The LAN market: segments at a crossroads

Although high-growth opportunities abound outside the United States for the sale of basic connectivity to first-time corporate clients, in order for strong growth in the local-area network (LAN) market to continue vendors must focus on emerging opportunities driven by new technologies and emerging customer segments. Perhaps the most fundamental factor driving this change is that the corporate market (especially in the United States) is approaching saturation. Basic connectivity (file and print services, e-mail, etc.) needs are fully understood and widely implemented by corporate IS. Corporate users are more concerned about next-generation issues like connecting as yet unconnected portions of the enterprise (remote offices) and upgrading existing connections to enable new applications. Outside corporate IS, significant potential exists in the small-business, small-office /home-office (SOHO), and home markets.

To capitalize on emerging technologies, IT firms must position themselves for success without alienating existing customers and jeopardizing revenue from markets in decline. For example, although 3Com needs to offer switching solutions, it doesn't want to emphasize this technology at the expense of its significant hub business. This strategy may sound simple, but its execution is not. Not only do new products and technologies complicate the sales process, but existing customers may not initially understand such products and/or may feel threatened by them. Geographical differences in technology adoption can make this already difficult balancing act particularly troublesome.

To capitalize on new customer segments, firms must offer complete solutions targeted at specific customer segments and needs. This is not to say that every customer type requires a different product. In some cases this may be true, but even when it is not, different distribution channels are likely required (each of which have their own requirements), as well as differentiated pricing, packaging, etc. Many IT vendors have missed this point and have tried to sell the same product, through the same channels, to the year's hot new customer segment. If a product is not positioned distinctly -- if not in feature/technical terms, then in merchandising terms -- it will likely fail to capture this new customer segment and will dilute the overall marketing message.

Despite their presence at this crossroads, LAN hardware and software markets continue to grow impressively. From a base of more than $17.1 billion in worldwide revenue in 1993, IDC forecasts the LAN segment to expand to a whopping $36.3 billion market in 1998 (a compound annual growth rate of 11%). Not only is this growth driven by the ongoing penetration of LANs into large businesses -- as they continue to grow, extend, and upgrade their networks -- but also by the implementation of LANs in small and medium-sized businesses (and even in some homes and home offices).

Of the segments that make up the LAN market, backbone switches and servers are among the fastest growing in terms of revenue, with 1995 to 1998 CAGRs of 81% and 22%, respectively. The server market's continued robust growth is particularly outstanding since it is also the largest LAN market segment. Given average server prices of $10K and up, its status shouldn't be too surprising; no wonder PC vendors are stumbling over one another to increase their business in this segment.

Several LAN segments, however, will start to decline, not because of weak overall demand, but in response to technology shifts. After peaking in 1997, the PC network interface card (NIC) will fall off, victim of market saturation. Both the NIC and hub segments are shifting to higher-speed technology, which may provide a short-term window of opportunity for new or existing players to gain share.

PC LAN Operating Systems

Of the segments that make up the overall LAN market, the PC LAN operating-system (OS) segment is arguably the most mature. The segment is, however, far from stable. To the contrary, the LAN OS market is changing so dramatically that today's key players, including Novell and Microsoft, are fighting for their lives. Perhaps the strongest force driving this change is the broader architectural shift to a client/server architecture.

The need for basic file and print services is widely understood and implemented in businesses; nevertheless, users are demanding more from their LANs (and the OSs that run them). A client/server architecture, for example, demands more at both ends of the wire: both the server and the client must be smarter and more capable. Viewed another way, the pendulum of user focus is swinging back -- the focal point of user interest has shifted from the extremes of the server-centric mainframe era and the desktop-focused PC epoch to a point somewhere in between. The emergence of workgroup applications (like Lotus Notes) and network-capable desktop OSs (like Windows 95) exemplify this trend. Such tools blur the distinction between "desktop" software, like applications and desktop OSs, and "network" software to the extent that users may question what general IS infrastructure, not to mention LAN OSs, their information needs require.

