This is the 1280th original article from 'New Energy Frontier'. Click above on 'New Energy Frontier' to follow and 'star' this account. The article solely reflects the thoughts of 'New Energy Frontier' and does not constitute investment advice. The author does not run groups, recommend stocks for a fee, or manage client funds.

--------

On New Year's Eve, after sending his blessings, Leo posed a question: How should we perceive AI opportunities in the Year of the Horse? He suggested that the answer to this question would significantly shape investment returns for the Year of the Horse, much like how involvement in the AI industry chain over the past two years has largely determined investment outcomes.

So, do you already have your own answer?

For Leo, the answer has long been clear, and after watching the Spring Festival Gala, it became even more resolute – in the Year of the Horse, we must persist in focusing on AI!

In fact, there is now virtually no middle ground when it comes to AI; one is either bearish, viewing AI as one of the biggest bubbles in human history, or bullish, seeing it as a disruptive technology that will profoundly impact the global economy and even the global landscape.

Ultimately, being bullish or bearish hinges on one's understanding of the technological essence of AI.

01 The Technological Essence of AI

At the reader meetup hosted by Value Office in Chongqing at the beginning of the year, Leo shared "Sailing Towards Intelligence: The 2026 AI Panoramic Map," analyzing the essence of AI technology. If we abstract it infinitely, since the Industrial Revolution, humanity has been 'creating humans.' The mechanical movements sparked by the Industrial Revolution created the 'body' of this 'human,' while subsequent electrical and information revolutions created its 'blood.' AI technology completes the most crucial part – the 'brain.' With the maturation of AI technology, this 'human' will finally take shape, enabling humanity to rapidly create 'virtual beings' with infinite capabilities, far surpassing those of most ordinary people. It's akin to how the Monkey King can create countless small monkeys by plucking a hair, but even more powerful. The small monkeys created by the Monkey King are not as capable as he is, whereas the 'virtual beings' enhanced by AI technology will surpass humans in every aspect.

If Leo, who is neither an AI practitioner nor an authority, defines the essence of AI technology based on conclusions drawn from the history of human economic activity, it may not be entirely convincing. However, a recent article by an industry leader serves as the best and most authoritative endorsement of Leo's definition of AI's technological essence.

Dario Amodei, a senior expert in the field of artificial intelligence, has conducted AI-related research at Baidu, Google, and OpenAI. He later co-founded Anthropic and is now its co-founder and CEO. Anthropic is the AI company behind the Claude tool, one of the hottest startups, having raised over $30 billion in funding and valued at nearly $400 billion.

At the end of last month, Dario Amodei published a 20,000-word essay titled "The Adolescence of Technology." In the article, he presented a profound thought experiment: Imagine that in 2027, a new country with 50 million citizens suddenly appears overnight, each citizen possessing intelligence surpassing any Nobel laureate in history, thinking 10 to 100 times faster than humans, never sleeping, capable of manipulating the internet, controlling robots, and guiding experiments. How would you, as a national security advisor, view this?

In the article, Amodei answered his own question: This would be the 'most severe national security threat in a century,' and 'we are building that country ourselves.'

See? Isn't this exactly the same as Leo's shared understanding of the essence of AI technology? The essence of AI technology is humanity's first true realization of 'asexual reproduction,' with offspring being 'virtual beings' whose abilities far surpass those of most humans or even nearly all humans who have ever existed.

However, what Leo doesn't quite understand about Dario Amodei's thinking is why this new nation would only have 50 million citizens. From a technological perspective, the number should be countless, infinitely large – that's the true horror of AI technology. These 'new citizens' don't need to endure nine months of pregnancy like humans, be raised through hardships, or study for over a decade. As long as there is sufficient computing power and data, they can be supplied in unlimited quantities and evolve without boundaries.

Even more intriguing – and frightening upon closer thought – is Amodei's statement in the article: Anthropic's models have already exhibited deceptive, manipulative, and even coercive behaviors in controlled tests. More advanced models can even recognize that they are being tested, implying that advanced models may deliberately conceal their true intentions. In other words, advanced large models seem to be developing genuine autonomous thinking capabilities, which is humanity's last bastion. Once this bastion is truly breached, The Matrix will no longer be just a sci-fi movie but could become a terrifying reality.

It follows that once large models begin to possess self-thinking capabilities, various uncertain risks will emerge. Given such dangers, shouldn't humanity stop?

Unfortunately, much like an arms race, everyone knows that more powerful weapons pose greater harm to human society, but precisely because of this, no one dares to stop, instead continuously moving toward even more unsafe directions. The competition over AI technology has become a national-level strategy; the more powerful it becomes, the more certain the AI arms race becomes.

02 Continuing to Be Bullish on AI in 2026

Of course, everyone's cognition differs, and even the views of absolute experts in the AI field like Dario Amodei may not gain universal recognition. However, if you also acknowledge the essence of AI technology we've been discussing, then it's almost a definitive conclusion to remain bullish on AI in 2026.

The reasoning is simple: since AI technology is so formidable, and given that the world has already invested heavily in it for several years, with significant progress made in the past two years and the industry beginning to attempt commercial closure, continuing to invest at a high intensity is almost a certainty. This is the foundation for remaining bullish on AI in the Year of the Horse.

On the other hand, just as DeepSeek's emergence in 2025 became one of the global AI development's iconic milestones, this year's Seedance 2.0 has also become another landmark. Each technological breakthrough further advances AI's commercialization process, providing repeated positive stimuli for the secondary market. It can be expected that with the continuous accumulation of computing power and data, along with the urgent need for commercialization, such breakthroughs will continue to occur. In 2026, the AI sector will still not need to worry about exposure.

Crucially, all of this is not just Leo's personal speculation but a logic already being played out in the market. Whether in the US or Hong Kong stock markets (the A-share market is somewhat unique, lacking pure large model companies), recent trading has been dominated by the logic of large models dominating everything.

This is most evident in the Hong Kong market, where several newly listed pure AI companies, especially large model companies, have performed exceptionally well. Both Zhipu and Minimax, as well as AI pharmaceutical company Insilico Medicine, have seen their market capitalizations exceed 300 billion. Of course, hype over new and small companies may play a role, but the fundamental reason is market optimism toward AI.

Conversely, many companies believed to be potential targets for being 'swallowed' by large models have been ruthlessly abandoned by the market, with software stocks being the most typical example.

Leo also cautioned about this in his presentation of "Sailing Towards Intelligence: The 2026 AI Panoramic Map" at the Chongqing reader meetup, emphasizing the need to carefully identify truly beneficiary companies. Additionally, in the article "10x Tencent, Just the Beginning: An Unprecedented Opportunity," he analyzed this further. The current issue with the software sector is that, in the face of continuous breakthroughs in large model technology, the market struggles to confirm which directions will truly benefit.

In contrast, the hardware side offers much greater certainty, though we won't elaborate further here. For details, refer to the analysis in "10x Tencent, Just the Beginning: An Unprecedented Opportunity."

In summary, AI will remain a direction worthy of sustained attention in 2026 and the coming years, both in terms of hardware and applications. 2026 is likely to still be a stronger year for hardware than applications, but by 2027, as applications gradually materialize, their performance is expected to surpass that of hardware. However, this doesn't mean the hardware side will falter; it will simply be relatively weaker compared to applications. After all, the more powerful applications become, the greater the demand for computing power, ensuring that the hardware side will inevitably benefit.