The DOS gold mine is shrinking

From Techrights

Jump to: navigation, search




Microsoft Memo

To: Steve Ballmer: Joachim Kempin; Paul Maritz; Russ Werner; Nathan Myhrvold
From: Bill Gates
Cc: Jon Shirly, Jeremy Butler, Richard Fade, Jon L20:35, 9 February 2009 (UTC), John Sabol. Peter Neapert, Mike Maples, Scott Oki
Date: May 18, 1989
Subject: Operating System Strategy

My recent trip to Europe gave me time to gather data and think about our oerating system strategy.

The DOS gold mine is shrinking and our costs are soaring - primarily due to low prices, IBM share and DR-DOS. Making Windows a strong product benefits our gold mine and protects it in the following ways:

DR-DOS. I doubt they will be able to clone Windows. It is very difficult to do technically. we have made it a moving target and we have some visual copyright and patent protection. I believe people underestimate the impact DR-DOS has had on us in terms of pricing.

IBMs market share increase. The new IBM low-end machines are designed and priced assuming they can achieve 1M in sales. In addition to their other sales. This will destabelize many of our customers and cost us a lot of revenue. I think IBM's MCA strategy and the move to OS/2 and networking will increase their share. IBM's failure to exploit the 387 or portable markets is not likely to be repeated. I am impressed with their seriousness. We can make money from them with Windows - retail yields the highest dollars, but OEM makes it a standard. OEM is the best long term.

Lower prices. Microsoft can't get more than the folowing royalties for DOS alone.

EndUser FOB $ %

500	 250	 5	2	8086

1000 500 10 2 286 2000 1000 20 2 386 HD 3000 1500 30 2 386SX HD

With Windows offered by Compaq and IBM I think we can get almost two times what we could get ourselves. I don't think any amount of system software can do better - even the world's best OS/2.


Going full bore on Windows leaves the folowing major problems:

1. OS/2 consistant 2. OS/2 more powerful 3. OS/2 slower to catch on - PR - UNIX 4. IBM 5. Use of OEM channel

Minor problems include:

6. Language support of Windows

1. Cannavino said he thinks they ought to embrace Windows if mirrors creates decent applications. If not, he says we have a dilemma. His point is that a lot of apps will use mirrors and that will determine how people perceive OS/2. If mirrors is okay, then he is sure we can make OS/2 look good.


2. High end. Better paging, objects, file system, 32 bit linear, SAA, true multitasking, new technology. Over time we will be fine. I am really keen on having OS/2 be really strong in the following areas:

1990 volume

250k - IBM lovers, EE, Office

100k - developers

80k - servers

? - POSIX - I want to run Xenix binaries so we can attack multiuser. Olivetti says this would be the way to sell OS/2 in '90 in Europe. (Paulma - who can write a memo on the feasibility of this?)

50k - Japan

100k - DTP

50k - Retail (NCR)?)

100k - Engineering

? - All SUN applications. I need to review their catalog. Lets get them all. Who is thinking about this? With 2.0 what are the obstacles.?

150k - Lotus lovers

Adding it up like this gets to 880k. Is it realistic?

3. UNIX is strong for a varity of reasons, multiuser is more important - far more important - than we give it credit for. Government standards are a problem. I think we need to invest more people in this. I have been very sucessful explaining how a desktop binary standard is better than a source standard when I get to talk to the right people. The EEC meeting has me feeling better about this.

Unique applications - we can fix this one with a focused effort. Love of RISC - the emerging liability. Other than our 850 project, 486 help, 586 help we can't do anything more.

Decent PR should turn our Windows victory into a Microsoft OS/2 victory. I think renaming DOS/WIndows "OS/1" may have real merit.


How will we deal with Windows for installed base - we could do a retail upgrade with Windows. How will we get the utilities out? When does IBM find out what we are doing with COmpaq. These are vry serious issues. Who will answer them?


Another issue to resolve is whether we allow non-Windows apps like compilers to use high memory with Windows or give them some way to look like pseudo-Windows applications.

The main difference between this memo and Nathan's mail of MAy 8th is that. I don't want to hold back at all on Windows/386 and he want it to be withheld. We have pots of existing Windows/386 OEM deals - we can't cancel them and tell them Windows/286 v3.0 is the substitute. We have a Windows/386 retail product today. There will be PR that we are suppressing features to help the failing OS/2. Desqview will have paging - DOS extenders exist. I am optimistic that IBM, extra ISVs and features will bail us out and delaying until 1991 and < 1M is okay.

I want us to release Windows/386 without deletions as son as we can. It will confuse people somewhat. Some users will expect an "upgrade deal." It will delay OS/2 somewhat but it will help us to get IBM to allow us to make OS/2 reasonable.

I would like to see memos from people on what they paln to do in their area based on us moving ahead with fall Windows/386 v3.0.

Full Exhibit

Personal tools
Search entire domain