The R&D team of Texas Instruments, their strength is beyond doubt.
Most of Xingyi Semiconductor's core R&D talent currently comes from Texas Instruments, and they are the cream of the crop, so their strength is naturally impressive.
Without any surprises, the smart home SOC chip, fast charging chip, and power management chip have all successfully gone through tape-out.
Next is trial production.
If trial production also goes smoothly, then full mass production can begin!
At that time, Xingyi Semiconductor will transition from the PPT era into a new era of official mass production!
Although these three chips are not high-end chips, everything is difficult at the beginning.
With these three chips leading the way, subsequent high-end mobile phone chips, baseband chips, and even next year's high-end SOC chips with integrated baseband will naturally follow.
Developing a 28nm SOC chip with integrated baseband directly this year is too difficult; after all, even Qualcomm will only achieve it next year.
No matter how capable the people poached from Texas Instruments are, it will still take until next year.
After all, baseband issues are even harder to resolve than processors, and integrating baseband makes it even more challenging.
Therefore, Wang Yi had to make a trade-off, slowing down the R&D of SOC chips with integrated baseband, and switching to an external baseband solution.
Separately developing a 28nm processor chip and a 28nm baseband chip.
This external solution, similar to Nvidia's Tegra 3 and Qualcomm's 8064 flagship processor in the second half of the year, greatly reduces difficulty and is sufficient.
It won't be too late for Xingyi Semiconductor to launch its integrated baseband Xingyi SOC in the first half of next year, when Qualcomm releases its integrated baseband Snapdragon 800 SOC.
After all, next year's Nvidia will still use an external solution.
This strategy is more stable and safer.
Moreover, with separate R&D, the processor not integrating the baseband can largely cut modules not needed for TVs, optimize video decoding capabilities, and be used as a TV 4K chip!
This way, it kills two birds with one stone!
Even without cutting features, directly mass-producing more units and using mobile phone chips as 4K TV chips is also fine.
Xiaomi TV generation one in the previous life used the Snapdragon 600 chip, which was the same Snapdragon 600 used in mobile phones.
Although this solution has slightly higher material costs than the cut-down version, it saves a series of processes like re-modifying the design, re-tape-out, re-trial production, and re-opening production lines.
Overall, the cost is actually lower.
After all, with so much code in chip design, changing just one line requires a re-tape-out.
The cost of a single tape-out is not low.
The chips Xingyi Semiconductor taped out this time are 40nm low-end chips, with relatively low tape-out prices, around 800,000-900,000 US dollars, which is about five million RMB per run.
But the price for 28nm is much higher.
Currently, a single 28nm tape-out costs 5 million US dollars, which is over 32 million RMB.
A few years later, when the 28nm process matures and capacity is abundant, the tape-out price will drop to 2 million US dollars.
These things are unavoidable.
Moreover, a tape-out is not necessarily 100% successful.
Xiaomi's Surge S2 chip in the previous life failed multiple times during tape-out, costing hundreds of millions of RMB, which even the wealthy Lei Bus couldn't bear, forcing him to abandon the S2 chip R&D.
Similarly, given the high current 28nm chip tape-out costs, Xingyi Technology can directly choose to use its 28nm mobile phone chips as TV chips, saving on re-R&D and re-tape-out.
Next year, they can then develop more high-end, professional TV image quality chips.
This is all perfectly fine.
Having thought this through, Wang Yi came to the semiconductor department.
Williams smiled and greeted him:
"As expected, all three chips successfully taped out on the first try!"
"Very good. Xingyi Semiconductor Department has made a great contribution. Each member of George's R&D team will receive an additional three months' year-end bonus!"
"I thank Chairman Wang on their behalf!"
Williams said with a smile.
The R&D for these chips was done by George's department, so the bonuses will go to George and his team.
As for Williams' department, their bonuses depend on whether the mobile phone chip R&D succeeds.
If R&D succeeds, more bonuses.
If R&D fails, they'll go hungry.
Wang Yi has always been clear about rewards and punishments, even within the same department, he implements quantified management, refusing to allow a "big pot of rice" mentality.
"Oh, by the way, Chairman, all three chips have undergone trial production, we should officially name them. Also, the mobile phone chip we are currently developing also needs an official name!"
