The main risk is a sequential chain: energize
Sweetwater, secure funding, receive and commission GPUs, then contract capacity at good economics. None of those steps is implausible alone, but the path is tight enough that
dilution or timing slippage can absorb much of the upside even if long-run AI demand remains strong.