Elon Musk discussed the construction and operational approach behind xAI's Colossus mega-datacenter, built at an accelerated timeline.

In a recent post on X, Musk outlined how xAI accelerated its data center construction through three primary tactics: paying above-market wages to secure the best crews immediately, moving facility components between locations to optimize speed, and deploying industrial robots that operate continuously without shift changes or recruitment pressures.

**The Colossus cluster results**

The scale of this approach is most visible at Colossus, a sprawling data center complex spanning the Memphis, Tennessee, and Southaven, Mississippi, region. (xAI has since been folded into SpaceX's AI division following a 2026 acquisition.) Colossus 1 was built from a repurposed Electrolux manufacturing site and reached operation in July 2024 after just 122 days of construction, launching with approximately 100,000 NVIDIA H100 GPUs.

Colossus 2 moved faster on several fronts. After xAI acquired the site in March 2025, the facility reached roughly 200 MW of cooling capacity within six months. By September 2026, it housed 110,000 GB200 chips and 440,000 GB300 chips. Musk plans to add three additional batches of 220,000 GB300 units each, potentially bringing the cluster past 1.2 million advanced NVIDIA processors by year-end. Investment in the Colossus cluster had exceeded $35.8 billion by mid-2026.

**Power strategy and local concerns**

xAI bypassed lengthy permitting delays by bringing its own power infrastructure—currently drawing hundreds of megawatts from rented mobile gas turbines, with plans for a permanent 1.2 GW natural gas plant. This approach, combined with temporary cooling setups and prefabrication off-site, allowed the company to sidestep standard interconnection timelines. Local opposition has focused on air quality, noise, and unpermitted equipment; xAI has committed to removal timelines extending into 2027.

**Revenue generation**

xAI signed a multi-year rental agreement with Google valued at $920 million monthly, covering approximately 110,000 GPUs from October 2026 through June 2029. This arrangement demonstrates the cluster's ability to generate revenue beyond xAI's proprietary models.

The unprecedented pace—a major cluster operational in 122 days—now sets the benchmark for the broader AI industry. The $35.8 billion commitment signals sustained confidence in rising AI compute demand, while the massive GPU order represents a significant portion of NVIDIA's advanced chip production. Regulatory changes affecting temporary power sources or delays to the permanent gas plant remain key risks.