After the layoffs, I became a tech godfather

Chapter 101 Inventory of AI Light Entrepreneurship Projects
"Xiaosha can't make the hardware?" Hao Cheng also sighed when he saw the relevant comments. It's not that he can't do it, but that even if he did, the current equipment can't manufacture it!

But then again, what does it mean to be involved in hardware?
Does it count if I find suppliers for everything and put together a mobile phone? Does it count if I am testing a children’s robot now?
What does it mean that these things, plus the blessing of Xiaosha, are far more competitive than similar products?
Hardware and software are integrated. Without a system, no matter how powerful the hardware is, it is just a pile of metal lumps, right?
Hao Cheng shook his head and stopped paying attention to these things. Instead, he searched for what projects Silicon Valley had been working on in the past year or two and which AI startups Wall Street investors were investing in.

CoreWeave? A large-scale GPU computing resource provider?
"Oh, this should be safe in the short term. In terms of hardware, it mainly requires computing power cards. Even here, a lot of computing power cards are consumed."

Nvidia has no rivals in the field of GPU or various inference and computing cards. AMD is a nobody, and Moore's Thread and Cambrian are just children's play. As for Ascend, its production capacity is a mystery.

In addition, NVIDIA started very early and formed a certain ecological moat, so it is really difficult for other companies to make a difference.

Ascend's ecosystem is slowly recovering, but the supply problem is too great.

One by one, they are either stuck in ecology or in production.

"Nvidia is really cheap. This company's cloud computing power is fast and it should last a long time."

Hao Cheng shook his head and continued reading: "Inflection AI, chatbot is the core product? That's dead!"

Hao Cheng sentenced this startup company, which raised $13 billion, to death. With Xiaosha around, what kind of chatbot can still have a market? Are you kidding!

"Oh, that's not right. Xiaosha seems to only support Chinese now." As soon as this thought came out, Hao Chengji shook his head again:
"Even if it's in Chinese, it still can't compare even if AI is used to translate it back and forth. This thing is dead."

AI translation is so simple that Xiaosha doesn’t even need to enhance his professional abilities to defeat all translators.

"MosaicML, a large language model." Hao Cheng seemed to be thinking. Xiaosha was so powerful, so it stands to reason that such a language model was the most dangerous.

But as far as he knows, all the big models including ChatGPT, Doubao, Qianwen, and Yuanbao are still undergoing training and iteration. Even each company is frantically adding cards in order to achieve qualitative change through quantitative change, slowly find the rules, and then peek into the secret of Xiaosha's successful training.

This is also one of the reasons why Nvidia's computing power cards have continued to rise recently - although there are some restrictions in the country, in a country the size of a city like Singapore, the sales of computing power cards continue to rise, and it goes without saying where these things are going.

As for stopping training of large models, there are also some.

According to what Hao Cheng learned, one well-known company is Baidu, which has stopped iterating its Wen Xiaoyan app.

But this is not surprising. It is not the first time that Baidu gets up early and arrives late. It has abandoned countless projects, and this one is not a big deal.

"OmniML, a company acquired by Nvidia, whose main product is Omnimizer, can compress the size of machine learning models so that large models can run on smaller edge devices, such as mobile phones, drones, cars and other terminal devices?"

Pure end-to-end testing? Hao Cheng pondered, Xiaosha runs on the server and cannot be used offline.

The advantage of end testing is that it can run offline, another advantage is that data confidentiality is higher, and the third advantage is that it is more real-time.

As for data confidentiality, it goes without saying that Xiaosha has gained enough trust, so it is not a problem at all. As for real-time performance, it is not very meaningful unless it is a scenario with very high requirements for decision latency, such as intelligent driving, which requires cloud training and terminal operation.

As for the rest, they were killed instantly by Xiaosha. The future of this thing is also unclear.

……

Hao Cheng looked through a number of AI startup projects invested by Silicon Valley and Wall Street, and finally came to the conclusion that not many can survive. In this round, 99% of them will be killed by these domestic AI young entrepreneurs!

However, what a startup called "Adept" is doing caught Hao Cheng's attention - what they want to do is to [build models that can take action in the digital world].

After training, this model can use all existing software tools, APIs and apps, and is a way to achieve general intelligence.

"Isn't this what Cao Song is doing now?" Hao Cheng took a closer look and found that it was also a little different: "This model is to train AI to use the current software, API and APP.

"But our plan is to provide a foundation to guide developers to develop applications suitable for both humans and AI. This concept is exactly the opposite."

