The comments section has readers asking about investment issues in the AI field. Coincidentally, I was chatting with friends these past few days about this topic, especially some companies in the large model and application sectors.

If we do not consider any external factors of the AI field (such as storage, electric power, photovoltaics, non-ferrous metals, etc.), and focus solely on industries and companies directly related to AI, my friends and I essentially share a common viewpoint:

This field currently seems too uncertain; not to mention that some existing companies could be disrupted at any time, even newly emerging companies that seem to be very promising could likely be eliminated quickly in the near future.

Because the technology in this field is developing too rapidly and the competition is too fierce, there are truly very few companies that have high barriers to entry and have formed solid business models.

Among the new companies in the application sector, I currently do not see any that have established a visible moat. Only companies in the underlying architecture category have somewhat developed their own moat.

Among the underlying architecture companies, I am mainly focused on two sub-sectors: one is large model chips, and the other is large models.

Regarding large model chips, at present, I believe the only company that has established a certain moat is NVIDIA.

NVIDIA's biggest moat is currently in training chips, while the application growth in the AI field is shifting faster towards inference. In the inference chip domain, while NVIDIA does have some advantages at present, those advantages do not seem obvious, and many competitors are eyeing the market.

However, among the many competitors in the inference chip field, even though some are quite powerful (such as Google and its TPU), I feel that they lack a unique quality. In other words, I don't see them as potential leaders in the inference chip sector in the future.

So in the inference field, I feel that the future leader will either still be NVIDIA, or be a latecomer striving quietly in some corner of the world, or may not have even been born yet.

Regarding large models, I have been closely watching two companies: one is OpenAI, and the other is Anthropic.

The competition between these two companies is very fierce, and both are vying for an IPO, reportedly targeting an IPO this year.

Among these two companies, I am more optimistic about Anthropic.

Initially, OpenAI had a tremendous first-mover advantage in terms of technology, layout, and business, but it seems now that for various reasons, OpenAI has noticeably stalled.

In application areas, for enterprise applications (to B), Anthropic clearly has the advantage; for user applications (to C), although OpenAI is currently in a leading position, Anthropic is rapidly closing the gap with significant momentum.

Business layout and development can actually be improved and caught up with, but some defects are quite difficult to make up for. Such defects often lower the ceiling for a company’s development, leading to long-term issues in the company's growth.

This defect is the company's culture.

In the past year or two, especially after reading the book by Duan Yongping, I have become particularly concerned about a company's corporate culture. For a startup, its corporate culture is essentially the founder's temperament and values.

I apply this standard to evaluate projects in the cryptocurrency ecosystem (such as Ethereum), as well as to OpenAI and Anthropic.

In my earlier articles, I compared the founders of these two companies. At that time, my impression of OpenAI's founder (Altman) was quite distinct: I felt that his values were problematic.

When I compared Altman and Dalio (the founder of Anthropic), I used their actions of banning users from mainland China as an example.

Although both advocated and implemented bans on mainland China, their motivations were completely different:

The former was entirely profit-driven, aiming to dominate the ecosystem; while the latter was based on broader and more public safety considerations.

Here, I do not comment on whether Dalio's views regarding mainland China are correct, I merely want to express that his starting point is not solely based on narrow self-interest and commercial considerations.

Later, a more well-known incident occurred that further highlighted the stark differences in their fundamental colors.

Due to the exposure of the Venezuela incident, Anthropic firmly opposed the US Department of Defense, not allowing the department to use its AI tools for military purposes.

My first reaction upon seeing this news was:

Dalio essentially adhered to the "Three Laws of Robotics" defined by his predecessor Asimov, and this adherence to and persistence in values underscored the humanistic foundation and value commitment of a tech worker.

In this incident, Altman, on the other hand, played a rather disreputable role, where all actions ultimately pointed towards pure commercial interests without any adherence to a bottom line.

I do not deny that in the face of politics, entrepreneurs sometimes have to bow down and compromise, but people can see when it is necessary to bow, when it is not, and when it is merely an opportunistic approach.

In all these incidents, Altman increasingly resembles a politician and a pure profit-chaser, rather than an entrepreneur who creates great companies.

If an entrepreneur is a politician, purely profit-oriented, can users really trust the products that such a company creates? Can they even trust to use them long-term?

Take the US Department of Defense as an example; if it becomes deeply tied to such a company, isn't there a concern that one day its systems could be completely hijacked, leading the Department of Defense to be manipulated by OpenAI? Isn't there a worry that in such a situation, that kind of person might do something without a bottom line?

This is akin to the concerns some predecessors had: it is like handing nuclear weapons to a gorilla, which would be a catastrophe for humanity.

So overall, in the current AI field, the two companies I focus on that are forming a certain monopoly advantage are NVIDIA and Anthropic.

For these two companies, if the price is right, I would buy some; if the price is not right, I will continue to observe.

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。