The 1985 Semiconductor Depression vs. the 2022 Semiconductor Demand Shock
In late 1983 and into 1984, times were amazing and the best they had ever been in the semiconductor industry. Until they weren’t.
Related and recent articles
• 5 Things You Need to Know About U.S. Stock Buybacks • Semiconductor Socialism — Why Such a Rush to Pass the CHIPS Act? • The 1985 Semiconductor Depression vs. the 2022 Semiconductor Demand Shock • WTF! Our Taxpayer Money Will Pay for Intel’s $130 Billion in Stock Buybacks and Dividends? • 5 Things You Need to Know About U.S. Stock Buybacks • (Part 1) The 5 Most Important U.S. Economic Events Over the Last 50 Years • Did This Happen by Accident to 89% of America’s Stock Market Wealth?
(Subscribe to receive email notifications when I post new articles.)
(Quick note: If you have not read “WTF! Our Taxpayer Money Will Pay for Intel’s $130 Billion in Stock Buybacks and Dividends?” I recommend this article also. There is important and relevant content here that you are unlikely to see covered by mainstream cable news media companies.)
Times were good in the semiconductor industry in late 1983.
Really good.
The initial boom in personal computer manufacturing was in full swing, and all of the then-new PC companies were ordering semiconductor chips like there was no tomorrow. The book-to-bill ratio — a closely watched metric in the semiconductor industry — registered its highest reading ever at 1.66 in December 1983.
Everyone in the semiconductor industry was pumped, and every semiconductor company was hiring aggressively and staffing up in every department as rapidly as they could.
That’s why it was so surprising when only 12 months later, the book-to-bill ratio dropped to its lowest reading ever at 0.58; everyone was in panic mode; and the industry was entering what would become the worst two-year downturn that it had ever seen.
What happened?
And what can this tell us about what might happen in 2023 and 2024?
What was going on in 1983 in the semiconductor industry and in the just-being-born personal computer industry?
In 1982 and 1983, there were probably 15 to 20 small, medium, and large companies with some kind of a presence in the new personal computer (PC) industry. They were all working hard to bring their own vision of a personal computer to the marketplace and hoping to dominate what everyone expected would be a massive new industry segment.
But that only sets the stage. The story gets a lot more interesting.
As each of these companies made their projections for how much of the market they thought they could win, it was typical for each company to be planning on getting a large chunk of the PC market for themselves. It’s probably safe to say that few companies went into that new business unless they were hoping to win at least 20% of the then-new market.
Here’s the problem.
If you have 15 companies each ordering semiconductor chips with the idea in mind that each of them is going to have at least 20% of the market, then do the math.
15 companies × 20% market share per company = 300% of the market.
But remember — there’s only “100%” of any market available to be won.
Basically, in the aggregate, all of these companies placed orders for chips that represented probably at least 3 times the actual personal computer demand that the marketplace was going to require for chips over that next year or so.
So that alone meant that the industry was probably in bubble territory and was going to have an unhappy ending.
But wait, it got worse.
So much worse.
What was the Book-to-Bill Ratio metric?
It was simply:
(1) the dollar value of all of the ORDERS RECEIVED by semiconductor companies in a given month
divided by
(2) the dollar value of all of the ORDERS SHIPPED by semiconductor companies in that month.
If the ratio was greater than 1, that meant that companies were receiving orders worth more than the orders they were shipping. And that meant the semiconductor companies were building a backlog of orders.
Backlogs of orders (up to a certain point) were a GREAT thing.
Backlogs meant that if there would be a downturn in orders, the companies could dip into their backlog of orders as a cushion and keep their factory lines and shipments running at a steady level.
In a manufacturing business like that, knowing that you could count on steady demand made manufacturing planning much easier, and it was generally good for profitability.
Bottom line, a Book-to-Bill ratio >1 was a good thing.
For example, a Book-to-Bill ratio of 1.2 meant that for every $1,000,000 of orders that semiconductor companies shipped in a given month, they were receiving orders worth $1,200,000. So they were building $200K worth of backlog that month for every $1M they shipped.
Conversely, if the Book-to-Bill ratio were LESS THAN 1, that meant that the semiconductor industry had to eat into whatever backlog cushion of orders that they had already on the books from previous months in order to keep their production levels steady.
If semiconductor companies had some amount of backlog cushion, they were able to use that cushion to smooth over cyclicality in orders and short-term downturns.
