WhiteLine Daily, consolidating the thinking of the Wu Shuo team to provide readers with the most valuable information and analysis of the day—helping them catch the trend shifts in the AI era.
1. Alibaba Cloud targets data center capacity of over 20 GW by 2032 and releases the Zhenwu V900 chip
Alibaba announced at the Cloud Rally Conference that it aims to increase Alibaba Cloud’s global data center capacity to more than 20 GW by 2032. Its subsidiary, T-Head, also released the AI chip Zhenwu V900, planning to start mass production and commercial release in the first quarter of 2027. The company said the new chip’s performance is three times that of the previous-generation M890. It can be used to build clusters of up to 500,000 chips for training and running large AI models. Alibaba CEO Eddie Wu said that strong customer demand for AI is driving growth in cloud business revenue, but supply chain constraints still limit the pace of capacity expansion.
One-sentence analysis: If a self-developed chip can be mass-produced on schedule, it will provide Alibaba with more options to expand its compute supply. However, the target of more than 20 GW still requires power, construction, and capital to be投入 in sync.
2, FT: SB Energy, seeking a valuation of about $50 billion, slows down its IPO
According to a report by the Financial Times, SB Energy, a U.S. data center developer backed by SoftBank, has slowed the process for its first public offering (IPO). The company is seeking a valuation of about $50 billion, but investors have raised concerns about the sustainability of its valuation and AI compute power demand. SB Energy currently has no data centers in operation and is waiting for investors’ sentiment toward the industry to stabilize. The report also says that the company’s potential debt financing plan of nearly $5 billion faces insufficient investor interest. Insiders said SB Energy still plans to go public and continues to evaluate market conditions.
One-sentence analysis: SB Energy’s decision to slow its上市 process shows that even support from tech giants may not be enough to get the public markets to accept its expected valuation. Its compute-expansion plan still needs to be tested by investors’ assessment of demand and returns.
3, Google signs to support capacity expansion of U.S. nuclear power units, planning to add about 96 MW of electricity capacity
Southern Company’s Georgia Power subsidiary has announced that it has signed a nuclear power capacity expansion agreement with Google. The company plans to increase Georgia Power’s electricity capacity by about 96 MW by upgrading relevant units at the Vogtle and Hatch nuclear power plants. The upgrades involve equipment such as turbines, pumps, motors, and cooling systems, using existing nuclear facilities to generate more power. Georgia Power said the arrangement is intended to meet the clean-energy needs of large customers while avoiding expansion-related costs being borne by other customers that did not participate in the agreement. The agreement is still subject to approval by the Georgia Public Service Commission.
One-sentence analysis: Google’s power-purchase commitment is being used to support capacity expansion of existing nuclear power generating units, which helps ensure that additional electricity demand corresponds to additional generation capacity.
4, California signs seven data center bills, requiring companies to bear grid and water-supply upgrade costs
California Governor Newsom signed seven data center regulatory bills requiring companies to bear the costs of upgrading the relevant power grid and water-supply infrastructure, to prevent costs from being passed on to other users. The new laws also strengthen disclosure requirements for water use, electricity use, and land use. Proposed projects must provide local governments and water utilities with information on water volumes, water sources, water-use efficiency, and drought-response plans, helping localities assess the project’s economic benefits and resource burden. On the power-supply side, the new laws also require data centers to comply with California energy procurement requirements and to bring additional clean energy supply to the grid.
One-sentence analysis: With clearer requirements that data centers cover grid and water-supply upgrade costs, project budgets need to include supporting expenditures beyond servers and buildings.
Today’s main line:
AI compute expansion requires both financing and the落实 of power and public infrastructure support. Only when the announced construction scale has secured funding, secured power supply, and the related costs are controllable can it gradually be converted into operational assets. These conditions together will affect the construction timetable and investment returns of data centers. Going forward, it remains to be seen whether customer demand can translate into continuous orders, and whether stable cash flow can be generated after the newly added compute capacity comes online. For developers of data center projects that have not yet started operations, the durability of demand and the outlook for capital recovery will also influence financing terms and valuation.