"Then let's name them by category,"
Wang Yi mused slightly:
"Smart home, abbreviated as SH is a bit cumbersome, so let's just use H for home. Our first smart home chip will be called Smart Home H1 chip."
"Power, so the first power management chip will be called P1."
"Charge, the first fast charging chip will be called C1!"
Williams agreed wholeheartedly:
"This naming is simple and clear, and even if we have more chips later, it won't be confusing."
Wang Yi nodded:
"Yes, and these low-end chips are not mobile phone chips; they don't require high performance and don't need annual upgrades. On the contrary, one generation of R&D can be used for several years."
By the time the next generation is needed, it will likely be many years later, when the Xingyi wafer fab achieves high-process manufacturing, freeing up the 28nm process, to create second-generation 28nm smart home, fast charging, and power management chips.
Before 2017 in the previous life, most smart home chips were 55nm, with 40nm being the top.
It was only after 2017 that new 28nm products gradually appeared.
As for 10nm and 7nm smart home chips, that's unrealistic and a pure waste.
After all, 10nm and 7nm mobile phone chips can sell for over 1,000.
But a smart home chip, at most, has a unit price of a few tens of yuan, so it's not worth using a high-end process at all.
Williams said with a smile:
"That's right, if we mass-produce more units at once, the cost can also be spread to the lowest! After trial production, we need to officially start production. Chairman, how many orders should we place for those three chips?"
Wang Yi fell into thought:
"How many orders to place, this is an important question! The more we order, the lower the foundry cost, and the lower the overall cost. Of course, not too many, to prevent excessive inventory."
"Exactly." Williams handed over a document, "Chairman, this is TSMC's quote, and our calculated costs. Please take a look."
Wang Yi took it and examined it carefully.
Williams had done his work very thoroughly, not only marking the foundry fees, material costs, and amortized total costs corresponding to different production volumes, but also indicating the current purchase prices of similar chips.
This made Wang Yi very satisfied.
Currently, Xingyi phones and Xingyi tablets use power management chips purchased from companies like Qualcomm and Nvidia.
Wang Yi's large purchase volume meant that the prices offered by the suppliers continuously decreased.
For example, the two power management chips for xphone 1 came from Qualcomm.
Xiaomi bought two for a total of 10 US dollars, but they gave Wang Yi the price of 7 US dollars.
Later, as xphone 1 orders increased, it dropped to 6 US dollars, meaning only 3 US dollars per chip, or 20 RMB, which was considered a dirt-cheap price.
But Qualcomm still made a killing!
The reason being, the material cost of a power management chip is only a few dimes!
For chips, material costs are very low, and mass production foundry fees are also very low.
The biggest cost is R&D.
Qualcomm has a large output, so the R&D costs are spread out.
Add in TSMC's foundry fees, and the total cost of a single power management chip is at most one dollar!
Therefore, whether Qualcomm sold it to Wang Yi for 20 RMB, 30 RMB, or to Xiaomi for 35 RMB, Qualcomm still made massive profits.
If you can't make it yourself, you have to spend an extra twenty or thirty yuan to buy it.
These things are unavoidable.
Twenty or thirty yuan per phone, for ten million phones, that's two or three hundred million yuan!
xphone shipments exceeded 15 million, and two power management chips per phone meant Qualcomm earned six or seven hundred million yuan.
Now that Wang Yi has achieved self-R&D of power management and fast charging chips, he can also drive prices down.
The material cost provided by TSMC is also terrifyingly low, only a few dimes!
And the R&D expenses for George and his team developing these chips weren't much; the main cost was tape-out fees.
Then there are foundry fees and production line costs.
Adding all these costs together gives the total cost.
The more you produce, the lower the amortized cost.
As Williams calculated, producing one million P1 chips would amortize to a cost of 30 yuan per P1 chip, which is 50% more expensive than buying Qualcomm's power management chips!
But producing ten million P1 chips would amortize to a cost of only 5 yuan per P1 chip, which is only a quarter of Qualcomm's lowest price to Wang Yi.
If directly producing one hundred million P1 chips, the amortized cost per P1 chip would be only 1 yuan.
Yes, it's that cheap.