But if they ran out of backlog, then it was like a car engine running out of lubrication — the engine was going to seize up, and the car would come to a halt...and might crash in the process.
Keep this in mind — what you just learned is going to come in handy in a moment.
And that kind of engine seizing up along with a car crash is exactly what happened to the semiconductor industry starting in 1985.Double-ordering and triple-ordering in the face of long lead times for semiconductor chip deliveries
Back in the 1980s, if delivery times for semiconductor chips were stretching out and becoming 3 months instead of 2 months . . . or 6 months instead of 3 months, then it was common for electronics companies to also place a second order with another semiconductor company for the same quantities of the same chips.
They would do this “double-order” for a little bit of insurance as a back-up in case the order with the first semiconductor company either took longer than expected to be delivered or just wasn’t going to get delivered at all (for whatever reason.)
But, as you probably already have guessed, these double orders would cause the semiconductor companies to push delivery times for new orders out even further.
And in 1983 as more and more orders piled up with semiconductor companies, some of these electronics companies even began triple-ordering the same chips. Double-ordering wasn’t enough.
Can you just feel this bubble getting larger and larger as you read each sentence?
So in January 1984, the December 1983 Book-to-Bill ratio was initially pegged at an all-time high of 1.62 . . . and revised the next month upward to an even higher all-time high of 1.66.
And then the bubble began to pop!
The farther into 1984 that the industry got, the more obvious it became to each of these personal computer makers that their sales were not living up to their original business plans and expectations.
Each of these PC makers had so many orders on the books with the semiconductor companies, that the nascent PC companies started to panic that they might have to actually take delivery on truly huge volumes of semiconductor chips that they weren’t going to have any use for.
So what did they do? The obvious thing. They started to cancel those double and triple orders. And they eventually work their way down to canceling the original first orders.
What was happening at the semiconductor companies while all of this was going on?
As the initial order cancellations came in, semiconductor companies started to decrease the delivery times on the orders that had the farthest out shipping dates.
And as PC makers started to get notified that the chip order that was originally going to have shipped in 6 months was actually going to ship in 6 weeks or 2 weeks, well, these PC makers got even more aggressive about canceling orders.
This bubble-supported tower in the semiconductor industry began to collapse with astonishing speed.
By December 1984, exactly 12 months after the all-time high for the Book-to-Bill Ratio (1.66), the semiconductor industry printed an ALL-TIME LOW reading for the Book-to-Bill ratio of 0.61, which was shortly thereafter revised down to 0.58.
Let me emphasize this.
It only took 12 months for the semiconductor industry to go from (1) truly believing that they were in the best business times ever to (2) being in utter panic mode and knowing that they were in the worst times ever.
That’s how fast things can change.
The Beer Game
What I described above is not unique at all to the semiconductor industry.
You see the same thing play out time and time again across a wide range of industries and businesses.
Wherever you have a supply chain with human beings at every link of the chain making decisions about how much inventory would be smart to hold, you will see this kind of behavior happen.
As a matter of fact, the problem is so common that back in the 1960s, MIT Prof. Jay Forrester developed a simulation tool called “The Beer Game” to teach important supply chain concepts, including:
- How single parts in a system influence other parts.
- How individual thinking differs from thinking out the overall supply chain/ecosystem level.
- Challenges faced in today’s supply chains.
- Potential solutions.
This video is an excellent 4-minute introduction to the beer game if you would like to learn more.
Takeaways, Part 1. What might cause a rapid downturn in the semiconductor industry in 2022 or 2023?
What does the semiconductor depression of 1985–1987 teach us about what might happen over the next couple years with today’s global semiconductor industry landscape that has evolved quite a bit over the past 40 years?
I am now shifting into pure speculation mode. I don’t have a crystal ball, and for most of this, I don’t have too strong a conviction about what might or might not happen. I’m just offering thoughts about possibilities.
In order for the semiconductor industry to get into trouble again the way they did back in the 1980s — and at other points in its history — there needs to be a big mismatch between supply and demand. I don’t think this supply/demand mismatch is likely to happen from a rapidly increasing supply. It just takes way too long to order, take delivery, install, and bring new semiconductor fabrication lines up to speed to the point where they can ship product for new chip-making supply to catch anyone by surprise.
So if there is going to be an unexpected mismatch between supply and demand, it seems more likely that any surprise is going to come from the demand side of the equation.
