From: Bill Todd [billtodd@foo.mv.com] Sent: Thursday, September 06, 2001 11:49 AM To: Info-VAX@Mvb.Saic.Com Subject: a summing-up so far Ken Farmer invited me to write up a point-of-view article for tru64.org for those who don't follow the newsgroups. I sent it to him last Sunday, and you all know what happened on Monday. Submitted an update early this morning, but it hasn't yet appeared on the Web site. Since CETS is fast approaching and people are doubtless wondering what they'll find to talk about there, I've appended it below: not much new and somewhat less detail than has surfaced in the total discussion, but sometimes detail can be overpowering. - bill Prologue: I sent this article to Ken the day before the HP acquisition offer was announced. Since I believe that the issues it raises remain important, I've chosen not to modify a word of it in hindsight but rather to add a postscript that attempts to tie those concerns into the new, wider picture. - bill Alpha: What Was Compaq Thinking? Over two months have now elapsed since Compaq signed Alpha's death warrant, but the reasoning behind the move remains as clouded as it was on June 25th. Some people believe that it was a legitimate business decision, others believe that it was either monumentally incompetent or the result of behind-the-scenes forces unrelated to shareholder value, and the majority are likely just uncertain and therefore at least somewhat worried. Ken Farmer invited me to present one side of this issue, since he felt that people who do not follow the newsgroup discussions (many of which have occurred in comp.os.vms and comp.arch rather than in comp.unix.tru64) might have missed the existence of significant disagreement with the official Compaq view of things that has been aired at tru64.org. He asked only that I attempt to keep the presentation moderately balanced, and I have to say that doing so will be a pleasant change from the often-heated newsgroup interchanges. So I'll begin by summarizing Compaq's explanations for its Alpha decision, and then try to explain the problems some of us have with those explanations. "Good-Bye, Old Paint" Compaq presented Alpha as a platform that had simply out-lived its value: it cost money to develop, commanded only modest market share compared with other server platforms, and showed little indication of improving that position in the future. Given that Compaq claims that its most promising future opportunities lie in service areas, it presented a platform-consolidation strategy in which the next generation of Alpha systems would be completed and brought to market, after which completion of Tru64 and VMS operating system ports to IA64 would allow that platform to take over the markets Alpha had satisfied. It's certainly possible to disagree with Compaq on Alpha's future potential and profitability, since many believe that had Compaq made any serious effort to support and promote Alpha, even starting as late as June 24th, it could have found Alpha more than worth keeping (since even being relatively neglected it was still as profitable as any other Compaq business area, and more profitable than most). And it's also possible to disagree on the future potential of Compaq's service business (especially the 40% annual growth figure based upon the acquisition of new service companies supposedly made possible by the money acquired from Intel as part of the Alpha deal), given that Compaq has had three years to make the service business it acquired from Digital bloom and has yet to do so. But corporations do need to be free to follow their visions, even possibly faulty ones, and if those were the only issues there would likely be relatively little active discontent over Alpha's demise, though still possibly a fair amount of regret for opportunities lost. Shattered Commitments The pill that may be the hardest for many to swallow is the fact that the June 25th announcement shattered the repeated, public, unequivocal commitments, both verbal and written, that Compaq has used since acquiring Digital to encourage customers to commit their businesses to Alpha's long-term future, including explicit commitments to that future's evolving performance and capacity. These commitments were unambiguously spelled out in a letter from two Compaq Vice Presidents (Bill Heil and Jesse Lipcon) that was posted publicly on the Alpha Web site from September, 1999, through July, 2001 (i.e., it finally disappeared a month or so after Alpha was axed). Though Bill and Jesse left Compaq late last year, the letter remained - and was used to demonstrate Compaq's public commitment to Alpha, right up until the June 25th announcement, in presentations to customers by a wide variety of Compaq functionaries including current Compaq VP&GM Rich Marcello. It was all the more forceful and convincing because it contained none of the usual disclaimers about 'forward-looking statements that may or may not actually come to pass': it just said, effectively, "You can safely bet your business on this future, because we are betting ours on it." Why would Compaq have gone out on such a limb? The letter appeared just after Compaq dropped support for Windows 2000 on Alpha (another instance in which Compaq reassured customers that its support for a platform was absolutely solid right up until that support was dropped) and Microsoft retaliated by dropping support for 64-bit Windows on Alpha. Customers were concerned that this might reflect questionable Compaq commitment to Alpha itself, and the letter was meant to be the strongest possible assurance that this was not the case. And it was effective reassurance indeed. Until June 25th, when, without consulting customers to find out how to change these commitments in some mutually-acceptable manner, Compaq simply broke them, unilaterally and with no hint of apology. That alone is