Shenhao started as a military training anchor

Silicon Valley: "China's AI industry has encountered a bottleneck, the capital market has withdrawn completely, the AI ​​competition has widened the gap, and the United States has been leading for more than ten years..."

Wall Street: "Chinese technology stocks are performing poorly, capital is cashing out at high levels, and investors are withdrawing their capital. It is expected that the industry will not recover in the next two to three years..."

London: "AI competition enters finals, OpenAI is far ahead, out of reach..."

"..."

All the major financial streets and centers around the world, as well as economic and technology newspapers, all popped up overnight to analyze the AI ​​market.

Devalue the domestic AI industry.

To be honest, the opponents' combined attack has indeed put the domestic AI giants in a difficult situation and they are unable to recover for the time being.

But these reports were too exaggerated, and Holly saw it as another war of public opinion.

Continue to put pressure on the domestic AI industry.

They know that they cannot use market rules to deal with Eastern political capital, and it is impossible to completely defeat their opponents or an AI giant.

Therefore, after the combination of attacks, negative public opinions were created one after another.

The purpose is to prevent domestic AI from reviving.

Even with the support of the banking consortium of the Big Four, continued losses will bring tremendous pressure to the company, the most obvious of which is large-scale layoffs.

I have to admit that the opponent's tactics are very sophisticated and every punch hits the target.

Here, Holly used the two capitals he had to buy nearly 3000 billion worth of technology stocks.

The loss caused by a slight improper operation is unimaginable.

He is under a lot of pressure now.

There is a glimmer of hope in my heart, which lies in the Tai Chi 99 chip developed by Everest Technology and the Ascend 910B chip that Huawei is about to launch.

The computing power of these two chips is comparable to that of Nvidia's most powerful chip of the previous generation, the H100.

In terms of the computing power of single-subject chips, Taiji and Shengteng have caught up with NVIDIA.

Although they cannot be compared with the G200 stacked chip developed under NVIDIA's new architecture, in terms of training large models, both chips can support the training of trillion-level large models.

The actual combat capabilities of these two chips have not yet been confirmed.

Even if the actual combat capabilities are excellent, they cannot escape the disadvantages in software.

There is no optimized and unified ecosystem.

Various domestic trillion-level models cannot be adapted to the two chips.

Unless you change the architecture.

Changing the architecture means completely abandoning Nvidia chips.

To be honest, it's quite risky.

After all, in the field of chip computing power, Nvidia is still invincible and has been evolving.

If domestic trillion-level large models completely abandon Nvidia chips and choose Taiji and Shengteng, whether the computing power evolution speed of the two chips can keep up with the training speed of the large models is a hidden bomb.

If they fail to keep up, the gap in artificial intelligence between the two countries will widen further.

There may even be a real bottleneck that stops you from moving forward.

The companies suffered huge losses and the industry regressed for several years. Few companies could afford the consequences of this gamble.

The computing power is not enough to roll out the financial snowball, and the model is also a bit weak.

How to find a gimmick to get the financial snowball rolling again and boost technology stocks again is a headache for Holly.

This is one of the few headaches he has experienced since he entered the technology industry.

After all, he holds 300 billion worth of stocks and is visibly losing money every day.

It's impossible not to be anxious.

While Holly was still thinking about market issues, two calls came in succession.

One called from the Ministry of Industry and Information Technology, and the other called from the Ministry of Science and Technology.

The two phone calls conveyed similar messages, that the country will provide financial and material support to get AI capital rolling as soon as possible.

As long as capital starts to flow, A-shares will stabilize.

Then slowly solve the industry’s problems.

I have to say that the leaders are very good at grasping the main contradictions and stabilizing the stock market first.

The bull market that had lasted for more than half a year would have disastrous consequences if it suddenly collapsed. Holly smiled.

Brother Huo dares to buy stocks worth 3000 billion yuan. Is he short of money?
What is lacking is not money, but the gimmick to get the capital rolling.

Faced with the dual pressures of the market and politics, Hawley can only say that he can still hold on.

In the past, the game with opponents was single-field and one-sided, and Holly frequently slapped his opponents in the face.

Now it is a competition between financial capital, system, industrial structure... and even national strength.

To be honest, the opponent is very strong.

When he was at a loss, his old friend Zhou Hongyi called him and communicated with him.

The two had previously shared the same views on the AI ​​application market.

That’s why there is a combination of Tai Chi 2.0 chip and billion-level small models.

This has proven to be a virtuous path.

Make AI an enabling tool that most companies can afford.

But as an industry leader, Big Model must also compete with its rivals.

An industry that is regarded as an industrial revolution will surely be able to leverage the huge capital market.

Affects the entire country's financial industry.

The opponent is now targeting this point.

In his communication with Zhou Hongyi, Holly compared the advantages and disadvantages of several large domestic models and gained some inspiration.

Is it possible to create a super large model that combines all the advantages of domestic trillion-level large models?
This will not only allow us to catch up with and match Chatgpt in performance, but also adjust the unified architecture and complete the ecosystem of domestic chips.

That is, the chip software ecosystem of Tai Chi 99 and Ascend 910B, which can match the computing power of NVIDIA H100.

The computing power problem was solved in hardware, and the chip ecology problem was solved in software.

If the concept is successful, domestic AI will take a big step forward, or even make a leap forward.

Zhou Hongyi was very shocked after hearing this.

While I admire Holly's great ideas, I also have some concerns.

The first thing is to bring together these domestically produced trillion-level models.

If it is just a structural change and mutual adaptation, it is relatively simple and will only require some effort from the technical staff.

The most difficult thing is that each model has its own profit model.

If you want to come together, you must be prepared to lose profits and sacrifice interests.

Tencent, Alibaba, Huawei, iFlytek... these companies may not be willing?
Their leadership may not be willing either.

While Zhou Hongyi raised his concerns, he also expressed his stance, "Mr. Huo, regardless of our personal relationship or the current difficulties, I am willing to cooperate with you to try to break through."

At critical moments, Lao Zhou is still reliable.

Holly also smiled and said, "Lao Zhou, if you give me this face, Ma Huateng and Cai Chongxin...wouldn't they be willing to give me this face?
There are only two paths now. Either continue to compete internally and see who has a more stable foundation?

Or let's build the house together first.

I'll give them a call myself."

Zhou Hongyi flattered: "In the entire entrepreneurial circle, only you, the leader, have such appeal.

But you also have to be mentally prepared.

After all, each party has its own huge interests at stake.”

Holly nodded: "I know what I'm doing."

After hanging up the phone with Zhou Hongyi, Holly made a few more calls.

Alibaba CEO Joseph Tsai, Tencent Chairman Ma Huateng, Huawei Rotating Chairman Ren Zhengfei, iFlytek CEO Liu Qingfeng, as well as the director of Tsinghua AI Zhipu Research Institute and the academician of the Chinese Academy of Engineering of the Institute of Artificial Intelligence of the Chinese Academy of Sciences.

These calls bring together the best of domestic trillion-dollar models.

After Holly's analysis, everyone expressed their opinions and agreed to sit down and talk face to face.

Holly personally prepared for this summit meeting, which was held in Shanghai.

Domestic trillion-level large models, Tongyi Qianwen, Hunyuan, Pangu, Zhufeng, Zhineng, Zhipu Qingyan, iFlytek Spark... gathered in Shanghai for the first time.

(End of this chapter)