262: Chapter 262 The Silicon Crown: AI Chips

8:00 PM, while industry giants were plunged into panic, the Chinese internet was jubilant.

Zhihu hot list #1: 'How to evaluate Xinghuo Chip securing $17.7 billion in orders in one day?'

Top-voted answer (anonymous, ID 'Silicon-Based Ascetic', suspected industry bigwig):

'Let's do some simple math:

The global annual demand for 28nm chips is about 10 million units. Xinghuo sold 3.6 million units today, 90% of which were sold to domestic companies. What does this mean?

Domestic electronics manufacturing's low-to-mid-range chip production is basically localized.

TSMC and UMC's 28nm production lines might have to shut down half their capacity.

The most terrifying part is the price: $2,800/unit is the industry cost line. If Xinghuo sells at this price and still makes a profit, it means the cost is likely below $2,200.

16nm/14nm orders reached 1.57 million units. This part of the market was originally the cake for Qualcomm, MediaTek, and Apple. Xinghuo used FD-SOI technology to achieve TSMC's 10nm performance at the 16nm node, with only 60% of the cost. This means:

Phone manufacturers have no reason not to use Xinghuo Chip.

Qualcomm and MediaTek's mid-to-high-end market is in jeopardy.

Apple might use Xinghuo Chip in their entry-level product lines.

7nm mass production starts Q1 next year, 5nm laboratory tape-out successful. This means:

In terms of process, Xinghuo is already on par with, or even slightly ahead of, TSMC.

Intel is still struggling at 10nm, and Samsung's 7nm yield is dismal.

In the next two years, the global high-end chip market might form a duopoly of 'TSMC vs. Xinghuo'.

Summary: Today might be the 'Independence Day' for China's semiconductor industry. From now on, we have our own high-end chip manufacturing capabilities and will no longer be held back by others.'

Likes: 285,000, Comments: 12,000.

Chinese microblogging platform hot search top 3:

#Xinghuo Chip doubles phone battery life# (Reads: 420 million)

#7nm chip coming next year# (Reads: 380 million)

#TSMC stock price plummets# (Reads: 350 million)

Popular posts on Chinese microblogging platform:

Digital Chat Station: 'Real-world test data is out! Testing 'Genshin Impact' with Xinghuo 16nm chip, the temperature is 9 degrees lower than Snapdragon 845, the frame rate is more stable, and power consumption is 35% lower. What does this mean? Next year, phone battery life might actually double. Huawei, Xiaomi, and OV are already fighting to get them; it's said orders are backed up until the end of next year. Qualcomm and MediaTek, you're in danger!'

Shenzhen Huaqiangbei backpacker: 'I received dozens of calls today asking if I have channels for Xinghuo 28nm chips. I want to have them too, damn it! The online system sold out in 60 seconds, and the scalper price is already $3,200. Xinghuo is trying to force the copycat factories to their deaths—but then again, if they use Xinghuo Chip, copycat phones could also have flagship-level battery life. This business... seems even bigger?'

Hupu Street legendary post:

Title: 'It used to be 'let's run a benchmark if you're not convinced', now it's 'let's compare power consumption if you're not convinced'. Xinghuo is trying to revolutionize Qualcomm and TSMC!'

Top replies:

'The JRs might not understand chips, but they understand phone battery life. If phone battery life can really double next year, I don't care about Qualcomm or MediaTek; I'll use whoever saves power.'

'Xinghuo's move: first establish authority with military hardware, then harvest with chips. A cannon in the left hand, a chip in the right; both cultured and martial.'

'It's said Xinghuo is also working on AI chips, automotive chips, satellite chips... This is trying to dominate the world.'

'Su Chen: I'm not targeting anyone specifically; I'm just saying everyone here is trash.'

8:30 PM, Xingchen Headquarters

Su Chen received an encrypted call from the leader of the MIIT.

'Comrade Su Chen, today's data... is very encouraging.' The leader's voice carried a smile. 'The ministry's preliminary estimate is that Xinghuo Chip can save the country at least $20 billion in foreign exchange expenditures this year and drive the output value of related industries to exceed 500 billion yuan.'

