MSX computer sales

Very good point to investigate the sold licenses to MICROSOFT. Although they would not remember, this could be the only way to get a trustfull number of sold machines.

I've checked a few old magazines and was able to confirm the data: The 4 Million machines sold shipped GLOBALLY milestone was reached in the early months of 1990.
It seems it was a big deal at the time, MSX magazines of the era ( MSX Magazine and MSX Fan) did report on it.

MSX Magazine had a two pages report on issue 03/1990 - pages 88,89.

はじめにも害いたように、この 1月をもってMSXシンの国内外 のメーカー出荷台数が400万台を突 破した。内訳は国内200万台、国外 200万台。お隣の韓国をはじめ、ヨ 一ロッバ、中近東、南アメリカな ど、世界中にMSXマシンが広まつ ているんだね。そのあたりの状況 も合わせて、来月号では400万台突 破記念の特集を組むぞ！

And then a LARGE (10+ pages) article in the 04/1990 issue and then again a 2nd part in the 05/1990 issue.
MSX Fan had a similar short blurb on page 28 of the 03/1990 issue, and both magazines kept the seal of the milestone on the front page for a few more months.
These magazines can be freely checked on archive.org

After the 4 million mark:

After the 80's most manufacturers had left, and only the MSX2+ machines (which had been on sale for a while) were still on the market; Then there were the A1-ST, the A1-GT and finally the OCM.

It is a known fact that MSX2+ was a market failure compared to earlier machines, but I don't know how many machines were sold in the 90's, what is known is that in those years the NEC machines had almost all the PC market share, the X68000 machines were filling the niche/power user market and the MD, PC engine and the old horse FC had ~100% of the games market.

The Turbo-R A1-ST sold well, but the GT was a failure, and the OCM was 5000 pieces.
Then we *possibly* have a few un-licenced machines, but nobody knows the number of those.

So all-in-all, the 5 Million (official) machines sold globally seem to be a plausible number.

I also found a small article with an interview with Nishi (MSX Magazine 92 summer/ Page9) who seems to confirm it, although it says 5 Million users, not 5 Million machines, and I don't know if Nishi can be trusted.

I also found an article on ASCII website where it mentions that the number of MSX Engines far exceeded the number of machines sold.
All those ICs ended up in many appliances (i.e. the video equipment mentioned above) because it made them easier/cheaper to manufacture, but those weren't necessarily computers.

Why is it a “known fact” that MSX2+ was a market failure? Personally I can see that initial sales for Panasonic, Sanyo and Sony were probably not as great as when the FS-A1 or HB-F1 were released but all in all a lot of MSX2+ machines were sold. If you look at sales on Yahoo Auctions the ratio is probably 1:1 for MSX2:MSX2+, that says something. And also Panasonic and Sony would not have made the HB-F1XV and FS-A1WSX if the HB-F1XDJ and FS-A1WX sold badly.

And to me Yahoo Auctions is like a window back in time. There are still a lot of 2+ machines offered there (I bought 12 only in the last week, and I did not buy all of them). Then there are machines like the Canon V-25 or the Victor HC-80 which only show up infrequently. But there must have been production runs of these and I don’t think they would just make 100 in a run (it would be cool to see photos of MSX production facilities come to think of it).

There is the corrected text:

According to Pentarou's explanations i would rather think a number close to 6M because MSX2 and even MSX1 kept selling it when the 2+ came out.

You left out the "compared to earlier machines" part.

But it should be pretty obvious, just by looking at how many manufacturers were still active in later years: Only only 3 manufactures and no official European/South American machines.

Shipped machines ≠ Sold machines. After '90 there may have been MSX1 leftovers that kept selling, but manufacturing of (official) MSX1 and MSX2 machines was a done deal.

And the 5M number is from Nishi.

It's possible, but those have already been counted in the 5 Million machines quoted by Nishi.

All official machines are already included in the 4 million number (up to '90).

And I already invited you to bestow some proof more substantial than "I believe/I think".