Reflection AI, an open-source AI lab, will pay SpaceX roughly $1.8 billion annually for dedicated access to Nvidia's GB300 NVL-class hardware at Colossus 2 — a facility near Memphis that sits alongside Tesla's existing Colossus cluster. The deal runs from July 1, 2026 through 2029, representing a multi-year committed revenue stream for SpaceX's compute business and a significant vote of confidence in GB300 supply availability at scale.

For NVDA investors, the signal is clear: hyperscalers and now well-funded AI labs are signing long-term, large-dollar commitments around the GB300 generation specifically, extending the capex supercycle narrative into 2027–2029. The key question is whether NVDA's supply chain can meet this wave of demand without margin pressure, and whether SPCE (Virgin Galactic — unrelated to SpaceX) sees any sympathy noise that fades quickly. SpaceX remains private, so the primary tradeable angle centers on NVDA as the hardware supplier anchored in the deal.