'This is the team's hard work,' Su Chen said.

'But I must also warn you, the international backlash will be very intense.' The leader's tone turned serious. 'TSMC has already filed an anti-dumping investigation application with the WTO, and the US Department of Commerce may be brewing new sanctions to restrict the export of various chip-related manufacturing equipment to China.'

'We are ready,' Su Chen replied calmly. 'We hold over 50% of the patents for FD-SOI technology globally. Regarding lithography machines, our 16nm lithography machines are already in mass production, and 14nm lithography machines are being tested. Core materials and equipment are basically produced by our own company now.'

'Good! It's good to have that confidence.' The leader praised. 'Additionally, the technical authorization for SMIC... the ministry has studied it and can support it. But you must set up a firewall to ensure core technologies are not leaked. You decide on the specific plan.'

'Understood.'

Hanging up the phone, Su Chen looked at his team: 'What do you think about SMIC's proposal?'

Li Wanqing spoke first: 'The risk of technical authorization is too high. FD-SOI is our core advantage; once leaked, TSMC and Samsung can catch up very quickly.'

'But SMIC's 28nm capacity is very tempting,' Gu Xinghe said. '500,000 units/month, equivalent to 125% of our existing capacity. If we can use it to produce low-end chips, we can concentrate all our high-end capacity on 14nm/16nm.'

Su Chen thought for a moment: 'We can authorize, but it must be tiered. Give SMIC the 28nm FD-SOI process package to help them improve yields and reduce costs. But do not authorize 16nm and above processes. And... we need cross-shareholding.'

'Cross-shareholding?'

'Xinghuo Chip will take a stake in SMIC, and SMIC will take a stake in Xinghuo Chip's packaging and testing business,' Su Chen said. 'Only by binding interests can we ensure technology is not leaked.'

'Will SMIC agree?'

'They have no choice.' Su Chen looked out the window. 'Either accept our conditions, get the 28nm process, and survive. Or continue to struggle on 14nm FinFET and eventually be eliminated by the market.'

This is the confidence of a technology leader.

9:00 PM, Su Chen stood in front of the floor-to-ceiling window on the top floor of the headquarters, looking at the city's myriad lights.

Downstairs, Xinghuo Chip's employees were celebrating. Today's $17.7 billion in orders meant everyone would receive generous bonuses.

But Su Chen was clear-headed.

'chips are just the foundation,' he said to Li Wanqing, Gu Xinghe, and Lin Wan beside him. 'The real future is in the silicon-based brain.'

He brought up the progress report of the Xinghuo AI Laboratory:

AI training chip 'Sui Ren 1' based on 14nm process, computing power reaches 1.5 times that of Nvidia V100, power consumption is only 70%.

7nm AI chip 'Sui Ren 2' is being designed, with a target of 5 times higher computing power.

Automotive-grade AI chip 'Xuan Yuan' has passed tests by Hongqi Automobile and will be used for L4 autonomous driving.

Satellite AI chip 'Bei Dou' is under development, used for autonomous decision-making of Starlink satellites.

'The chip is the torso, and AI is the brain,' Su Chen said. 'What we need to do is not just sell chips, but provide a total 'chip + AI' solution. From mobile phones to cars, from factories to satellites, all devices that need intelligence will use our chips and AI.'

He looked at Li Wanqing: 'Starting tomorrow, launch the 'Sui Ren Project'. Goal: within three years, make Xinghuo AI chips occupy a 30% share in the global data center market.'

'Three years? Nvidia currently holds 90%.'

'That's why we have to do it.' Su Chen's eyes were sharp. 'Just like today's chip launch—break the monopoly with a technological generation gap.'

Outside the window, the night sky of Shenzhen was illuminated by countless electronic screens.

In this city, tonight, bosses of thousands of electronic enterprises were recalculating costs, re-planning products, and rethinking the future.

And all of this was because one company had rewritten the power map of the global chip industry in just one day.

The silicon-based crown had been donned.

But the crowned one knew this was just the beginning.

The true throne was in AI, in intelligence, in the digital future connecting all things.

And that future was being ignited by Su Chen.