Oguz Erkan(@oguzerkan):$META compute optionality is ridiculously undervalued. $META is expected to have 5 GW of excess compute by 2030. Its cost of deploying 1 GW of capacity is reportedly around $38 billion. Wells Fargo estimates it can charge around $20 billion per GW with around a 73% net margin, resulting in $14.6 billion in net income per year. This means a 38% annual after-tax return on the initial compute capex. Even if we assume that its average total cost of deployment increases 50% by 2030 and settles around $57 billion while the resale price remains the same, we’ll still be looking at a 26% after-tax ROI. The market gives almost $0 value to this.

2026.08.26 10:58

$META compute optionality is ridiculously undervalued. $META is expected to have 5 GW of excess compute by 2030. Its cost of deploying 1 GW of capacity is reportedly around $38 billion. Wells Fargo estimates it can charge around $20 billion per GW with around a 73% net margin, resulting in $14.6 billion in net income per year. This means a 38% annual after-tax return on the initial compute capex. Even if we assume that its average total cost of deployment increases 50% by 2030 and settles around $57 billion while the resale price remains the same, we’ll still be looking at a 26% after-tax ROI. The market gives almost $0 value to this.