After all, it's something whose material cost is only a few dimes.
This is the power of self-R&D.
Take the Beidou terminal chip, for example.
Initially, we couldn't make it, and foreign companies sold it for 1,000 RMB each.
Once we mass-produced our first generation of self-developed terminal chips, foreign companies lowered their prices to 500 yuan each.
After we mass-produced the second generation of self-developed chips, foreign companies again dropped the price to 200 yuan.
Then 100 yuan, 50 yuan...
After several generations, domestic terminal chips fully caught up with foreign chips, and they lowered the price to 6 yuan.
Yes, the terminal chip that once cost 1,000 yuan later cost only 6 yuan, and they even begged us to buy them.
After all, their R&D costs had long been amortized.
Now, the total cost is only a few dimes.
Let alone selling for six yuan, they would make a profit selling for one yuan!
Chips are just that special; material costs are very low, the key depends on R&D costs and production volume!
Similarly, the fast charging chip C1 and smart home SOC chip H1 have similar prices.
If one hundred million units are mass-produced, the amortized cost per chip is only 1 yuan!
Wang Yi really wanted to order 100 million units directly, but he also had to calculate if the market could absorb them.
With this in mind, Wang Yi began to carefully consider.
The demand for fast charging chip C1 and power management chip P1 would depend on the sales of future Xingyi phones, Xingyi tablets, and Xingyi TVs.
In fact, a charging chip is also a type of power management chip.
The only difference is that a charging chip only handles charging management.
A power management chip, on the other hand, handles overall power management, such as cutting off part of the power supply under certain operations.
The simplest example: when the screen is off, only the screen's power control is cut, but the power supply to other components of the phone remains, which is achieved through the power management IC.
As mobile phones upgrade, more and more power management chips will be needed.
A 3G phone only needs two power management chips, which is sufficient.
A 4G phone, however, needs 4-6 power management chips!
A 5G phone even needs 8-10 power management chips!
Currently, xphone and Xingyi Tablet X1 both use two Qualcomm power management chips.
And xphone pro and Xingyi Tablet X1 pro both use two Nvidia power management chips.
But by Xingyi Generation 2, the number of power management chips will need to increase.
Even if Xingyi Generation 2 doesn't support 4G and remains a 3G phone, it will still need 2 P1 power management chips, plus one C1 fast charging chip.
If it supports 4G, that would be 4-6 P1 power management chips, plus one C1 fast charging chip.
However, for a 2012 phone, there's no need to use 4G.
After all, commercial 4G in China started in 2014.
This year's iPhone 5 isn't a 4G phone either.
The reason being, the iPhone 5 in the previous life used a Qualcomm baseband that supported 4G frequencies, but the Qualcomm transceiver it used did not support 4G, so it couldn't receive 4G signals.
Therefore, the iPhone 5 was just a 3G phone.
It wasn't until next year's iPhone 5C and iPhone 5s that 4G was supported.
Similarly, Xingyi phones don't need to rush into 4G.
Although 4G has long started in Europe and America, 4G in China won't be until 2014, and Xingyi phones cannot be sold overseas yet, so there's no rush to implement 4G until the second half of next year.
Thus, over the next year and a half, a Xingyi phone will require two power management chips + one fast charging chip.
The same goes for Xingyi tablets.
Calculating the time, the P1 power management chip successfully completed trial production and began mass production in March.
It could be used in Xingyi phones and Xingyi tablets as early as April.
With self-developed chips, there's naturally no need to continue purchasing Qualcomm and Nvidia power management chips.
That's right, after consuming the inventory of Qualcomm and Nvidia power management chips costing 20 yuan each, Xingyi phones and Xingyi tablets will fully switch to self-developed chips costing one yuan each!
Over the next year, xphone will continue to sell, targeting the mid-range market.
With monthly sales of 1.7 million units, annual sales can reach 20 million units.
xphone pro will also continue to sell, with another 20 million units shipped in the coming year not being a problem.
Xingyi Tablet X1 is on global sale, with another 20 million units shipped in the coming year, facing no pressure.
Xingyi Tablet X1 pro, as the strongest quad-core tablet, is a global bestseller, with hopes of shipping 30 million units in a year.