We have already seen over the past 2 years what happens when (1) semiconductor demand increases unexpectedly and (2) supply gets constrained at least somewhat just because of global supply chain issues resulting from pandemic-related issues.
(For instance, think about car companies not being able to ship almost-completely-assembled $40,000 cars because there are still waiting on delivery of a few $5 semiconductor parts.)
What if new Covid variants or Omicron subvariants keep making it difficult for businesses and factories to maintain the levels of production that they want to? This would certainly have ripple effects and “bullwhip effect” propagating back up through the supply chain all the way back to the semiconductor companies.
Even a relative Covid success story like Japan is in danger of seeing exactly this scenario play out:
Japanese companies are temporarily shutting offices or suspending production as they battle a record wave of COVID-19, disrupting businesses in a country that has until now weathered the pandemic better than most advanced economies.
Automakers Toyota Motor Corp and Daihatsu Motor Co last week halted production line shifts because of employee infections. KFC Holdings Japan Ltd has had to shut some fast-food restaurants and move staff to fill gaps, while Japan Post Holdings Co has temporarily shut more than 200 mailing centres….
Work stoppages and no chatting at lunch: Japan Inc grapples with COVID (Aug 04, 2022)
It’s not hard to imagine similar issues here in the U.S. in the coming months and quarters.
Takeaways, Part 2. If it does happen, what might the repercussions look like?
Ok, if it were to play out like this, semiconductor companies could quickly go from too little capacity to too much capacity.
Boom-and-bust cycles have been part of the semiconductor industry’s DNA for half a century — there is nothing to suggest that these boom and bust cycles have disappeared forever.
Well, if this were to happen, the good news is, the shipping times on your new iPhone or MacBook Air are likely to be a lot faster. The products might even be a bit cheaper than they otherwise would have been.
The “less-good” news is that the substantial investments in the semiconductor industry that the federal government and the State of Ohio are making will start to look foolish, at least in the short-term.
The mainstream media might even take this opportunity to finally start questioning the wisdom of pouring tens of billions of dollars in taxpayer money into semiconductor companies that had blown hundreds of billions on stock buybacks and cash dividends over the past 10 to 20 years.
I mean, if the semiconductor industry ends up being awash in excess capacity, then plants will potentially stop operations because they don’t have the order volume to keep the fab lines running. At that point, Ohio and the Feds having committed tens of billions of dollars over the next few years — and, I suspect, more like hundreds of billions over the next 10–20 years — will be embarrassing, to say the least.
Takeaways, Part 3. In summary, here are the key takeaways for today:
- Things can change a LOT faster than you expect.
- Things can change a LOT more than you expect.
- Think hard about who is actually going to get left holding the bag if the bubble pops. Is it the companies? Is it the governments who have provided these incentives and cash? Is it the politicians who spent “the people’s money” to get themselves quick boosts in the polls in advance of the 2022 midterm elections? Will it be the people themselves whose tax money funded all of this…and might have to keep funding it for a long time…? What might any/all of these players do to minimize their own reputational damage and push it toward someone else? What will this almost-certain game of musical chairs mean for the semiconductor industry going forward?
- If you have been thinking about ordering, say, a new Macbook or replacing your now-old iPhone and you hear anything significant about a supply chain disruption that would impact TSMC — I’m thinking in particular of any real threat from China toward Taiwan — go immediately to your nearest Apple Store and buy whatever they have in stock that is closest to your ideal device configuration. Don’t worry about buying the perfect device fit. Get what you need while you can, and don’t second-guess yourself. Same thing with any other big purchase you have been considering if a sudden semiconductor shortage would hurt your ability to make the purchase.
Related and recent articles
• 5 Things You Need to Know About U.S. Stock Buybacks • Semiconductor Socialism — Why Such a Rush to Pass the CHIPS Act? • The 1985 Semiconductor Depression vs. the 2022 Semiconductor Demand Shock • WTF! Our Taxpayer Money Will Pay for Intel’s $130 Billion in Stock Buybacks and Dividends? • 5 Things You Need to Know About U.S. Stock Buybacks • (Part 1) The 5 Most Important U.S. Economic Events Over the Last 50 Years • Did This Happen by Accident to 89% of America’s Stock Market Wealth?
(Subscribe to receive email notifications when I post new articles.)
Again, thank you for reading, subscribing, clapping, and sharing — your time and attention are deeply appreciated!