Microsoft is aware that network and desktop software are in this sense converging. In an effort to capitalize on this trend and sell more Windows NT Advanced Servers (NTAS), it is leveraging its Windows (both 3.x and 95) desktop operating systems, its Office application suite, and its software development tools (like Visual Basic) to the hilt. For example, in order for software to be eligible for the Windows 95 logo, it must be written to a specification that makes certain it also runs under Windows NT and thus ensures a wealth of NT applications and developer support.

It should come as no surprise that Microsoft is employing a particularly aggressive strategy to push Windows NTAS. Despite a presence in the LAN OS market for many years, through LAN Manager and now Windows NT, Microsoft has failed to achieve real success. Since the LAN OS market represents the key to the enterprise for Microsoft, this shortcoming is particularly nagging. In order for Microsoft to go to the next level -- to move beyond the desktop and into the enterprise -- it must steal share from Novell, the LAN OS market leader.

Novell is leveraging the directory service functions in NetWare 4.1, highlighting the ability to develop and manage complex distributed network applications. The company's extensive installed base, combined with the mission-critical nature of its networks, will make Novell extremely difficult to overcome under any circumstances. In order for Novell to leverage its distributed network strategy, it must do a better job of getting users to upgrade to NetWare 4.x, something the vendor has been struggling with for years.

Overall, the LAN OS market grew 19% in terms of licenses, but decreased 4% in end-user sales value because of strong pricing pressure and a general transition to lower-priced products. During 1994, 1.0 million PC LAN OS licenses, representing 19.5 million PC nodes, were shipped worldwide. The worldwide dollar value of the licenses shipped in 1994 was $2.3 billion, down from $2.4 billion in 1993. Growth is expected to slow after this year with worldwide revenue expansion falling from 15% in 1995 to 4% by 1998. By 1999, IDC forecasts that 180.0 million PCs (installed) will be running a network operating system and that the dollar value of PC network operating-system shipments will exceed $3.8 billion.

Market Shares: No Major Shifts

In terms of market share, there were no significant shifts in the PC LAN OS segment in 1994. Novell's NetWare continued to account for approximately two-thirds (65.6%) of nodes shipped worldwide. Although NetWare 4.x shipments increased, sales were somewhat disappointing. NetWare 3.x remained Novell's dominant product, while 2.x shipments were minimal.

Microsoft increased its total market share to 9.4% of worldwide 1994 nodes on the basis of increased use of Windows for Workgroups as a peer-to-peer network operating system and sales of Windows NT Advanced Server. Windows NTAS, utilized as a file and print server, totaled 52,000 licenses (5.5% market share). Windows 95 should accelerate the peer-to-peer trend given its ease of use, plug-and-play architecture, and robust network support. In fact, Windows 95's networking may be the product's sleeper feature, as it is a dramatic improvement that is infrequently mentioned.

In 1994, IBM OS/2 LAN Server increased its sales to 92,500 licenses and 1.6 million nodes worldwide (9.1% share). Meanwhile, Artisoft's LANtastic garnered sales of 683,000 nodes worldwide in 1994, earning a 3.5% market share, while Banyan's Vines accounted for 1.9 million nodes, or 6.1% share.

Definitions

IDC defines a PC LAN operating system (OS) as software that enables PCs attached to local area networks to communicate with the network server (whether dedicated or nondedicated) and with other PCs on the network. PC LAN OSs must have the following functionality:

* File sharing among the PCs on the network

* File storage and retrieval to and from the file server

* Shared print services on the network

Additional features generally considered by IDC to be part of a PC LAN operating system include the following:

* Ability to run and customize multi-user (network) applications between multiple PCs or between PCs and the server. LAN electronic mail and database management systems are the prime examples of these applications.

* Network administration, directory, management, and security functions.

* Provision for LAN-to-LAN communications and gateways from the PC LAN to other systems (IBM mainframe gateways and remote dial-in).