Adding in Xingyi Generation 2 in the second half of the year, with sales estimated at 30 million units.
All combined, over the next year, Xingyi Technology will sell 120 million mobile phones or tablets, requiring a total of 240 million power management chips and 120 million fast charging chips!
Adding in Xingyi TVs and smart home products, which also need power management chips...
Like AI speakers, they need power management chips, and other products do too.
In essence, almost any smart home device requires a power management chip.
Thus, the demand for power management chips will need to increase!
With this in mind, Wang Yi looked at Williams:
"William, directly order 300 million power management chips. In one year, have TSMC produce 300 million for us and drive the price down to the absolute minimum!"
"300 million units?" Williams was startled, "Chairman, do we need that many?"
Wang Yi nodded:
"300 million units will only be enough for one year, maybe not even a full year. Moving forward, all our phones, tablets, smart home devices, and TVs will use our self-developed P1 chips."
"Okay, a massive order of 300 million units, that will further reduce costs. The amortized cost is estimated to drop to eight cents!"
"Excellent!" Wang Yi beamed, "Originally, a power management chip cost 20 yuan, and we'll get it for 80 cents. For 300 million units, we save nearly 5.8 billion yuan. This is the power of self-developed chips!"
"Indeed, of course, the most important thing is our large output. Three hundred million units. If we only had three million units, the cost would be more expensive than buying from Qualcomm!"
Williams chuckled:
"So how many fast charging chips C1 should we produce?"
Wang Yi thought for a moment.
Fast charging chips are used less frequently; one per device is usually enough.
Most smart home appliances are plug-in and don't need fast charging chips.
At most, the next generation of power banks will require fast charging chips.
In that case, 150 million units should be enough.
However, considering the future promotion of the X-Charge 1.0 fast charging standard and the Type-C standard, Wang Yi decided to produce more.
"For the C1 fast charging chip, produce 200 million units in one year for now."
"That many? Won't we be unable to use them all?"
Williams was a little worried.
Wang Yi waved his hand:
"It's fine. When we launch the X-Charge 1.0 fast charging standard, achieving 10W fast charging and fully ushering in the fast charging era, how will our competitors react?"
Williams' eyes lit up:
"Competitors will want to quickly follow up on fast charging, but companies like Qualcomm can only offer 5W slow charging. So if they want 10W, they will have to buy our Xingyi Technology's C1 fast charging chip!"
Wang Yi nodded:
"Exactly, and our Xingyi Technology's C1 fast charging chip must be paired with our P1 power management chip to be used."
Williams was overjoyed:
"At that time, competitors will only be able to use our fast charging solution, purchasing our fast charging chip + power management chip. Then, two P1 power management chips + one C1 fast charging chip will be sold as a bundle. Qualcomm sells for 20-35 yuan each, we'll sell for 20-30 yuan each, 60 yuan for a set of three. The cost is only three yuan, that's huge profit!"
"Precisely, that's the strategy."
Wang Yi smiled slightly.
This was his overall strategy: after self-R&D, not only are they cheaper for internal use, but they can also be sold externally for substantial profits.
Once sold externally, 200 million C1 chips and 300 million P1 chips might not even be enough.
They would need to increase production capacity!
That's fine; in the early stage, the fast charging chips will be used internally, and then sold externally a few months later, increasing capacity as needed.
By the time Qualcomm's fast charging chips are almost developed next year, Xingyi Technology's fast charging chips will be aggressively priced to seize market share.
In the past, foreign companies used to do this to domestic companies; now Wang Yi is simply doing the same.
"Chairman, what's the order quantity for the H1 smart home chip?"
Williams asked again.
Wang Yi thought for a moment:
"The annual sales of a single smart home product are around ten million, but if there are many types of products, the total quantity will go up."
Smart homes require many chips; low-end ones use MCU chips, while high-end ones use the SOC smart home chip H1 developed by Xingyi Technology.
Logically, high-end products would use H1, while low-end products could simply purchase MCU chips from other companies.
But the awkward part is that even competitors' low-end MCU chips cost several yuan each, which is even higher than the self-developed cost of the high-end H1 smart home chip!
After all, they need to make a profit!
