"Computing-Power Collaboration" is Still Adding More Value! Has AI's HALO Moment Arrived?

The value of “computing and energy collaboration” is still increasing.

As this term was first included in the government work report and explicitly listed as a national-level new infrastructure project, the market responded enthusiastically: the power sector experienced a surge in daily limit hits, with wind power, energy storage, and green electricity becoming hotly pursued assets.

In a rare long article, Jensen Huang also defined the energy layer as the fundamental principle of AI infrastructure. He emphasized that real-time intelligent generation requires real-time electricity, with each generated token resulting from electron movement, heat management, and energy conversion into computing power. Currently, energy supply has become an urgent bottleneck for large-scale AI development.

Expanding the view globally, it becomes clear that the key to AI success has shifted from computing power arms race to power supply. In this future-related competition, China has taken the lead, while Europe and the US are catching up. The importance of power supply is rising, and AI is entering its own “HALO moment.”

Back in December 2023, when ChatGPT first emerged and sparked a global AI frenzy, China’s National Development and Reform Commission and other departments issued the “Implementation Opinions on Deepening the ‘East Data West Computing’ Project and Accelerating the Construction of a National Integrated Computing Network,” which for the first time proposed a “dual cooperation mechanism of computing power and electricity.”

More than two years later, in March 2026, this concept was officially elevated and included in the government work report for the first time. A relevant official from the State Council Research Office emphasized during interpretation that “we must leverage the advantages of our national power grid system.”

According to Zheshang Securities, this marks a strategic leap from technological exploration to top-level national design for “computing and energy collaboration.” It is not only a deepening of the “East Data West Computing” project but also a necessary path to address the two core bottlenecks of “electricity cost” and “green power absorption” in the AI era.

With OpenClaw driving a surge in token demand, China’s large model invocation volume has increased significantly, creating opportunities for high-latency inference computing power to go overseas. Caitong Securities reported that the accelerated construction of western data centers not only solves wind and solar power absorption issues in the west but also forms global competitiveness with low-cost green electricity at 0.3-0.4 yuan per kWh, becoming a key breakthrough in the integration of energy and computing power.

▌Europe Faces “Electricity Shortage”

The European Commission recently approved the Clean Energy Investment Strategy, aiming to mobilize more private investments into grid construction, innovative clean energy technologies, and energy efficiency improvements. The EU will cooperate with the European Investment Bank Group to implement this strategy, which will provide over 75 billion euros in financing over the next three years.

In the press release, the EU admitted, “The current geopolitical situation reminds us once again that Europe’s over-reliance on imported fossil fuels poses risks. Citizens and businesses are deeply concerned about high energy prices.”

The “AI electricity shortage” is also a problem Europe faces. For example, Italy’s data centers are relatively small but have ambitious expansion plans. A2A CEO Renato Mazzoncini pointed out, “Since January 1, 2025, the total demand for new data center connections to the grid has reached 70 GW—while Italy’s peak electricity demand is about 50 GW, and Milan’s peak is only 1.5 GW.”

In this context, the EU is also turning its attention to “computing and energy collaboration.” The European Commission plans to convene member states, grid operators, and data center developers to work together; additionally, the revised Energy Efficiency Directive is expected to introduce new data center rating schemes to encourage integration into the energy system.

Notably, due to companies’ inability to wait for grid bottlenecks to ease, they are turning to self-built power sources. Recently, Europe’s first microgrid directly connected to a data center was officially put into operation. Located in Dublin, Ireland, it was developed by Pure Data Centre Group and AVK, a power solutions provider. Microgrids, which generate, store, and distribute electricity locally, have been widely used in the US.

▌Silicon Valley Giants Attempt to “Transform the Power Grid”

Similar situations are happening in the US, which has long suffered from an “AI electricity shortage.”

Recently, President Trump stated that his administration has told major tech companies they must build their own power generation facilities for data centers.

Subsequently, Google, Tesla, and five other companies in the power equipment and data center supply chain announced the formation of the “Utilize” alliance this week, aiming to improve US grid utilization, reduce electricity costs, and significantly increase power demand.

The alliance pointed out that US grid utilization has been persistently low; more efficient use of existing grids could save American consumers over $100 billion in electricity costs over the next decade. They listed solutions including battery storage, demand response, and virtual power plants—technologies that have rapidly emerged over the past ten years but are still underutilized.

This underscores how critical electricity is to these tech giants—moving from waiting for grid connection and self-building power sources to actively attempting to deeply transform the energy landscape.

As Elon Musk previously said, the real bottleneck for AI development is “power generation.” Using his xAI data center in Memphis as an example, connecting high-voltage transmission lines takes a year, forcing temporary use of gas turbines and battery packs to supply power.

▌"Aggressive HALO Assets"

Recently, the “HALO trading” wave swept the globe. Behind this, investors are no longer blindly chasing lightweight asset narratives but are embracing heavy assets, driven by “AI panic” and “counter-AI” trades.

However, the emergence of “computing and energy collaboration” balances these two, finding an AI “HALO moment”—a safe haven asset against “counter-AI” risks and a growth asset aligned with “pro-AI” development.

Qianhai Open Source Fund classifies industrial metals, power equipment and grids, oil transportation, and logistics as “offensive HALO assets.” Their core value lies in “the greater the AI development, the higher the demand,” possessing both the “hard asset” qualities of HALO assets and sharing the dividends of AI growth.

Everbright Securities also pointed out that the essence of “HALO trading” is the re-pricing of capital on “scarcity” assets. In the context of continuous AI technological iteration, the truly scarce resources are the energy and power systems supporting computing power and the infrastructure maintaining social operation. China, with its comprehensive supply chain system and global leadership in new energy, power equipment, strategic metals, chemicals, and shipbuilding, will become a hard asset in the AI era. With industrial upgrades and industry consolidation, its strategic value may undergo a systemic revaluation.

Technology must learn to coexist with the physical world. Historians often associate the Industrial Revolution with coal and oil, but now the AI revolution appears deeply intertwined with electricity. As the market begins to measure “bits” in watts, AI is entering its own “HALO moment.”

(Article source: Caixin)