.1 in the prior two sensitivities, we see that an adjustment of onlye=Morgan Stanley Research estimates.Source: Morgan Stanley Research 2% would have just as much of an impact, demonstrating the importance of those first two assumptions in the server model.Likewise, as we alter the mix of virtualized versus non-virtualized servers being shipped, we see a similar level of Exhibit 23volatility with small changes in the assumption. Similar to the Increased VM Density per Server Reduces Overallway public cloud providers compress workloads: With 6.5 Server Growthworkloads running on one virtualized server (as our forecast 2011 2012esuggests), an increased demand for virtualized servers will Server Serverlead to a significant reduction in demand for physical servers growth growth (%) (%)(at a 6.5:1 rate in 2011). For every 1% point reduction in the VM per host 6.9 1.0 7.2 0.6mix of non-virtualized server shipments as a percentage of 6.8 1.8 7.1 1.3the total, we would expect to see an impact of more than 2% 6.7 2.5 7.0 2.0on the overall server market growth rate. 6.6 3.3 6.9 2.8 6.5 4.1 6.8 3.5 6.4 4.9 6.7 4.3 6.3 5.7 6.6 5.1 6.2 6.6 6.5 5.9 6.1 7.4 6.4 6.8 Source: Morgan Stanley Research 15

×