Tesla's Cortex 2 AI training cluster at Giga Texas is getting a significant energy storage upgrade. Aerial observer Joe Tegtmeyer spotted 18 freshly delivered Megapacks along the western side of the facility, with 8 of them already seated on their permanent beam and pier supports — a clear sign the installation timeline is moving fast.
Where Cortex 2 Stands Right Now
To understand why these 18 units matter, some context helps. According to permit filings and previous reports, Giga Texas started with roughly 130 Megapacks installed to support the first phase of Cortex 2. By February 2026, approximately 150 units had been delivered and partially installed, with room flagged for around 100 more. Permits filed as of May 8, 2026 indicated crews were working toward roughly 250 total Megapacks — essentially doubling the original footprint.
The first 250 MW phase of Cortex 2 went live in April 2026. The full cluster is designed to reach 500 MW of total power capacity, housing approximately 100,000 Nvidia H100 and H200 GPUs at complete deployment. That scale of compute demand requires equally serious power stability infrastructure — which is exactly what the Megapack expansion is there to provide.
Why Energy Storage Is Non-Negotiable at This Scale
AI training clusters at the 500 MW scale don't just consume enormous amounts of power — they need that power delivered with exceptional consistency. Grid fluctuations that would be a minor nuisance for a factory floor can corrupt training runs or damage sensitive GPU hardware. Megapacks act as a buffer, smoothing out supply irregularities and providing backup capacity during grid events.
The math on 250 Megapacks is substantial. Each unit stores around 3.9 MWh of energy, meaning a full buildout would represent roughly 975 MWh of on-site storage — nearly a gigawatt-hour sitting alongside one of the world's largest AI supercomputers.
The Bigger Picture: Texas as Tesla's AI Backbone
Giga Texas is increasingly functioning as more than a vehicle manufacturing site. Cortex 2 joins the original Cortex cluster (which supports Dojo and FSD training workloads) in making Austin the center of Tesla's AI compute strategy. The energy infrastructure surrounding these clusters is being built out with the same urgency as the compute hardware itself.
Worth noting separately: Tesla is also constructing a new Megapack manufacturing facility in Brookshire, Texas — near Houston — with Megapack 3 production targeted to begin in late 2026 and an annual capacity goal of 50 GWh. While those units are primarily destined for grid-scale energy storage customers, the proximity of that manufacturing ramp to Giga Texas's own growing Megapack appetite is a useful alignment of supply and demand.
With 8 of the 18 new units already on permanent supports less than a day after delivery, the pace of installation suggests Tesla is pushing hard to get the second 250 MW phase of Cortex 2 fully powered and operational. The gap between current installed capacity and the 250-unit target is closing quickly.
Sarah focuses on Tesla Energy, SpaceX missions, and the broader Musk AI portfolio. Former data analyst in clean energy. Based in San Francisco.
Sources verified at publish time. Spotted an inaccuracy? Email editorial@basenor.com.
