Perhaps some officials will not understand: how can the depreciation of equipment in the semiconductor industry be so fast?

In other mechanical and electronic industries, can't equipment depreciation be replaced in 10~15 years? This has not been considered for tinkering and continuing to be used.

Moreover, why did Zhang Zhongmou say that "equipment replacement is not bad, but because of Moore's law, when the year is up, it will fall behind, and it will be eliminated directly before it is broken"?

A little knowledge of the chip foundry industry and Moore's Law will not be surprised by such a periodic law.

Because TSMC is also like this in later generations, basically every 3 years, it has to change a wave of equipment and the latest technology, and then the price of processed products of the next level of technology begins to drop.

If you are reluctant to "squeeze toothpaste", others will overtake you, and you will not even be able to maintain your current profit margin.

Everyone is forced to "squeeze toothpaste" by the speed and cycle of research and development.

In this way, others will know that they can't catch up with you, and they will always lag behind you a generation and a half after spending a lot of research and development costs, so they will completely give up resistance and obediently admit your monopoly profits.

Take the current Intel thing as an example to analyze, for example, when Intel's 486 appears in 88, then the processing technology and processing accuracy equipment prepared for the 286 in 82 will be in a state of small profit operation, and the processing technology prepared for 8086 in 80 years will be processed and handed over to small factories that make single-chip microcomputers, and large factories such as Xiangji will no longer be able to look at it.

Basically, it is "486 windfall profit period, but also 386 medium profit period, 286 small profit period, 8086 external processing period".

According to this algorithm, the depreciation rate of the production line of foundry enterprises in the chip and semiconductor industry is only 6~8 years to be eliminated.

That is to say, when the factory was built last year, the original equipment procurement investment of 6.7 billion US dollars spent by Gu Biao was basically disposed of in 91~93. (Of course, the production over the years has also brought back the cost, and the income has to be continuously invested in the purchase of new equipment)

And this kind of "processing" is not directly thrown away, it is nothing more than something bought in 85 years for 400 million US dollars, when you sell it in 91, you may be able to discount 50 million US dollars and sell it to those small factories that make single-chip microcomputers at that time, or take two more years to survive to 93 years and sell it for 2000~30 million US dollars.

This is the fate of "not running to death" that chip foundry companies must endure, they must keep running and die if they stop.

At the same time, it is precisely because "the equipment is not bad, but backward and disposed of when the age is up", so as long as an OEM company can find a sufficient number of orders to ensure that your equipment operating rate is very high, then the depreciation cost of your equipment will be much lower than others, and your competitive advantage will snowball.

It is precisely like this, even if Xiangji does not make money when it does OEM 68,000 raw materials for Tiankun, this list will be on the list. Because the larger you are, the lower the equipment depreciation and amortization.

……

The crux of the matter now is that Zhang Zhongmou is confident that he will integrate production capacity and improve efficiency within two years to reach 20 million CPUs per year, while maintaining annual depreciation and amortization of $100 million.

However, if 20 million pieces of Motorola 68,000 are really produced, can this market digest them?

At present, Motorola's 68000 series CPU, the largest customer on the planet, is Gu Biao's Tiankun game console.

Other arcade machines, such as Sega's arcade machines, as well as some niche brands of personal computers, will also choose this CPU, but after all, the sales are too small to use much at all.

The game console of Tiankun, since its launch in 83, has only sold 300~4 million units in the first year (after all, it has only been sold for 7 months).

In 84, because of the perfect ecology and the abundance of game software, it began to blow out, reaching a growth rate that doubled that of Atari at its peak (also because game consoles were cheaper than in the 70s), and at the end of the year, the market stock of Tiankun game consoles exceeded 10 million units.

85, 86, will continue to grow, especially after the 86 years of "Tiankun learning machine" came out, high and low configuration at the same time, the sales tide will be more rapid.

However, even so, according to the estimation of Tiankun Entertainment, by the end of 1986, the total historical sales of Tiankun game consoles in the world will not exceed 24 million units...... It would be nice to sell 3 million units in half a year.

Excluding the ones that have been thrown away in the past few years, the global stock is about 24 million - including 15 million units in the United States market, 3 million in the local market, and more than 5 million in Europe and other markets combined.

This result is actually much better than Nintendo's FC in the same period in another time and space history, and the FC in another history was sold until the launch of SFC in 90, and it accumulated more than 50 million sales. And you must know that this kind of product will definitely sell more in the next few years, because Moore's Law will make things lower in price.

(Note: The official statistics of FC sales are 63 million, but that was until the beginning of the 21st century, when it was completely discontinued, and the last 10 million or so was the sales in the 90s, when FC technology was already very backward, and there were still consumers in third world countries who were buying.) ）

In other words, according to this account, Zhang Zhongmou increased the production capacity of Xiangji to the scale of "20 million pieces per year in 1988", and the Tiankun game machine could not be digested at all.

Tian Kun does not have 20 million game consoles a year to sell.

Tiankun can't do it, and other niche products can't be digested.

In order to achieve the ideal state of "amortizing the depreciation of the lithography machine and other packaging equipment to full capacity" to obtain additional profits and competitive advantages, Zhang Zhongmou had to find a new digestion customer that was almost slightly inferior to the Tiankun game machine.

He wants to bet that in the next two or three years, in the field of global personal consumer electronics, there will be another explosive model that needs at least millions of units every year.

Otherwise, obediently accept "make $5 less per CPU" and give up the pursuit of making full use of resources.