sufficient reason for some customers never to do business with current Compaq management again: some minimal level of trust is important to business relationships, and Compaq's leadership just doesn't have it. I'll provide my copy of the letter for Ken to post if he feels he can: it was, after all, publicly posted for close to two years by Compaq. A slightly sanitized version (possibly just an earlier draft) of this letter still appears on the Compaq Web site, though less prominently http://www.compaq.com/hpc/news/news_alpha_sept99.html ): it's just enough milder in the last lengthy paragraph to significantly blunt its impact compared with the version that's no longer there. Smoke, FUD, or Worse? Unfortunately, there's more. In apparent attempts to justify a move which it must have known would engender major criticism, Compaq claimed that Alpha in the future would have had "no substantial performance benefit" over contemporaneous IA64 platforms (Rich Marcello, quoted in the June 26th EETimes: http://www.eetimes.com/story/OEG20010625S0105 ). And Terry Shannon plus members of both the VMS and Tru64 development groups publicized Compaq's internal assertions that the doubts about Alpha's future performance potential (compared with IA64's) came from the Alpha engineers themselves (I think there may also have been some more formal public statement to this effect, but I can't find it at the moment). It wasn't long before said Alpha architects started to come forth (one publicly, more privately) to state that not only had they said nothing of the kind but that in their opinion Alpha had every bit as much performance promise (and future advantage over IA64) as had been forecast for years and explained in great detail in a 1999 Compaq white paper (alpha_ia64.pdf, which can still, at least as of this writing, be found as one of the papers at http://www.compaq.com/hpc/ref/ref_systems.html ; additional comparative analysis can be found in Paul DeMone's 'Insider' columns at www.realworldtech.com). Later, a high-level Alpha development manager responded internally to questions by stating that the move had been a 'business decision' rather than due to technical issues. Recently, it has been suggested (in http://www.theinquirer.net/27080104.htm , an article which is disturbing in other respects as well) that the decision to kill Alpha came from the CTO's office rather than from more substantive portions of engineering. So there are several possibilities, none of them very attractive: 1) Marcello and his superiors were actively lying, 2) Marcello and his superiors were actively lied to, or 3) a pivotal corporate decision was made on the basis of incompetently-researched information that no one bothered to verify, despite the decision's clearly sensitive nature. But, as in other areas, Compaq has issued no official clarification since the announcement rhetoric. In a bizarre twist, concurrently with the above public claims of Alpha's newly-discovered inadequacy Compaq was more quietly (and inconsistently) spreading rumors via employees (VMS VP Mark Gorham apparently among them) that Intel had purchased the rights to Alpha technology to save the faltering IA64 architecture, which would become so Alphabetized under the covers that the result would be entirely acceptable to those Alpha aficionados who might otherwise be upset (see http://www.theinquirer.net/28060117.htm , plus comp.os.vms posts from 'Alphaman' of that date). This tactic was much more short-lived, however - possibly because Intel got wind of it and was not amused. Voodoo Economics Assertions were put forth by possible Crown Prince Mike Winkler (in http://www.zdnet.com/eweek/stories/general/0,11011,2780645,00.html and picked up by other eager external voices, Mr. Shannon's first among them) that the expense of continued Alpha development was simply not justifiable, but while no internal voices have been raised up to contradict this it appears to stand up to examination no better than the above-described technical fig-leaves. Leaving aside the likelihood that any real effort to support and promote Alpha could have significantly expanded its market share and profitability, existing VMS-related profit alone ($800 million annually as of a bit over a year ago) far more than covers the $300 million annually Winkler claimed Alpha development cost, let alone the annual $150 million quoted in the aforementioned Marcello interview http://www.eetimes.com/story/OEG20010625S0105 ) which may be a more realistic estimate of what Compaq stands to save by canceling Alpha chip development (and is also more consistent with the projected development costs in the 1999 and 2000 annual reports). Add in at least $400 million in annual profit for Tru64-related system sales (a March Compaq slide presentation listed VMS-related annual revenue as $4 billion and Tru64's as $3 billion, so pegging Tru64's profit at half VMS's seems conservative), plus a smidgeon for Alpha Linux system sales, and the Alpha business looks healthier than any other Compaq has: Tandem systems likely have similar margins, but at $2 billion annually according to the March slide presentation less than 1/3 the Alpha revenue; PCs have barely broken even for the past couple of years, and while 'industry-standard servers' grew in revenue and profitability last year when the dot.com bubble burst their profits apparently disappeared with it. I could go on, but this is already getting rather lengthy. Anyone hungry for more details can find them in my July 19th comp.arch post rebutting Terry's repeated assertions that Alpha development was "economically unsustainable". So, What Was Compaq Thinking? The picture I've painted above is not a pretty one, strongly suggesting either staggering incompetence or blatant dishonesty (or both). But while I've said as much in my more heated moments, there's still