IDC considers the following products to be PC LAN operating systems: Novell NetWare family (Personal NetWare, NetWare 2.x, 3.x, and 4.x), 3Com 3+ and 3+ Open, IBM PC Network Program, IBM OS/2 LAN Server, Microsoft LAN Manager (OS/2 or Unix), Microsoft Windows NT Advanced Server (NTAS), Microsoft's Windows for Workgroups (if used for peer-to-peer networking), AppleShare, Banyan Vines, Digital's Pathworks, Artisoft LANtastic. In addition, any LAN OS designed around Novell's Portable NetWare, Microsoft's OS/2 LAN Manager, or LAN Manager/Unix are considered PC LAN OSs.

Both 1993 and 1994 figures are actual. Forecasted revenue figures are based on the average sales value of LAN OS products, where the average sales value is the average price the customer pays.

TCP/IP

Twenty years after its invention, TCP/IP is emerging as the foundation for next-generation networking. Perhaps the most general force driving TCP/IP adoption is the desire of organizations to build tighter links between users and applications through client/server computing. Other related trends that account for the tremendous growth in TCP/IP use include the following:

* Growth in the use of Unix systems in commercial environments

* Increased role of TCP/IP in application development

* Embedding of TCP/IP within leading desktop and server operating systems

* The migration of enterprise networks to TCP/IP

* Expanded use of TCP/IP networks for remote access

* Broader Internet usage in organizations/World Wide Web mania

PCs Driving Growth

The largest and most competitive market for TCP/IP products is the PC segment. From revenue of $439.6 million in 1994, the worldwide market for PC-based TCP/IP products will grow to roughly $1.7 billion in 1999. Meanwhile, the worldwide percentage of PCs with TCP/IP will grow from 5.5% in 1994 to 66.1% by 1999 as TCP/IP is embedded in major desktop operating systems (like Windows 95) and firms continue their migration to network-centric computing.

The PC-server platform will emerge as the platform best suited to meet PC clients' demand for enhanced network services. As a result, the worldwide percentage of PC servers with TCP/IP will increase from 18.2% in 1994 to 66.1% in 1999. In their efforts to meet this demand, customers will leverage their desktop operating system expertise by using the related PC server operating system (such as OS/2 or Windows NTAS) and by shopping through the same distribution channels.

Issues

The administration and management of large networks has migrated from mainframes to network operating systems within PC LANs. In conjunction with this shift, these services are transitioning from single-vendor solutions to more distinct products available from multiple suppliers. TCP/IP-based products exemplify this trend in that today many users are piecing together their own solutions using several products from multiple vendors. Other key issues include:

* Administration -- Currently, the administration of TCP/IP networks is time consuming and difficult. For TCP/IP to continue to grow as a network standard, simpler TCP/IP configuration and management tools are required. The Dynamic Host Configuration Protocol (DHCP), which enables the dynamic allocation of IP addresses, should help in this regard.

* Security -- Distributed networks, like those based on TCP/IP, require new security controls. Thus far, fire-walls have addressed this concern, but ultimately customers will need solutions that ease their ability to maintain a secure a network.

* Windows 95-Windows 95, with its integrated 32-bit TCP/IP stack, has upped the ante for vendors in the TCP/IP market. Now more than ever vendors must provide higher-level applications if they want to be included in customers' purchasing plans.

* Universal client desktops -- Customers are demanding universal clients so that via one bundled package they can connect to multiple types of hosts/servers (IBM mainframe, AS/400, VAX, etc.). As a result, vendors like Attachmate and Wall Data are finding that they now compete with TCP/IP vendors like FTP Software and Net-Manage. Pricing will fall as single-purpose solutions get squeezed into lower-margin products while at the same time integrated products compete with higher-priced single-use products.

Recommendations

If TCP/IP is to become the base network standard, vendors must provide a full range of product services (stacks, applications, management tools, security) combined with a strong service and support organization. These services, which were traditionally built into the operating environment, are just emerging in distributed TCP/IP networks.

