In the morning, I saw the news that BlackRock has acquired the American power company $AES. Of course, the deal is still in the negotiation stage and has not been completed yet, but this event itself already indicates that, in the phase of rapid AI development, what is scarce is not only storage but also electricity, and it must be said that "electricity" is the foundation of all computational power.

Therefore, the reason for BlackRock's acquisition of AES, or the reason BlackRock is interested in "electricity," is likely one of the following:

1. The "essential need" of the AI era is shifting from chips to electricity and grid connection.

Although GPUs and memory are scarce products, they can be achieved by expanding production capacity, but insufficient electricity, the need to queue for grid connection, and inadequate substation capacity turn computational power into a joke. Moreover, obtaining electricity is much more difficult compared to GPUs and memory, and the costs required are also much higher.

2. Electricity is equivalent to a "toll booth" for computational power.

Chips are the physical architecture, and models are the higher-level applications, but all paths ultimately return to "watts." As data center expansion becomes a long-term trend, electric assets are like toll booths on highways, with more rigid demand, longer-lasting bargaining power, and the need for more infrastructure investment as expansion continues. Private investment becomes even more difficult.

3. Electricity is the "cash flow" of infrastructure.

The business model of electricity is not merely because of rising prices but because it is suitable to become assets with "long-term contracts and predictable cash flow." As long as there is a demand for computational power, electricity is essential. For example, AES recently signed a 20-year power supply agreement with Google.

4. What AI needs is not just "electricity," but "stable electricity."

AI data centers require stable loads 7x24, and they may need to expand at any time. Therefore, what is needed is not only electricity itself but also deliverable electricity, ideally guaranteed electricity.

5. When "electricity" becomes the foundation, the next layer of scarcity falls on the "power supply chain" (by the way).

Here, "chain" does not refer to blockchain but a chain of connections. When many people discuss power bottlenecks, their first reaction is to think of power plants, grid connection, and substations, but the electricity that AI truly needs is not the kind you can use right away once connected; it needs to be processed all the way to become computational power.

For example, from the power grid to the machine room, a significant electrical transformation must take place: high voltage comes in, low voltage goes out. From the machine room to the cabinets, distribution architecture, redundancy, UPS, PDU, and busbars must be addressed. From the cabinets to the servers, AC needs to be converted to DC, and 12V or 48V needs to be transformed into electricity that GPUs and HBMs can actually use.

Going further upstream to the equipment side of the wafer fabs, etching and deposition processes essentially use extremely fine and controllable electricity to drive plasma. Unstable electricity doesn't just slow down the machines; it leads to decreased yield and increased costs.

This is also why companies like Advanced Energy $AEIS (which is different from AES) will become more crucial in the AI era. Electricity is important, but how to "make electricity usable" is itself part of the electric supporting infrastructure.

Simply put, when the market begins to realize that "electricity" is the foundation of computational power, everyone will definitely pay attention to the source of "electricity." However, how "electricity" is transformed into computational power is a hidden and crucial layer. This is also why I think such "power supply chain" companies deserve to be re-evaluated.

@bitget VIP, lower rates, better benefits

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。