Elon Musk Cheers NVIDIA’s Vera Launch After SpaceX Gets First Units

NVIDIA has tapped SpaceX as one of the first customers to deploy its new Vera CPU, the chipmakers debut processor designed specifically for agentic artificial intelligence (AI) workloads. The collaboration drew quick public backing from SpaceX chief executive Elon Musk, who reposted the announcement on X and joked that the chip lived up to its name. NVIDIA Ships First Vera Units to SpaceX NVIDIA confirmed that early Vera silicon went directly to SpaceX, OpenAI, Anthropic, and Oracle Cloud Infrastructure. The chipmakers hyperscale division hand-delivered the first units, framing the rollout as the start of a wider commercial deployment. The Vera CPU houses 88 custom NVIDIA-designed Olympus cores and supports up to 1.2 TB/s of memory bandwidth using LPDDR5X memory. NVIDIA claims the chip runs agentic sandbox workloads up to 50% faster than competing rack-scale CPUs while doubling efficiency. A new 256-CPU rack configuration sustains more than 22,500 concurrent agent environments at full performance, according to NVIDIA. The system is part of the broader Rubin platform revealed at GTC in March 2026. Musk Reacts on X Musk posted his reaction shortly after NVIDIAs official AI Infrastructure account thanked SpaceX for testing the chip. Vera nice, Vera nice 🤌 Elon Musk (@elonmusk) May 19, 2026 Elon Musk, Source: X His enthusiasm follows the recent folding of xAI into SpaceX, which now operates the merged AI division under the SpaceXAI brand. The unit already runs the Colossus 1 and Colossus 2 supercomputers in Memphis. Why Vera Matters for Agentic AI Vera signals NVIDIAs first serious push into CPUs aimed at AI agents rather than the GPU products that dominate its revenue. The launch coincides with rising enterprise demand for agent-based workloads that orchestrate many smaller models. NVIDIA chief executive Jensen Huang has called agent-based services the companys next multi-trillion-dollar opportunity. Customers lined up for deployment include Alibaba Cloud, ByteDance, Meta, CoreWeave, Lambda, and Nscale, alongside SpaceX. The breadth of early adopters suggests NVIDIA intends to extend its AI infrastructure dominance well beyond training silicon. Whether Vera meaningfully shifts CPU market share will depend on how quickly customers like SpaceX move from testing to full production. Several rivals, including AMD, are also working on competing chips aimed at the same workloads.