It is not enough for a TCP/IP vendor to offer narrow solutions such as PC Internet access. In order to be a major player, it must provide a portfolio of server-based offerings such as management, administration, accounting, and security in conjunction with client applications. These pieces should be available separately, but as a package they should give users coherent control of the network. To capitalize on the growing migration to TCP/IP, vendors must articulate a clear strategy to customers and leverage TCP/IP's broad appeal in corporations. In addition, they must provide comprehensive solutions that integrate diverse systems in the enterprise. Customers will base their mission-critical network on TCP/IP and thus will select vendors that are experienced and committed to providing high-level network management, administration, and configuration tools.

Definitions

The Transmission Control Protocol/Internet Protocol (TCP/IP) is actually a set of protocols developed in the early 1970s according to Department of Defense specifications for the communication of systems from multiple vendors. TCP/IP is the standard network protocol for Unix, and a family of applications has evolved on top of it. These include Telnet (terminal access), FTP (file transfer), SMTP (mail), and SNMP (management).

PC Network Interface Cards (NICs)

The worldwide PC NIC market will continue to grow at least until 1997 although that growth will be driven largely by markets outside the United States. In fact, the NIC market will peak in the United States in 1996. Not only are prices falling, but the market is approaching saturation. From worldwide revenue of $3.7 billion this year, the PC NIC market will peak at $4.2 billion in 1997, after which revenue will steadily decline. Given this outlook, some NIC vendors are diversifying into higher-growth markets. 3COM, for example, is aggressively pursuing opportunities in LAN backbone equipment like hubs, routers, and switches.

To the extent that a market opportunity exists for non-high-speed NICs, it is outside the United States and Western Europe. The Asia/Pacific and Latin America regions are particularly attractive because they are growth markets until nearly the end of the century. More specifically, the Rest of World, or ROW, market (basically, the world, except for the United States and western Europe) for ethernet adapters will grow from 5.1 million units in 1995 to a peak of 9.2 million units in 1998. In 1999, however, shipments will begin to drop off to 8.7 million as the ROW regions start to become saturated, much as the United States will in 1996.

From a NIC technology standpoint, ethernet dominates today, and it will continue to account for the majority of both unit shipments and revenue through 1999. Meanwhile, token ring unit shipments will be flat essentially through 1999, although the value of these shipments will steadily decline as prices fall. Finally, after rivaling token ring from a unit-shipment perspective in 1997, high-speed NIC shipments (like 100Base-T, 100VG AnyLAN, and ATM) will take off, accounting for roughly half as many shipments (and roughly half as much revenue) as ethernet in 1999. Growth in high-speed NICs will be driven by heightened bandwidth requirements for business applications including LAN-based multimedia (like online training and education) and advanced Internet use.

Definitions

IDC forecasts the following network interface card products: ethernet, token ring, 100Base-T, 100VG AnyLAN, FDDI/CDDI, desktop ATM, and wireless LANs. The revenue figures are based on the average street price of a PC LAN adapter multiplied by the weighted shipments by vendor (to ensure that low-volume vendors do not skew the average street price value).

Routers and Hubs

Routers

The worldwide router market will continue to experience steady growth: from 328,370 units in 1994 to 923,025 units in 1999 (1994-1999 CAGR of 23%). Revenue, however, will grow more slowly than units (9% CAGR 1994-1999) as the low-end segment of the router market booms driven by remote-site connectivity and deeper penetration into midsized and small corporations. The high-end router market will lose unit shipments to LAN switching products in 1997 and 1998, a fate the rest of the router market could share long term.

Hubs

The LAN hub market is about to reach its prime. The good news is that in 1998 more money will be made on hubs worldwide than ever before. The bad news is that following this peak the market will steadily decline as prices fall (for mainstream ethernet and token ring products), technology matures, and the percentage of PCs connected to LANs reaches saturation. In the United States, the LAN hub market will peak even sooner -- in 1996.

From a technology standpoint, sluggishness in the ethernet and token-ring segments will cause the overall market to shrink. High-speed hub sales represent an opportunity, however, since shipments should swell as customers migrate to faster network technology. Thus, to be successful in the U.S. LAN hub market, vendors must aggressively transition to the high-speed hub offerings (for example, fast ethernet, switched ethernet, and ATM) while also positioning themselves in other growing LAN backbone opportunities (like switches). Regions outside the United States will remain growth markets for several years, even for basic ethernet and token-ring technology.

