While IBM executives have for some time talked in general terms about Network Stations, the actual deal between IBM Corp and Network Computing Devices Inc that led to the imminent IBM announcement is quite recent. On June 27, IBM and Network Computing signed the product development contract they call an alliance agreement. Under the terms of the deal, Network Computing would initially develop a version of its Explora tailored to IBM’s requirements and later do a more advanced desktop box based on the Explora design and extensions to OS/400 that support Network Stations. The Phase II product and associated software are supposed to be completed so that they can be made generally available by IBM in July 1997. The agreement runs to December 31 1998, but IBM has the right to extend its term until December 31 2000. Notwithstanding the intention of both parties to honor the agreement, IBM insisted on the right to terminate the contract without cause on 60 days’ notice, although in that case some of IBM’s rights and obligations to Network Computing (mainly those related to the licensing of intellectual property) would persist. Further, the deal could be ended if Network Computing files for bankruptcy or goes into liquidation, possibilities that are unlikely but not entirely out of the question.

Embarrassment

IBM agreed to contribute to the product development process and to set up the Sun Microsystems Inc hardware on which Network Computing runs its software development. IBM will give Network Computing an AS/400 to be used in the development and testing of the support code being written for it. Among the goals of the development effort are the extension of Network Computing’s network support to Token Ring; it is limited to Ethernet at this time and the adaptation of the extended Explora product for use with all IBM hosts (S/390, AS/400, RS/6000 and iAPX-86-based servers) will require Token Ring. Network Computing is also supposed to make its best effort to obtain a license for additional software under terms permitting the software license t o be passed to IBM. All specific references to this software have been kept confidential to protect IBM from embarrassment should Network Computing be unable to obtain the desired license, which it may well have done before details of its deal with IBM became public. If so, it is possible that the original owner of the software is Microsoft Corp and that the code would enable the Network Computing extensions to Windows NT to be more readily ported to AIX, OS/2 or another environment. Another possible source of code is Citrix Systems Inc, which is the primary source of software that turns Windows NT into a multi-user graphical operating system. Whatever company Network Computing promised to approach for a software license, it is clear fro m the way Network Computing and IBM have withheld details of their agreement that IBM would rather not have the information made public. Network Computing will make Network Stations, or, more accurately, manage their manufacture… but not necessarily all of them. IBM has the right to assign a portion of the production to another supplier or make some units itself, although it will still have to pay Network Computing royalties on every box it buys until December 31 2000. At that point, IBM wil l be deemed to have fully paid Network Computing for a perpetual license to all the intellectual property inherent in the Network Stations, except for what the manufacturing contract between the two calls Major Enhancements and Custom Enhancements.

Rights and royalties associated with such anticipated but not specifically foreseen changes in the product will be governed by specific contracts created to meet situations as they arise. The fraction of the production run that IBM must buy from Network Computing remains confidential, but until IBM can generate a large volume of orders, Network Computing will almost certainly be IBM’s sole supplier. It currently has assembly performed by a subcontractor in Thailand. However, that could change and, if IBM orders large volumes of the machines, both Network Computing and IBM would presumably favor the use of additional assemblers. Also, Network Computing has never been geared up for the unit volumes the Network Station will generate if it b ecomes a successful product line. The relationship between IBM and Network Computing is based on the assumption that IBM will make its Network Stations world wide products. IBM expects some customers to acquire monitors for use with the devices, while others will use monitors they already own. The potential market for Network Stations could be enormous if they are priced attractively and are easy to integrate with legacy systems and emerging applications. The worldwide installed base of 20 to 30 million dumb tubes and low-end personal computers attached to IBM hosts suggests that a successful Network Station would warrant production volumes of at least 1m units a year and possibly as many as 3m. IBM is believed to ship 2m to 3m 3270s and 5250s a year, just about all of them to users of S/390 and AS/400 systems, plus 5m or 6m personal computers. Outside estimates of the number of IBM personal computers sold for use as workstations on IBM hosts are all over the map; our own estimate is that a quarter to a third of IBM personal computers are sold for use as clients on S/390, AS/400 or other IBM hosts. A successful networkstation could detract from some of IBM’s personal computer business, but it would far more often be sold as a n upgrade to a 3270 rather than a downgrade of an intelligent client. And even if there is some impact on personal computer sales, IBM is only one of many suppliers of personal computers to its processor customers and, at many sites, not even the most significant. Until other vendors can offer appealing alternatives – and it is possible that none will do so – IBM will have the Network Station market all to itself. Even if IBM has a head start, two serious issues cloud the future of the Network Station. First, there is no assurance that IBM’s NetworkStation product or even the concept will catch on very well. Second, if sales of the device do not rise fast enough, Network Computing may not be able to survive in its present form. IBM will be offering its Network Station in competition with a number of other similar products as well as a number of cost-effective smart workstations, including its own personal computers. At some sites, the most appealing alternative will be used personal computers. Secondhand 80486 boxes from major as well as minor suppliers are cheap and abundant, and all the big leasing companies are active in this market.

Green screen users

Increasingly, lessors offer asset management plans that provide for the upgrades and reconfigurations commercial users have learned to accept as a byproduct of progress in hardware and software technology. Because their functionality derives mainly from code supplied by servers, Network Stations will be less volatile than personal computers, but only if they are designed with sufficient processor power, graphics capability and memory capacity to withstand the pressures that send healthy but obsolete personal computers to the graveyard. The appeal of the Network Station concept stems in large measure from the presumably high support costs associated with personal computers. Because it is very difficult to control the way personal computer s are acquired, managed and used, large organizations experience very high per-user costs, particularly according to studies done by consultants. Organizations whose experience resembles those described in the horrific consultants’ reports will certainly consider Network Stations as alternatives if the Network Stations permit end users to perform all the same functions for less money. But will it be the undisciplined end users that now have personal computers who get the Network Stations or the green screen users who have no inclination to do the kinds of things that increase support costs?

By Hesh Wiener

From Infoperspectives, (C)1996 published by Technology News of America Inc.