My Fintech Empire

On Tuesday, April 4, Nvidia's stock price continued to maintain a strong bullish surge at the opening today after yesterday's surge. The intraday increase once reached +26%, and the market value soared to US$150 billion.

But not long after the company's stock price exceeded $150 billion, an accident happened.

A congressman from America's side jumped out and called for the federal government to curb the expansion of the computing power of the Eastern countries on the other side of the ocean. He even bluntly said that relevant bans and regulations should be introduced to prohibit North American chip manufacturers from supplying them.

And not long after, America's new talker commented on his personal social media account, saying that this was a great suggestion and well worth considering.

As soon as this news came out, the stock prices of companies such as Nvidia and AMD plunged sharply, especially Nvidia's stock price, which turned green and fell by 16 percentage points.

The heads of these international chip manufacturers were immediately anxious, and the shareholders behind them were also anxious. Wouldn't this be ruining people's financial paths? They have already announced a seven trillion plan, with a market size exceeding one trillion US dollars. If this business is done No matter how big the loss is, everyone has to be anxious.

The news spread to the country, and many media cited the report, which also attracted a lot of attention.

At this moment, Fang Hong was also discussing this matter via video chat with Chen Yu. In the study room on the second floor, Fang Hong said with a positive attitude: "...even if Lao Mei doesn't do this, we must take the initiative to discuss this matter with Chen Yu." NVIDIA and other international manufacturers are ready to cut the gap. This time we spend 7 trillion to build a unified national market for storage and computing services. This is an excellent opportunity to reshuffle and set new standards. "

Chen Yu nodded: "It makes sense. Seven trillion is an unparalleled right to speak."

Fang Hong continued: "It is no longer difficult to catch up with the leading level of international manufacturers such as NVIDIA at the hardware level. The gap with their most advanced consumer-grade chips has been shortened to only one generation. The real difficulty is the set of chips developed by NVIDIA. What is C, a parallel computing platform and programming model that is strongly related to its GPU?

Chen Yu answered: "Cuda."

Fang Hong: "Yes, that's it."

The Cuda ecosystem is similar to the relationship between Intel chips and Windows systems in the Wintel ecosystem. It includes many levels of resources, such as compilers, debuggers, rich library functions and various software tools.

Chen Yu in the video said: "It has been ten years since Cuda was launched. Many algorithm engineers are already familiar with and accustomed to using the tool library and programming language provided by Cuda. Their algorithms, models and applications are basically developed based on Cuda. Even many engineers learn Cuda programming from the beginning. If they switch to a new AI chip and cannot use Cuda, many people may feel at a loss if they need to rewrite the codes of all algorithms, models and applications.”

"It's like suddenly telling users who have been using Windows systems for many years that they can no longer use Windows and switch to Linux. It is necessary to rewrite all Windows-based algorithms and software. This is an almost impossible task for many people. Changing such a fundamental Things cannot be completed in a short time, they require long-term accumulation and adaptation. Many algorithm engineers have been using Cuda for seven, eight or even ten years.”

Fang Hong said firmly: "When Duan continues to suffer from chaos, it is urgent to rebuild a Cuda software ecosystem similar to Nvidia. The longer it takes, the greater the sunk cost will be, and the greater the user's dependence on it, the more disadvantageous it will be to us." With the launch of this 7 trillion plan, we have the market size advantage and we will set the standards. Now is the best time to build the MHAC ecosystem. Doesn’t Cuda boast of its excellent compatibility? Then let it be compatible with us, not us. Compatible with it.”

The so-called MHAC ecosystem is the abbreviation of "Matrix Heterogeneous Computing Architecture". It is an open ecosystem and other types of GPUs can be connected to it. Currently, there are not many developers in the MHAC ecosystem, but it is the best time to expand the scale of the ecosystem.

The market size is the right to speak, and the Seven Trillion Plan is the right to speak, including the current second-generation large models whose technology is open source, and it is inevitable that the next three generations of large models will also be open source.

