Those who regularly read The Memory Guy’s blog posts know that I am concerned that today’s boom will result in a collapse sometime in the near future. With that in mind, my reporting of August’s record-breaking revenue number won’t be treated as a welcome success, but as a good thing to be viewed with some caution.
First the news: The Semiconductor Industry Association (SIA) announced that August semiconductor revenues, as reported in the WSTS 3MMA number (3-month moving average) were $53.12 billion, the highest on record.
Is this number reliable? The SIA, the US lobby for the semiconductor industry, is also the sales arm for the World Semiconductor Trade Statistics body (WSTS) which is basically a group of accounting firms who anonymously collect and compile monthly statistics from any meaningful semiconductor supplier who can be convinced to share detailed sales figures: Revenues, unit shipments, etc. by product type and region. That’s a lot of data.
WSTS then produces a report based on these companies, and its own estimates of the few companies that don’t participate, and publishes that information to be sold to industry participants. These participants then can use WSTS numbers, which are reputed to be as unbiased as possible, to measure their own success.
Now that you understand how very important and reliable this data might be, let’s talk about August, the most recently-reported month in the WSTS database, and a record-breaking month for semiconductor revenues. Surprisingly, the SIA missed this fact in its press release.
Actual monthly revenues were lower, at $52.93 billion, second to March 2022’s $54.63 billion, and this past July’s actual number of $52.97 billion.
Note that the reason that the 3MMA was the highest on record was that this was the first time in history that revenues for three successive months were above $50 billion. Note, too, that August’s revenues were lower than July’s, possibly indicating the very beginning of the downturn that Objective Analysis has been predicting for over a year.
It’s too early to tell if this is a trend. The DRAM and NAND flash spot markets, famous for providing a leading indication of the market’s main direction, has been relatively stable lately, with NAND flash prices seeing a drop of close to 30% since May, with an abrupt 20% reduction in August, but DRAM decaying much more slowly, at a 20% annualized rate, after a relatively dramatic 10% drop in March.
Most important is the fact that the recent spot market activity is not similar to the March DRAM drop or the May NAND flash drop, although August’s NAND decline may be partly to blame.
Still, the lack of a certain correlation between memory pricing and August’s lower revenue figure hints that the August decline in WSTS monthly semiconductor revenues is not a cause for concern.
In short, it’s really hard to tell from this data whether or not to worry.
Still, Objective Analysis will be closely watching this market and reporting it to our clients.
If you would like to know that you have the most current market data, seasoned with a deep understanding of what it means and where it is likely to lead, please contact us at http://Objective-Analysis.com.


Hi JIM – I thought I would share with you the Cowan LRA Model’s monthly ‘take’ on the WSTS’s August 2024 HBR sales numbers relative to its latest expectations for worldwide sales and yr-o-yr growth forecast for this year.
Based on the recently posted – (10-10-24) – WSTS’s August 2024 global chip sales HBR, the Cowan LRA forecasting model is predicting a full year 2024 global semiconductor sales result of $610.6B which jumped up from last month’s forecast of $603.5B. This newly calculated sales forecast number corresponds to an updated 2024 yr-o-yr sales growth prediction of 15.9%, which is significantly up from last month’s result of 14.6%, and is almost equal to the WSTS’s SPRING 2024 market forecast of 16.0% released at the beginning of June, 2024.
It should also be highlighted that according to the WSTS’s published August 2024 HBR, the cumulative year-to-date sales through August 2024 came in at $395.574B which when compared to last year’s corresponding August 2023 YTD sales number of $331.328B, yields an actual 2024 cum YTD (thru August) sales growth of 19.4%.
Therefore, the actual cum YTD sales growth through August 2024 likewise jumped up strongly from last month’s reported result of 18.1% indicating a possible strong uptick – implied inflection point? – for 2024’s sales growth which is higher than both the Cowan LRA Model and WSTS are presently predicting. Stay tuned, time will tell.
Regards. Mike C
Mike! Good to hear from you.
Your model has certainly been a stable member of the numerous forecasts the industry uses. Thanks for letting us all know what you are seeing at the moment.
This cycle and the last two (2018 & 2022) have been particularly challenging to predict, since they have all involved demand changes – something that used to be really rare in the industry. Before 2018 there was the 2008 Global Financial Collapse, and before that the Internet Bubble Burst in 2001.
Now we have AI being built out in advance of appreciable revenues. This looks like it could create an AI capacity glut.
Time will tell.
Meanwhile, keep cranking on your Cowan LRA model. The industry needs to know all the different forecast approaches, so that they can choose one they feel comfortable with when making production plans.
Best,
Jim