Definitions

IDC forecasts the following hub products: ethernet and token-ring LAN hubs (chassis, stackable, and unmanaged); network management modules; 100BaseT, 100VG AnyLAN, FDDI/CDDI hubs; workgroup LAN switches (switched ethernet/token ring); and workgroup ATM switches (25/155Mbps). The following products are not included: backbone LAN switches (switched ethernet/token ring/ FDDI), backbone LAN ATM switches, router cards in hubs, remote-access cards in hubs, terminal server cards in hubs, file servers in hubs, and network-management software applications.

Microsoft's Strategy: $100/PC/Year

With the launch of Windows 95, Microsoft has defined its OS strategy beyond the earlier "Windows everywhere" campaign to focus on a few specific customer/market segments. Instead of highlighting a questionable effort to put Windows on a multitude of different platforms, Microsoft is focusing on providing a few basic customer segments with compelling operating systems, which it will regularly upgrade. In addition, the vendor's revenue benchmark has come to light: $100 per PC per year.

In the past, Microsoft talked of having Windows on desktops, servers, pen computers, set-top boxes, office equipment, etc. Perhaps someday the software giant will achieve this lofty goal, but today its mainstream OS strategy is far simpler, and includes a mainstream desktop OS (Windows 3.x and 95), a workstation-class OS (Windows NT Workstation), and a server OS (Windows NT Advanced Server).

Windows 95 will extend Microsoft's dominance in mainstream desktop OSs. Even from an installed-base perspective (as opposed to unit shipments), by the end of this year 9.5% of PCs worldwide (nearly 20 million PCs) will already be running Windows 95 (compared to 46.8% for Windows/DOS, 22.4% for DOS only, 8.0% for Mac OS, and 5.4% for OS/2). In 1996, Windows 95's share of the worldwide PC installed base will jump to 27.4%, or 68.5 million units (compared to 35.2% for Windows/DOS, 15.6% for DOS only, 8.1% for Mac OS, and 6.7% for OS/2). The migration in shipments of new systems will, of course, be even more aggressive. In 1996, a whopping 62.7% of the PCs shipped worldwide will run Windows 95.

The workstation and server fronts will be much more difficult as Microsoft faces formidable and entrenched competition. The vendor will make inroads in these arenas, but the level of dominance Microsoft enjoys on the desktop -- just over 80% of PC shipments worldwide in 1995 -- will wait until the next century, if ever. Not until 1997 will more than a million PCs ship worldwide with Windows NT Workstation (although this will be more than double the 750,000 units in 1996), and even this number will represent a mere 2.4% of total PC shipments. Windows NT Server is having an outstanding year and will triple its worldwide license shipments in 1995 (vs. 1994) to 165,000, thus more than doubling its market share to 13.3% of the market. By comparison, Novell, the market leader, will ship 662,000 licenses of its NetWare products, to gain 53.4% of the worldwide PC LAN OS market. Further gains for NT Server will be steady but much more gradual. In 1999, Microsoft will ship 390,000 Windows NT Server licenses and earn 22.3% of the market, compared to Novell's 845,000 units and 48.3% share.

Over the next few years, Microsoft will need to further segment its OS product line as the market expands and user requirements diverge. By 1999, it will add a distinct consumer OS (leveraging Bob and Windows 95/97) and perhaps even a set-top box OS. At the same time, though, Microsoft expects to merge its workstation-class offering (Windows NT Workstation) with the corporate desktop product (Windows 95 and 97). Segmentation of this sort is wise since home and business PC users' requirements for a desktop operating system are already very different.