It can be said that we need the right to speak in the market, and now we have it, and we need the right to speak in technology, and we now have it. Achieving strong binding with algorithm customers is not a fantasy. Many algorithm engineers are used to CUDA and can also get used to the MHAC ecosystem. Their companies will urge them, unless their bosses are indifferent to the 7 trillion blue ocean market and no longer want to make money here.

This wave of operations by Fang Hong has both technical elements and a one-stop effort. Choosing to embrace MHAC will also make it easier to obtain financing support from the capital market, and Fang Hong will also give these companies a higher price-to-earnings ratio.

Finally, Fang Hong said: "In the second half of the year, you will also open source the third-generation large model, so that it can be strongly bound to MHAC. The capital market will also cooperate and send assists. Those who follow us will be able to integrate easily." To the endowment.”

Chen Yu nodded and said, "No problem."

Fang Hong doesn’t care that some betting companies can easily get financing from him. It doesn’t matter. It doesn’t hurt to let some betting companies collect some wool. What he wants is to create a general trend.

What's important is that even if you are used to Windows, you have to switch to Linux. If you don't make such a change, the boss of the company you work for will have to fire you. If you change, you will remain the same.

Because if your boss doesn't do this, the investors will withdraw their capital. He can raise funds from the market because the investors have more or less support from the stars behind them. The biggest logic of investing in you and giving your company a higher price-to-earnings ratio than its peers is based on the reason that you want to use MHAC. Therefore, the boss of the company must want engineers from the MHAC ecosystem. There are not enough market-related human resources. In order to grab MHAC engineers, the bosses will give higher salaries, and higher salaries will attract more engineers to join this camp.

This is the general trend. Fang Hong has made a great determination and is willing to invest heavily. If he doesn't invest heavily, he can't create this general trend.

After the video call with Chen Yu ended, Fang Hong also looked at the fermenting news on the Internet. America on the other side of the ocean said that it would curb the expansion of computing power of the Eastern power and asked companies such as Nvidia to provide computing chips, etc.

This news spread back to the mainland and caused a lot of panic in the circle, causing the existing graphics cards and other hardware equipment on the market to rise in price again.

After a while, Fang Hong closed the relevant information page and logged into his personal Weibo account, quickly editing a paragraph of content and updating it:

[I remember when the cooperation between Xingyu Technology and Qualcomm was terminated, I had a conversation with Qin Feng. He said self-deprecatingly at the time: There are all kinds of medicines in the world, but there is no regret medicine. If there is anything to regret, it is that Xingyu Technology did not engage in semiconductor manufacturing because it believed in global division of labor at that time.

As a company, it is impossible to do everything, and now it has been slapped in the face by reality. You can't make high-end chips yourself. Others will block you with a card, resulting in the mobile phone business that is far ahead of it almost standing still and losing a large part of the market.

If we had gone all-in on semiconductor manufacturing in the early years, perhaps there would not be such a bad situation today.

The road ahead is long and hard, but now we are starting from the beginning.

Matrix Quantum has learned the lessons of Xingyu Technology and has been unwaveringly determined to be independent in science and technology from the beginning. Qunxing has also been crazy about making up lessons in the entire semiconductor industry chain. Now, our opponents have not only failed to defeat us, but have made us stronger. Companies such as Xinwei Semiconductor have made gratifying progress.

Over the years, Matrix Quantum has made continuous major technological breakthroughs in both software and hardware. Although there is still a gap with the world's most advanced products, this gap has been shortened from the huge gap of four or five generations in the past to within one generation. I believe it will not take long to achieve parity with the world's advanced level.

By the way, Matrix Quantum has decided to expand its production line and increase its production capacity by more than ten times, and fully provide hardware supply guarantee for the construction of the integrated storage and computing center infrastructure project. The initial plan is to provide 70% of the supply demand, and the remaining 30% gap will be reserved for other domestic manufacturers to choose the best. 】

……