Blackstone is Wagering $30B On Japan’s Artificial Intelligence Industry

Alternative asset manager Blackstone is investing $30 billion into AI datacenters in Japan over the next 3–5 years. This was announced by Blackstone’s President & COO Jonathan Gray in a recent Nikkei Asia article published on June 23rd 2026. The goal is to have over 1 gigawatt (GW) of computing power available, equivalent to that of a nuclear power plant, but solely focused on AI servers operating 24/7.

Datacenters are the brick-and-mortar behind all AI applications. When you send a prompt to ChatGPT or Claude, there’s a physical location somewhere in the world with rows of servers, air conditioning units, and fiber optic cables that process your request and return a response. These facilities are expensive to construct and take years to design, so choices made today will shape the landscape of AI infrastructure for much of the remainder of this decade. Currently, there’s a shortage of supply relative to demand.

1Japan becomes Asia's AI compute hub

Japan has become one of the most desirable locations for AI infrastructure in Asia. It has a resilient electricity network, political stability, a skilled labor force, and a strategic location that positions it well as a gateway to the Asia-Pacific. Other players are also investing heavily here; Alibaba recently launched its 5th datacenter in Japan and Brookfield, another large infrastructure investor, expects its Japanese division to grow faster than its businesses in Hong Kong and Singapore combined.

Blackstone has existing exposure to Japan via its investment in AirTrunk, a datacenter company it purchased in 2024. Earlier this year (March 2026), AirTrunk closed a ¥191.6 billion ($1.24 billion) green bond, which it plans to use to grow its Tokyo-based AI campus. This latest announcement represents a significant increase on top of that base.

It’s interesting to hear how Gray views this investment opportunity. He noted that “the downside of not having enough computing power is greater than the upside of a potential AI bubble.” This is coming from someone who’s paid to think about these things. If even the biggest players in infrastructure think that not having enough capacity is worse than having too much, then there must be some real conviction behind the idea that there will be sufficient demand for AI services.

2What this means for AI costs and availability

Over the last couple of years, the price of AI has dropped dramatically. Something that would have cost $1 to run in 2024 now costs fractions of a cent. Part of this is due to increased competition, improved software optimization, and ongoing development of purpose-built AI infrastructure. Each additional GW of AI optimized compute coming online globally increases the total amount of capacity in a tight market (since ChatGPT was released). Increased competition leads to lower prices.

Japanese compute will be particularly useful for applications targeting Asian audiences where reduced latency and local data flows are important. However, the impact will be felt worldwide as each large investment changes the equilibrium and puts downward pressure on prices, benefiting everyone building on top of AI, no matter where their customers live.