Hao Cheng could understand this, because the original intention of the current software, whether it is computer software, mobile phone software, etc., is to be used by people. Besides, before Xiaosha appeared, there was no AI that could use APP so powerfully and freely.

But now, Xiaosha has appeared, and the use of AI APP will become the norm.

What Adept does is actually something Xiaosha can do, but it just simulates human operation, directly "touching" the screen to use the APP, which is a level of conversion and a little inefficient. At most, when there is no one in front of the computer, the screen is still moving and working, which looks a little scary, that's all.

"But this entrepreneurial team is pretty good as they have this kind of thinking!" Hao Cheng finally gave this idea, but he was also not optimistic about the outcome - because they met Xiaosha - they wanted to do the same thing as Xiaosha, but were obviously far behind.

……

"Mr. Hao, Zhu Yue, Apple's vice president and managing director of Greater China, has sent a business visit letter. How should we respond?" When Hao Cheng was evaluating the future of these projects one by one, Li Qingbo brought this news.

"Oh? Such a slow reaction?" It's not surprising that they came, let alone a vice president. Even if Cook came, Hao Chengdu wouldn't find it strange. It would be strange if they didn't come - Apple's sales in China have been halved again in the past two months.

Nonsense, such a powerful Xiaosha has no plan to develop an IOS version. Normal consumers will no longer choose such a product.

If Apple didn't have any special plans or intentions, it would be a fatal mistake to think of visiting at this time.

"Apple has no research or accumulation in AI. The news I got was that Cook visited Brockman, who told it that OpenAI will solve the problem by March next year at the latest."

Li Qingbo reported that he had found relevant information after receiving a business visit from Apple: "Solving the problem means getting an AI with capabilities similar to Xiaosha. Cook believed it."

"Oh? Why did you change your mind now? It's only December!" Hao Cheng asked curiously.

"As you know, Huawei has been planning overseas press conferences for some time, including Dubai, Malaysia, Paris and even London," Li Qingbo said. "Apple can tolerate Dubai and Malaysia, but they find Paris and London fatal."

Hao Cheng is of course aware of Huawei's actions. The overseas version of Xiaosha still needs the interface upgrade cooperation here: "They can really tolerate it. China is their most important market."

"After all, it's an American company, and we have to look at the current situation of both sides. If Xiaosha is such a powerful thing, if it is accepted in China, will it be accepted abroad? If not, it will be a problem, and if it is accepted, it will be even more of a problem.

"If Apple loses the Chinese market, it will be amputation at best. Even if all four limbs are cut off, it can still survive. But if it loses the European and American markets, it will be beheaded."

Li Qingbo analyzed: "But if Huawei brings Xiaosha to Europe, it will be completely different. For them, it will be a knife if they stick their heads out or hide their heads. One good thing is that America has not imposed any restrictions on Xiaosha and has never mentioned it.

"Of course, it may also be because we were founded only a short time ago, less than two months at most."

Although Xiaosha caused a big fuss, it was only established less than two months ago, and many decisions may not be made that quickly.

"Is this better for us or for Apple?" Hao Cheng shook his head speechlessly: "Forget it, we can still meet and talk. We treat all companies equally and never discriminate."

Li Qingbo was also stunned for a moment, and realized that Hao Cheng really didn't care about the so-called Apple, or even the so-called America.

But if you think about it, there is really no need to care about what they think or do.

The domestic market is the largest single mainstream market, and there is enough room for growth. Even if Xiaosha never goes abroad and stays in China, its business model can still be run, incubated, and grown healthily.

……

The next day.

"Hello, Ms. Zhu." Zhu Yue was very anxious. After all, the area she was in charge of was Greater China. The so-called Greater China region consisted of a small group of Hong Kong, Macao and Taiwan, but the most critical market was mainland China.

The market in mainland China has fallen to a point where it is no longer viable this month. The market share is now the same as Samsung's, and the ranking has entered the other sequence, which has never happened before.

She even chose to give up treatment before the headquarters made a decision - she herself was using Xiaosha and had become addicted to it, so how could she convince consumers to buy Apple?

If you don't join, this will be a dead end, unbreakable.

Even if she chose to join, Zhu Yue had no idea how much hope there was for a reversal, and the conditions given by the headquarters - she herself thought Cook might be crazy.

But, she still has to fight for it, this is her last chance.

　Happy New Year, book friends!
　　Asking for a monthly ticket!
　　
　
(End of this chapter)