SpaceX Names NVIDIA Exclusive Chipmaker on Vera Rubin Architecture; Alpamayo 2 Super Ships

NVIDIA got a market boost after Elon Musk, on SpaceX's first earnings call since listing, said SpaceX will build exclusively on NVIDIA's Vera Rubin architecture, calling it 'the best AI computer' available. The endorsement — from a company that also owns xAI and could plausibly pursue custom silicon — sent NVIDIA shares up more than 3%. Musk added that xAI's computing power is 'fully bound to NVIDIA,' reinforcing NVIDIA's grip on the highest-profile AI buildouts even as Anthropic, OpenAI and Meta pursue in-house chips.

Separately, NVIDIA made Alpamayo 2 Super commercially available — its frontier open reasoning model for autonomous vehicles. CEO Jensen Huang's launch post (14,865 likes) described a model that 'understands and reasons through the complex world — thinks before it acts,' positioned as a backbone for robotaxis, trucks, shuttles and delivery vans. NVIDIA also debuted a Synthetic Video Detector NIM microservice reaching up to 92% accuracy and surviving compression and re-encoding, plus Nemotron 3 Ultra for top agent tasks, available via NIM and Hugging Face weights.

The competitive subtext is that NVIDIA's dominance is being contested precisely as it's being reaffirmed: the biggest labs are building custom chips to reduce dependence, yet the biggest deployments (SpaceX, xAI) are doubling down on NVIDIA. Both can be true — custom silicon takes years, while frontier training needs the best GPUs now.

NVIDIA reports Q2 FY27 results on August 26, guiding to roughly $91 billion in revenue (±2%). That number sits against a backdrop of the week's compute-scarcity theme — AWS sold out through 2028, Anthropic's $10B Volta deal — all of which ultimately flows through NVIDIA's order book. For readers, the watch item is whether the in-house-chip movement across labs eventually dents NVIDIA's pricing power, or whether demand simply outstrips any custom-silicon substitution for years to come.