In fact, even if you make less than $5 per piece, it is not worse than other peers in the world, at most it is the same as me - in the mid-to-late 80s, the demand for high-end CPUs on the earth was so sluggish, and no one could eat enough production capacity. It is the norm for all peers that the equipment cannot be fully connected until it is eliminated.

If you can do that, then you have an extra snowball advantage over others, which is an additional question.

Intel is currently producing its own products and does not look for foundries.

AMD, which competed with Intel in later generations and barely maintained it, has not fallen out with Intel yet, and has no self-built brand at all-

Although AMD was founded in 1969, for the first 18 years of its founding, until 1987, AMD existed as Intel's design outsourcer.

Historically, AMD got Sony's CMOS chip design/foundry order in February 87, and the waist pole suddenly stood up, feeling that there was no need to rely on Intel's snort to survive, and then immediately trembled, and in April of that year, it settled old grievances and overlord clauses with Intel, and began to fight a lawsuit, and finally AMD spent four or five years in a litigation cycle before officially winning the lawsuit in 91.

AMD's first completely independently developed and designed CPU, AM386, was not put into production until March 91.

Similarly, because there was no butterfly effect in the original history, Zhang Zhongmou's TSMC entrepreneurship was relatively late, and people only started in 87 years, and the speed of raising money was slower, and it was only in 90 years that production capacity was formed, and then after receiving a small order to brush up on the prestige of the rivers and lakes, it caught up with AMD in 91 "There is design and research and development, but there is a lack of foundry", and then TSMC and AMD hit it off, and both escaped the fate of "lack of external force in the first cycle of pushing the flywheel in the early days" and died.

But now, Gu Biao, the butterfly factor, has pushed Zhang Zhongmou and Xiangji three or four years faster, but AMD has not kept up, so this trump card that originally made Zhang Zhongmou's early production capacity successful in history cannot be played.

AMD can't count on it, and Nvidia (which NVIDIA makes the graphics card) can't count on it, and Nvidia doesn't know where it was at the end of the 80s.

According to history, Nvidia was established in 93 years, and it was only after looking at the success possibility of the AMD+ TSMC model that it was established, but it was aimed at the subdivision of "graphics card design + finding foundry outsourcing".

……

Without AMD, without NVIDIA, with Zhang Zhongmou's unopened vision, the options that can be thought of are very limited, and they are almost ready to come out with a little brain.

His concerns about major shareholders are explained in this way:

"My plan is to get the CPU foundry market for Apple's next-generation Macintosh computer by the time of '88, if Jobs can further expand the market share on the basis of the current generation.

I think that with the overall price decline curve of the personal computer industry, the number of people around the world who will buy the next generation of Macintosh computers after '88 should be able to fill the surplus capacity of our production of Motorola CPUs - if Steve Jobs can sell 3 million Macintosh next-generation computers a year.

I know that according to your layout some time ago, what you hope is to cooperate with Wang An Computer in the future. However, Wang An's computer is a word processing machine, which is originally a relatively cheap functional model and does not require a very good CPU. Even if we convince Wang to switch to Motorola's 68000, or other non-Intel products, we may be counting on them to use 'the highest-end CPU that just came out in 1988'.

In addition to Wang An Computer and Apple, several other major personal computer manufacturers that are currently mainstream are all Intel-based. Intel, on the other hand, is responsible for the production process itself and will not outsource. The more people in the world use Intel, the smaller the total market size that needs us to do foundry - so, in this case, we can only count on Apple. ”

Apple's Macintosh computer, which is the "MAC operating system" computer that the world is familiar with, was released in 1984, and the second generation was around 1988, and Motorola's 68000 series CPUs were used in history-

Although Zhang Zhongmou is not a traverser and a prophet, he can rely on his industry acumen to speculate that "since Apple is focusing on the design and development of the whole machine around the Motorola 68000 series CPU, as long as Motorola's CPU is upgraded, Apple will also launch an upgraded computer design."

This does not need to be a prophet, and anyone with discernment can deduce it.

Historically, in the field of personal computers, it was Apple, as Motorola's main ally. This alliance relationship began around 85 years, and entered the second generation in 88, and in 1991 it was an additional IBM that brought in IBM, under the joint research and development of Motorola, IBM and Apple, the AIM alliance standard was launched, and the PowerPC model was developed by Motorola, which is the next generation of the 68000 top version of the product, benchmarking Intel's 586.

(For IBM, it was a two-headed bet at the time, after Microsoft launched MS-DOS5.0 in 91, it began to expose greater ambitions, and began to prepare for BM to find that it led the "open system computer" plan, which may not be able to get the maximum benefit as a hardware integrator, and may be bypassed by Microsoft + compatible machine manufacturers.)

So Apple and Microsoft dog-eat-dog were added. Microsoft in 91 is already a bit of a tiger for IBM. Just because Microsoft and IBM were on good terms in '86 and Apple was on the other side of the equation doesn't mean that this relationship between friend and foe will continue. This confirms Winston's saying, there are no eternal enemies, no eternal friends, only eternal interests)

Wang An does not need the latest technology and the highest level of CPU, and other factories use Intel, and no matter how high the sales are, there is no need for foundry. So it's normal for him to turn his mind to expecting Jobs to be a little more powerful.

But Gu Biao obviously had other arrangements.

"Don't count on Jobs, he can't make it. But don't worry, by 1988 you'd see a new PC on the market that sold better than the MAC-II and used the Motorola 68,000 next-generation CPU. ”