In China, Nvidia is very contradictory.
Not only have to bear all kinds of geopolitical pressure, but also not willing to give up the vested interests of the Chinese market.
Between several drags, both sides of the story need to “give up” something in order to reach an agreement.
As a result, many enterprises both love and hate Nvidia.
Nvidia is like a mountain that is difficult to climb, running across the road of the development of science and technology in China.
After a long time, he developed a tangled mood: he wanted to get rid of him and set up his own door, and he was inseparable from him and cared for him in every way.
Obviously, technology did not build the Tower of Babel.
And how long can the greatness of Invida continue to be “great” in the Chinese market? 20 days ago at noon local time on March 18th the 2024 Nvidia GTC conference opened with a full capacity in the stadium with more than 10,000 people.
Everyone is waiting for the man in leather to “detonate” the industry and even the world again.
As a result, early in the morning of the next day, the major domestic media came in bursts of exclamations, such as “the strongest AI nuclear bomb”, “Wang fried”, “the strongest in the world”, “crazy” ….
It must be admitted that Nvidia does deserve these compliments.
As a rare supergiant in the industry, there is no doubt about the technical strength of Nvidia.
But the particularly important point is, will these high-tech chips be handed over to Chinese companies? Nvidia shows off its skills and doesn’t give up the Chinese market.
As soon as Sora comes out, artificial intelligence is the hottest thing to say about human GG.
this year.
Even Musk threw aside Tesla, the “number one in the world”, and went off the line himself.
And the generative AI model has a terrible dependence on computing power, that is, GPU.
In a word, in the era of artificial intelligence, the key to development is still in the hands of Nvidia.
At this GTC conference, Nvidia launched the new Blackwell architecture B200, which is twice the size of Nvidia’s previous generation AI chip Hopper, and integrates 208 billion transistors, which is 2.
6 times that of the previous generation chip.
Oh, it’s not over.
Two B200 GPU and Grace CPU are integrated together, and through the interconnection technology of 900GB/s ‘s ultra-low power NVLink chips, they become GB200 super chips.
Then layer upon layer, two GB200 super chips are installed on the motherboard and become a Blackwell computing node.
Eighteen such computing nodes form a larger “virtual GPU”.
As Huang Renxun said, “We need a bigger GPU.
If not, we will combine more GPU into a larger virtual GPU.
” As a matter of fact, as long as we can achieve 1: 1 > 1, Nvidia’s show-off is a complete success.
But it is a pity that this kind of tough chip is prohibited from being exported to China.
As early as October 2022, the United States introduced a policy to restrict the export of AI chips to China.
In addition, the Bureau of Industry and Security (BIS) of the US Department of Commerce issued export controls on advanced semiconductors and computing equipment to China twice in a row.
In the early morning of March 30th, BIS once again issued new rules and measures to “impose additional export controls”, comprehensively restricting the sale of AI chips and semiconductor equipment such as Nvidia and AMD to China.
Again and again, under the layers of increase, the “red line” is getting tighter and tighter.
Reuters said the new rules are intended to make it more difficult for China to obtain US AI chips.
Although Nvidia has also supplied some special chips through “castration” before, the development of artificial intelligence in China has indeed been strangled one step at a time.
Perhaps for fear of forcing Chinese companies into a hurry and spoiling their friendship, Nvidia chose to find another way to deepen its ties with China’s major auto companies in the automotive business sector with low revenue.
There is no need to defend Nvidia, which is “sandwiched in the middle”.
After all, from the point of view of the intelligent driving chip alone, Nvidia has also achieved a “distant” lead.
It is a very ingenious move to reopen a road to win over Chinese car companies.
Focus, in fact, compared with the overall market, Nvidia’s revenue in the automotive industry is not much.
According to Nvidia’s latest financial report, the auto business earned $1.
1 billion in 2024, accounting for only 1.
8 per cent of total revenue.
But it must be admitted that less than 2% of revenue is firmly locked in the self-driving market in China and even around the world.
, no Orin, no smart driving.
Such a consensus has been formed in China’s auto market.
Brands of major car companies, such as Weilai, Xiaopeng, ideal and extreme Krypton, all need to rely on Nvidia’s chips to establish corresponding self-driving teams.
It is no exaggeration to say that the mainstream smart car chip today is the Nvidia Orin.
When the intelligent driving ecology is established, the silent costs such as hardware, software, algorithms and so on continue to accumulate, so it is not so easy to get away.
Nvidia, which knows how to do this, continues to expand its “market share” while courting Chinese car companies.
At the Nvidia GTC conference in 2024, Huang Renxun announced three cooperative car companies for Thor chips: BYD, Haopin and Xiaopeng.
In addition, polar krypton, ideal car and Wenyuan Zhixing have also made it clear that they will design and produce the next generation of products based on Thor chips.
Whether CUDA is easy to use or chip technology is far ahead Nvidia at this time has obviously become a “mountain” on China’s self-driving industry.
Just imagine, if the so-called ban is transferred to the car market, will those self-driving capabilities based on Nvidia chips continue to take the lead? Is there any Plan B that can be implemented as a precaution? The good news is that Chinese start-up chip companies such as Horizon Cambrian and Black Sesame have gradually integrated into the automobile market and become competitive to a certain extent.
However, it is still difficult to shake Nvidia’s position in the industry.
It may be a good way to touch Tesla to cross the river.
If there is no accident, some aspiring car companies still have to develop their own “FSD” on their own.
Weilai should be considered to be in the front.
On 2023 NIO DAY, Weilai released the first self-developed intelligent driving chip Shenji NX9031, which adopts 5nm gauge-level process.
Li Bin said that one Shenji NX9031 can match the performance of four Orin chips.
Geely also has a bottom.
At the second Tech Day of Yijiatong, Shen Ziyu and Wang Kai, CEO of Core engine Technology, jointly released the high-level 7nm intelligent driving chip-Eagle Zhikai AD1000, which is directly related to the standard Invida Orin.
In addition, Horizon Expedition 6 is ready to start, and Momenta’s self-developed smart driving chip is already under way.
On the string, many “spare tire” chips are in hand, and the situation does not seem to be as bad as expected.
However, it must be taken into account that only after the test of the market and time, these smart driving chips with Chinese genes can be regarded as a real “spare tire to become a regular tire.
” In this period of time of Chinese chip Don’t overextend, Nvidia is still the undisputed “overlord”.
In the end, looking back on these years, it seems that from the beginning, Nvidia has opened the “Eye of God” and walked on the right path.
Games, mining, meta-universe, AI, autopilot.
In all the hot fields, Nvidia always gets a piece of the pie and makes a lot of money.
In the face of the huge Chinese market, Nvidia can not give up easily, but it does not have many choices.
In the face of the “ban,” Huang Renxun could only bow his head.
Perhaps it is precisely because of this “dilemma” that Nvidia has begun to actively woo Chinese car companies and is no longer willing to be a single supplier of self-driving chips.
Not long ago, Liu Yilin, senior director of self-driving products of Xiaopeng car, was rumored to leave.
Liu Yilin may join Nvidia to join Wu Xinzhou, who previously left Xiaopeng Motor, according to people familiar with the matter.
Although Liu Yilin refuted the rumor afterwards, Nvidia’s willingness to drive on autopilot was well known from the moment Wu Xinzhou was introduced.
Can Nvidia achieve “greatness” in the Chinese market with its auto business? Will there be a seat for Nvidia on autopilot in China? The future is full of uncertainty, let’s see how Nvidia plays its cards.
, return to the first electric network home page >.
