According to DataQuest's analyses, sales of the UNIX operating system will peak in 1996 and decline thereafter to a level of 1.8M/year by 2000. A YES claim states that DataQuest's market estimates for the year 2000 will report that UNIX unit-sales have peaked and have declined to a level of 1.8M/year. Whether they peak in 1996 is irrelevant to the claim.
Background: From: Wendell Craig Baker (wbaker@splat.baker.com) Date: March 14, 1995
Yesterday's Wall Street Journal (3-13-95) had a very interesting article entitled "Sellers of Unix Systems Adopt Standard against Microsoft" (page B6) on the planned announcement today of the so-called Common Desktop Environment by HP, IBM, Novell and Sun.
The article was extremely interesting by virtue of how this information was presented by the reporter. ``I think its too late,'' Robert Enderie of Dataquest is quoted as saying. There was a telling little table in the corner:
Unix Falls Behind Shipments, in thousands 1994 1998 actual projected Unix 1,855 2,195 Windows NT 555 12,552 Windows 3.1, 95 23,833 75,901 Note: Table assumes that NT eventually becomes an operating system for server computers linking desktop machines while future versions of Windows 95 become desktop systems linked to NT Source: DataQuest
Unix Falls Behind Shipments, in thousands 1994 1998 actual projected Unix 1,855 2,195 Windows NT 555 12,552 Windows 3.1, 95 23,833 75,901
Source: DataQuest
On March 8, Martin Reynolds, who is Director of Technology Assessment at DataQuest gave a talk at Stanford entitled ``Processors in the Real World.'' Most of the talk was on how DataQuest goes about generating estimates of what Intel is planning to do. The direction of Intel being what controls the destiny of the processor industry at this point.
Significant here though was his analysis of Intel's current in-place fab capability. He had two slides full of Intel plants that are either blowing glass today or will be online in the next 18 months. AMD by contrast has just the one plant. Based on the analysis Reynolds stated that Intel's plan is to have ship 70 million P6 chips by Q1 1997. This will roughly double the installed base of the x86 architecture -- and almost half of them will be the 6th generation.
This makes claims that the Microsoft marketplace in '98 will be 37 times bigger than the whole Unix marketplace a bit more credible.
But the WSJ article was on standards and how the Unix has used them and how CDE is going to change that. I was amused that finally the reporters had identified the cartel behavior in the Unix marketplace (lack of standards, attempts on the part of vendors to differentiate themselves from their competitors by making their software unique, a strategy of selling computer systems by locking customers into their proprietary brand of software, etc.).
Enderle's predictions for DataQuest go on that HP, SGI and DEC are expected to ``rapidly adopt'' Windows NT and that Unix sales will peak in 1996 and decline thereafter to a level of 1.8M/year by 2000. Microsoft at that point is predicted to be selling 22M units of NT and 99M units of the Windows client OSes.
But what about the extra performance that you get with the RISC processors on which Unix runs? Isn't that worth paying something for? Yes, Reynolds said it is -- the market will absorb a 2X to 3X price jump for that extra 50 to 100% in performance that you get in RISC-based systems. But while the RISC architectures currently hold the high ground in terms of performance some of them have only fair-to-poor ratings in terms of price/performance.
Standard ambiguity clause: If the wording is found to be ambiguous the Panel will judge on the basis of the obvious intent. If the intent is ambiguous the Panel will judge on the basis of precise wording. If both are ambiguous or if both are clear but conflict, the Panel will look for a solution that causes the least damage to FX as a market and game.
CAUTION: This claim deals explicitly with *sales* of operating systems. Judgement will be based on Dataquest estimates for sales in the year 2000, which may or may not include actual sales of noncommercial or semi-commercial Unix or Unix-like systems such as FreeBSD or Linux. Copies of these operating systems downloaded for free will not be taken into account. The number of actual installations of UNIX (free or paid for) is irrelevant to this claim. If Dataquest estimates are not available, a suitable alternative may be found.
The essence of the UNIX claim is that the Dataquest estimates for the year 2000 will show UNIX sales at or below 1.8 million copies sold in that year. The "or below" is our interpretation of the claim's intent: requiring the sales to be precisely 1.8M makes little sense. The claim clearly indicates that 2000 is the year of interest. The precise year of UNIX peak (if any) is not relevant to judging the claim. A local minimum below 1.8 million for any year before 2000 is not relevant to judging the claim.
If Dataquest includes Linux/FreeBSD units sold on CD in its UNIX sales, we will count them also.
We expect to judge this claim near the stated due date.