Therefore, there's no need to purchase MCU chips at all; it's better to directly use the self-developed high-end H1 chip, even if it's overkill...
This is the difference between purchasing and self-R&D.
In his previous life, Huawei's self-developed processors broke through blockades, directly pushing tablet prices down to 2099 yuan, causing the entire Android tablet market to collapse!
There was nothing they could do; Huawei's self-developed processors had a fixed cost.
Competitors purchasing Qualcomm processors had to pay a hefty price and simply couldn't compete.
With this in mind, Wang Yi said:
"For smart home chips, let's aim to produce 100 million units a year for now, and adjust later based on market conditions."
"Okay!"
Wang Yi remembered the mobile phone chip matter:
"As for mobile phone chips, Qualcomm processors are called Snapdragon, so we'll call ours Kunpeng!"
Currently, Qualcomm's processor naming scheme is still quite chaotic, 8660, 8064...
But early next year, it will all be adjusted: low-end Snapdragon 200, mid-range Snapdragon 400, Snapdragon 600, high-end Snapdragon 800.
The 28nm quad-core flagship processor APQ8064 released in the second half of this year is part of the Snapdragon 600 series.
And Xingyi Technology's first processor is also 28nm and uses an external baseband.
Its release time is half a year later than 8064, so it should have some performance advantage, but it won't be a match for Qualcomm's next-generation Snapdragon 800.
With this in mind, Wang Yi said:
"The first mobile phone chip will be called Kunpeng 700!"
Qualcomm Snapdragon 200, Snapdragon 400, Snapdragon 600, Snapdragon 800...
Xingyi chips will be Kunpeng 300, Kunpeng 500, Kunpeng 700, Kunpeng 900...
Follow Qualcomm's path, leaving Qualcomm no path to take, and even surpass Qualcomm.
"Kunpeng! That's impressive, even more impressive than Snapdragon!"
Williams was beaming with joy.
Wang Yi smiled slightly:
"Make the Kunpeng 700 with external baseband as soon as possible. Next year's new phones will feature Kunpeng 700. In the first half of next year, develop the integrated baseband Kunpeng 900. When Qualcomm's integrated baseband chips hit the market in the second half of next year, we must also launch our integrated baseband Kunpeng 900!"
"Alright, Chairman, rest assured, that's not a problem."
Wang Yi continued:
"As for the external baseband's name, the 3G baseband will be called Winged Dragon 3000! Subsequent 4G basebands will be called Winged Dragon 4000!"
"Okay!"
Williams firmly noted it down.
Wang Yi patted Williams' shoulder:
"Work hard. I look forward to the successful R&D and mass production of Kunpeng 700! Keep up the good work."
"Rest assured, Chairman Wang, I will not fail you!"
Williams guaranteed.
The GPU for Kunpeng 700 uses an ARM public model, so there isn't much difficulty.
The CPU R&D is also progressing smoothly with him leading the Texas Instruments team.
The key lies in the R&D of the Winged Dragon 3000 baseband.
After all, Texas Instruments is not specialized in basebands, and the engineers from VIA previously only worked on 55nm.
Now directly moving to 28nm presents considerable difficulty.
But there's still more than half a year, so there's a lot of hope!
Two days later, the three chips successfully completed trial production!
Due to Wang Yi's massive scale, with a total of 600 million chips combined, Xingyi Semiconductor directly became a super-large client of TSMC.
Although they were all 40nm chips, in 2012, that was still quite good.
As for high-end 28nm, it had just started production, and TSMC's capacity was limited.
Morris Chang was incredibly excited about the sudden order for 600 million 40nm chips and personally called Wang Yi:
"Chairman Wang, congratulations! All three chips have successfully completed trial production. TSMC will prioritize your order. So, production will begin tomorrow, March 16th. We will give you the lowest possible price and sufficient capacity, please rest assured."
"No problem, thank you then."
Wang Yi spoke calmly.
But he secretly thought that the honeymoon with TSMC wouldn't last too long.
The Xingyi wafer fab will be completed by the end of the year and start production next year.
If it can achieve 40nm mass production, these three chips' subsequent super-large orders of several hundreds of millions annually will all be transferred to the Xingyi wafer fab for self-production and self-sale!
(End of chapter)