The narrative logic behind the hot trend of HALO assets

Ask AI · How Can HALO Assets Become the Foundation Infrastructure for the AI Era?

By Xie Changyan Edited by Lin Weiping

Since 2026, while everyone is still amazed by— or even anxious about— the pace of AI iteration, the concept of HALO assets is drawing attention from capital markets. International investment banks such as Goldman Sachs and Morgan Stanley have also repeatedly emphasized key positioning in it.

This image may be generated by AI

HALO assets originate as an abbreviation of “Heavy Assets (heavy asset base)” and “Low Obsolescence (low obsolescence rate).” Their core characteristics are large asset scale, slow industry iteration, and stable cash flow. The key lies in their competitive barriers and asset quality. In the AI era, this type of asset will not be replaced, but it is also not simply “old-timer assets.” Instead, it serves as infrastructure for AI development— demand remains strong and even determines a country’s competitiveness in the AI era. It is an important support for the industrial realization of AI.

For example, the power sector is the core representative of HALO assets. In the future, the core of competition between countries will be computing power and electricity. Hydropower, thermal power, and new energy generation sectors such as wind and solar with storage all have heavy-asset characteristics. They are the energy foundation for AI big-data centers and the operation of information highways, and they belong to long-term key development directions. Industrial metals are also core HALO assets. Copper, aluminum, cobalt, nickel, and others have heavy-asset characteristics. The United States is actively stockpiling metals such as copper because the construction of big-data centers requires large quantities of copper wire, and the amount of copper used in electric vehicles is 4 times that of traditional cars. Tight supply-and-demand conditions are creating opportunities for these assets to receive a major repricing.

Since the start of the year, these assets have seen notable gains, confirming their investment value and also highlighting capital’s preference.

So, why have HALO assets suddenly become popular? The answer may be that most essential line in investing: invest in two directions— one is what cannot be changed by the world, and the other is what changes the world.

As AI technology moves from large models to penetrating the application layer, the market has begun to worry about its “creative destruction” of light-asset business models such as software and the internet. When software can be rewritten by AI, capital starts to feel lost: what, exactly, is the eternal moat? The answer may be those “things that AI cannot replace.”

Therefore, for the purposes of hedging risk and seeking certainty, capital is shifting from areas that may be disrupted to HALO assets with tangible barriers and slow technological iteration, thereby offsetting AI risks. And the more AI develops, the more critical the “shovel-seller” becomes. The hard-core logic is that AI’s own development actually strengthens dependence on tangible hard assets like HALO.

For example, every conversation you have with Doubao or Yuanbao, or every instruction you give to a lobster, all consume electricity behind the scenes. Data from the International Energy Agency (IEA) shows that in 2024, global data center electricity consumption was about 415TWh, and it is expected to soar to 945TWh by 2030, with an average annual growth rate as high as 15%.

From an investment perspective, in the AI era investing should focus on HALO assets as the core— emphasizing power, industrial metals, and other heavy-asset sectors with irreplaceable characteristics. At the same time, it is also necessary to take into account high-quality targets for AI applications such as humanoid robots, return to fundamental research, and only then can investors capture long-term investment opportunities.

(This article has been published in the Securities Market Weekly on March 28.)