Meta is seeking to raise up to $95 billion, about 126.0 trillion won, to build AI data centres. It plans to secure funding for large-scale investment by issuing shares and mandatory convertible preferred shares, aiming to gain an edge in the race for AI infrastructure.

On Oct. 26, local time, IT outlet SiliconANGLE reported that Meta announced it would begin a share and depository share offering worth about $35 billion. It also said it is pursuing a plan to raise up to an additional $60 billion early next year.

Of the initial offering, $22.5 billion will be raised through sales of Class A common stock. Nvidia, AMD and Broadcom also expressed interest in buying shares worth a combined $5 billion. No investment agreements have been finalised.

The remaining $12.5 billion will be raised by issuing Series A depository shares based on mandatory convertible preferred stock. The preferred shares are set to convert into common stock in August 2029.

Meta later plans to raise up to an additional $60 billion using non-voting Class C shares. Under a plan to sell shares in tranches depending on market conditions, the specific size and schedule have not been decided.

The funds raised will be used to build AI data centres and supercomputing infrastructure. Meta is currently building four AI superclusters in the United States with a capacity of more than 1 gigawatt (GW) each.

A flagship project is the 5 GW Hyperion data centre being built in Louisiana. Meta plans to bring 10 additional power generation facilities online to meet electricity demand.

It is also expanding cooperation with outside investors. Blue Owl Capital is participating in the Hyperion project, and the Texas data centre is being jointly 추진 with BlackRock.

Meta expects capital expenditure this year to reach $135 billion to $145 billion, about 179.0 trillion won to 192.3 trillion won. It plans to lower inference costs while expanding use of its AI model Muse, and also improve the efficiency of idle computing resources.

It also points to potential dilution of existing shareholders from large-scale share issuance and whether it can recoup AI investments as key challenges ahead.