In conjunction with these product goals, however, Microsoft also hopes to both increase and flatten its revenue stream. Basically, the vendor aims to get $100 per PC per year from revenues in operating systems, applications, CD-ROMs, etc. For consumers, the goal could go as high as $150 per PC. Since regular and incremental upgrades will play a critical role in achieving this revenue target, Microsoft hopes to get users to upgrade their OS every two years. Given that roughly 76 million PCs will ship worldwide in 1997, should Microsoft achieve just its $100 per system objective by then, the vendor would enjoy calendar 1997 revenues of $7.6 billion on the basis of new PC shipments alone (not counting installed base upgrades). Not bad for a firm with fiscal 1994 revenues of nearly $4.7 billion. (For Microsoft to keep revenues steadily growing at its fiscal 1993/94 growth rate of 24%, the vendor would have to earn nearly $8.9 billion in fiscal 1997.)

Not only do frequent upgrades help Microsoft achieve its $100 goal, however, they also help flatten revenues. For the past few years, Microsoft has in a sense been overly reliant on Windows 3.x for its revenues in that the OS has been the company's cash cow, and it is only now being truly upgraded. If Microsoft can successfully spread its revenues more evenly over a host of products, each with its own upgrade cycles, Microsoft would achieve a more regular and even revenue stream.

[GRAPHS OMITTED]

Worldwide PC LAN OS Revenue, 1993-1998

                  1993     1994     1995    1996     1997     1998

LAN OSs ($M)     2,461     2,354    2,697    2,986    3,137   3,257
% growth            16        -4       15       11        5       4

                     1993-98    1995-98
                    CAGR (%)    CAGR (%)

LAN OSs ($M)          5.8        6.5
% growth

The Gray Sheet -- Special Edition also includes IDC's latest forecasts for most IT market segments. In each category IDC applies the fruits of tens of thousands of end-user interviews and forecasts market growth through the end of the decade. In addition, The Gray Sheet-Special Edition examines the trends behind the numbers. It provides a concise explanation of the driving forces and inhibiting barriers in every segment of the U.S. IT market. We are extending our special offer to you as Gray Sheet subscriber to get the Special Edition for $100 off, or just $395, until the end of September. For more information, or to order your copy of The Gray Sheet -- Special Edition, contact Lisa Bloom at (508) 935-4236.

RELATED ARTICLE: PC Market Outlook Conference

Tuesday, September 19 -- The Red Lion Hotel, San Jose, California

The 1995 PC Market Outlook is a one-day conference sponsored by IDC's Personal Systems Group. The conference provides comprehensive market and technology forecasts, fresh primary research on emerging markets, access to IDC's expert analysts from around the globe, and a forum for discussion about the global computer business.

The PC Market Outlook is an excellent opportunity for anyone to inexpensively acquire a wealth of IDC's latest PC-related data and analysis. In addition, it is a chance to meet with me (Ted Julian, The Gray Sheet editor) and other IDC analysts. If you plan to attend, let me know so that we can get together (phone: 508-935-4237; Internet: tjulian@idcresearch.com).

The conference is scheduled to cover the following: the worldwide PC market outlook, consumer brand preferences and buying behavior, Windows 95, PC distribution channels, the future of mobile computing, emerging PC technology, and the PC as the gateway to the information highway.

To Register

* Call 1-800-343-4952 (or 508-935-4585) and ask for Dara Queen.

* Or fax your registration to 508-935-4015. Include your name, title, company, address, city, state, zip, phone, and fax.

Registration fees (bulk discounts are available for three or more)

* Full registration -- $595

* The Gray Sheet subscriber special rate -- $495

* IDC client registration -- $395 (if you subscribe to an IDC CIS in addition to The Gray Sheet)

Payment and Cancellation Policy

Payment must be received prior to attendance and may be made by check, cash, or credit card (Mastercard/Visa). Cancellation will be accepted until one week prior to the event.

Location

The PC Market Outlook Conference will be held at the Red Lion Hotel in San Jose, California (2050 Gateway Place, San Jose, CA 95110, phone: (408-453-4000). Although we reserved sleeping rooms at a reduced rate, they are nearly all taken due to a better-than-expected response for this conference. If you need a room, call the Red Lion immediately and mention the IDC PC Market Outlook Conference for the lower rate (however, we can't

COPYRIGHT 1995 International Data Corporation
COPYRIGHT 2008 Gale, Cengage Learning