something unsatisfying about not having any explanation for what might have motivated Compaq to be so unwise and/or so unethical. For years I've felt that Compaq's leadership really wasn't comfortable being responsible for significant technology, hardware or software. At least they've never seemed to devote any real effort to promoting it since Eckhard Pfeiffer was ousted. Which at least raises the question in my mind of whether the current leadership might be afraid of joining Pfeiffer if they were so rash as to pursue anything outside Compaq's traditional PC and 'industry-standard server' space. An alternative to fear (of the BoD?) could be internal politics. The 'Compaq Classic' Houston contingent has, unsurprisingly, seemed to have considerably more influence than those people acquired with Digital, and using it to expand their own fiefdoms rather than in what might be the best interests of the company as a whole it would hardly be unprecedented. Then there's the bean-counter disease, of which the new interest in the service business could be a symptom. Service industries don't require expensive development funding and technology responsibility, and since Compaq's service revenue was broken out from the associated system revenue it has looked pretty attractive. The fact that most of this service revenue might dry up if Compaq hadn't sold the hardware in the first place may be less evident to a bean-counter. And finally there's simple isolation. Perhaps the people who make decisions just haven't a clue about the possible consequences: after all, the remaining Digital customers have put up with many years of neglect or outright mistreatment without bolting, so what reason is there to believe they'll balk this time? But while I'm fairly comfortable that I understand a lot of *what* has happened surrounding the Alpha debacle, I'm nowhere nearly as sure that any of the above adequately explain *why* it occurred. And while if Compaq chooses to respond to this article it may be inclined to attempt some kind of factual rebuttal, I suspect it would be better advised to provide instead details of the reasoning it applied to the process and any additional insights it has developed in the months since: whether Compaq is yet aware of it or not, its survival may depend upon adequate answers to such questions. Postscript: Well, the events of September 4th certainly add possibilities to the speculations above. But they don't eliminate the concerns about Compaq's integrity and competence, which is why I decided not to modify the exposition but simply to add to it. We now know that plans were afoot to merge with another company with an even greater commitment to IA64 than Compaq has http://www.infoworld.com/articles/hn/xml/01/08/23/010823hnitanium.xml ) and no option equivalent to Alpha as a fallback (PA-RISC could carry HP servers for a few more years, but was never planned to carry them indefinitely). Given the investment HP already has in IA64 (including running hardware and software), expecting it to embrace Alpha would be a stretch. However, that merely changes the question to why Compaq would *want* such a merger. If it thought that this would be a good deal for its stockholders, all current evidence indicates that it was radically in error: HP's stock dropped 22% over the two days after the announcement, and Compaq's stock, unlike that of most acquisition targets, dropped 16% as well; before the announcement, both were at or near 52-week lows, while two days later both were at 5-year lows; analyst reactions have been almost universally negative, seeing no synergy in the deal but two years of confusion while the merger (if it occurs) is sorted out and advising purchasers to consider holding off on major commitments to products of either company until it has become clear which product lines will survive consolidation; HP's credit rating was reduced ... For the Tru64 and VMS camps, the news does not appear to be good either. "Eventually, HP will converge toward Linux, NT, and Unix, according to McDonnell" http://news.cnet.com/news/0-1003-200-7071077.html?tag=pt.msnbc.feed..ne_7071 077 ) hardly promises the VMS 'renaissance' that the latter group had hoped for, while Capellas' statement that the idea for the merger arose during discussions about Compaq licensing HP/UX makes his (let alone HP's) likely commitment to Tru64's future appear dubious at best (and since HP/UX is big-endian while Tru64 is little-endian, the likelihood of merging them compatibly seems slim). Couple this with the porting hiatus during which neither Tru64 nor VMS will be available on a platform with a perceived future, plus uncertainties due to the merger itself (as mentioned above), and additional customer attrition seems inevitable. The situation thus appears less confused but considerably bleaker than it did prior to September 4th. My guess is that for Tru64 and VMS to survive, the merger must fail, and fail relatively quickly. The result of such a failure could be a top-level Compaq management blood-bath, which doesn't strike me as a bad idea in any event, though an attitude-conversion of Biblical proportions on their part could be equally acceptable. I'll even go so far as to suggest that the only way Compaq is likely to survive is to revive Alpha: the EV8 design was close to completion, the EV7 engineers are still around (plus any who might return if there were some real guarantee that EV8 would become reality), Compaq retained Alpha development rights in the deal it made with Intel (that's what supposedly made the deal FTC-friendly), and all the financial numbers still seem to suggest that any money spent on Alpha development will be amply repaid in increased profit. No, I don't really expect that to happen. But nothing else on the horizon looks as if it will save the company in anything like its current form.