"Plenty of Capital, but Google-Blackstone AI Joint Venture Stalled on Data Center Construction"

Blackstone Invests $5 Billion in Joint Venture
Project Faces Setbacks Due to Infrastructure Shortages and Strict Regulations

It has been revealed that the cloud joint venture established by Alphabet, Google's parent company, and global alternative investment firm Blackstone is facing setbacks in building data centers. Despite their immense financial resources, they have encountered practical barriers such as shortages of infrastructure like electricity and equipment, local community opposition, and strengthened regulations, which have prevented them from moving forward with their projects. As the competition for investment in artificial intelligence (AI) accelerates, there is growing analysis that securing sufficient physical infrastructure to support the rapid surge in investment and demand is emerging as a new key success factor in the AI business.

According to Bloomberg on September 8 (local time), Project Braid, the joint venture between Alphabet and Blackstone, is reportedly facing such difficulties from the initial stages of the data center construction process. Project Braid is a business initiated through a $5 billion investment (approximately KRW 6.699 trillion) by Blackstone, with the goal of leasing Google’s proprietary AI chip, the Tensor Processing Unit (TPU), to clients starting next year.

Interior view of Google's data center located in Iowa, USA. Google.

To this end, Google had planned to build a data center in Cheyenne, Wyoming, USA, and entrusted the project to the data center developer Crusoe. However, Google lost confidence in Crusoe's operational capabilities and subsequently terminated the construction contract with them. Google then took over the project directly and is currently undergoing fresh permitting procedures. As a result, Bloomberg reported that there are now local rumors the project scale will be significantly reduced.

At other sites, infrastructure shortages became a stumbling block. The Texas project site was deemed unsuitable because it failed to secure the required transformers. Texas Governor Greg Abbott suspended new data center projects and called for a review of the electricity cost burdens on data center developers and technology companies.

Bloomberg pointed out that such delays in data center construction are becoming widespread across the industry. According to a report released by JPMorgan Chase this past May, more than 60% of the data center capacity scheduled for completion next year has not even begun construction yet. The lack of electrical equipment such as switchgear and a shortage of skilled labor are acting as bottlenecks, further delaying construction. McKinsey & Company has also analyzed that supply shortages for equipment necessary to build data centers have lengthened the wait time for some equipment to nearly a year.

Expectations that data centers will be completed on schedule have also dropped significantly. Senior officials at Project Braid now estimate there is only a 50% chance the data center business will meet its planned completion timelines. Three years ago, this probability was assessed at about 90%, indicating a reduction of nearly half.

The two companies are accelerating the process of securing alternative sites. It is reported that Project Braid has currently identified 29 potential data center candidate sites. Google has recently asked partner companies to search for additional land, while Blackstone is also mobilizing consultants and partner firms to seek out new candidates.

Bloomberg noted, “This may become an obstacle for Blackstone in its ambition to become the world’s largest digital infrastructure investment firm,” adding, “Google, which needs to build an ecosystem to rival Nvidia, faces a similar predicament.” Google entered into the joint venture with Blackstone to share some of the costs and business risks associated with data center construction. This year, Google plans to undertake as much as $205 billion in capital expenditures.

This content was produced with the assistance of AI translation services.