NVIDIA announced on August 24 that SpaceXAI will deploy its new Vera CPU across its agentic AI infrastructure. Elon Musk confirmed the same day on X that SpaceX had already designed a space-optimized Vera Rubin NVL72 system for an orbital launch in the fourth quarter of 2027, with broader deployment expected in 2028.
"SpaceX, in partnership with Nvidia, has designed a space-optimized Vera Rubin NVL72 system for launch to orbit in Q4 next year, with significant scale in 2028," Musk wrote.
Neither stock rewarded the announcement. NVIDIA closed down 2.91% at $208.48, according to Google Finance data. SpaceX, which has traded as SPCX since its June IPO, fell 1.44% to $135 and slipped a further 0.22% in after-hours trade to $134.70.
The first CPU built for agents is going to work at scale. @SpaceX is deploying NVIDIA Vera to accelerate the orchestration, code execution, and data processing that powers its next generation of agentic AI — keeping GPUs fed and agents acting fast.
— NVIDIA (@nvidia) August 24, 2026
From gigawatt AI factories… pic.twitter.com/xJrxAw9EtQ
The Starmind satellite and the problem of computing in orbit
SpaceXAI's planned first-generation Starmind AI satellite will be built around an optimized Vera Rubin NVL72 rack-scale system. That means the same computing architecture used in large AI factories on the ground will need to function in an environment that works against conventional data center assumptions. Power is constrained. Thermal dissipation without atmosphere operates under different conditions from anything a data center engineer plans around, and each component must be built to survive the physical stress of launch.
NVIDIA and SpaceXAI said they are adapting the Vera Rubin architecture to those orbital requirements while keeping a common software ecosystem intact across both terrestrial and space deployments.
Ian Buck, NVIDIA's vice president of hyperscale and high-performance computing, framed the ambition directly.
"Vera gives AI agents the CPU performance to act in real time — executing code, processing data and coordinating complex tasks. SpaceXAI is taking this architecture from massive AI factories to the next frontier of computing in orbit," Buck said.
What the Vera CPU was built to handle
NVIDIA designed Vera specifically for agentic AI workloads. The chip carries 88 NVIDIA-designed Olympus cores with NVIDIA Spatial Multithreading technology. Memory bandwidth reaches 1.2TB/s through LPDDR5X. NVIDIA said Vera completes tasks up to 1.8 times faster than x86 CPUs across workloads that include reinforcement learning and data processing.
The engineering rationale is grounded in how agentic AI actually runs. Between model calls, AI agents require substantial CPU resources for tool use and code execution. Task coordination across an agent pipeline adds further demand on CPU capacity that GPU-centric infrastructure was not designed to handle. Vera targets those gaps so that GPU capacity remains focused on model inference.
Mike Nicolls, president of SpaceXAI, stated the operational benefit.
"Vera gives us the CPU performance and memory bandwidth to run enormous amounts of orchestration, code and data processing while keeping GPUs doing what they do best," Nicolls said. "That means higher-performance AI agents and more useful work from every watt of compute."
Musk added a note on the physical design:
"Our design is significantly simpler, lower cost, denser and lighter than a traditional rack."
SpaceXAI plans to scale Grok on the same Vera Rubin platform
The Vera Rubin deployment extends beyond the Starmind satellite. SpaceXAI plans to use the platform to expand the AI infrastructure behind Grok as the company moves toward gigawatt-level computing capacity. The platform brings together NVIDIA accelerated computing, NVLink interconnect technology, Spectrum-X Ethernet networking, BlueField data processing, and NVIDIA software in an integrated stack. NVIDIA said the architecture is designed to drive down cost per token at scale.
The combined plan puts Vera Rubin at the center of SpaceXAI's infrastructure across both its ground-based AI factories and its orbital computing program.
Why both stocks declined despite the announcement
NVIDIA's earnings report was scheduled for August 26, two days after the partnership went public. The stock trades near its 52-week range against a backdrop of concerns over memory costs and export restrictions in China. Attributing Monday's decline to the Vera announcement specifically is difficult given those surrounding factors.
SpaceX's post-IPO record provides relevant context. SPCX debuted in June, reached a high of $225.64, then fell to an all-time low of $104.83 on August 3. Monday's close at $135 put the stock below its IPO price. In after-hours trade, SPCX fell a further 0.22% to $134.70.
Roman Zaika
Disclaimer: All materials on this site are for informational purposes only. None of the material should be interpreted as investment advice. Please note that, despite the nature of much of the material created and hosted on this website, HODL FM operates as a media and informational platform, not a provider of financial advisory services. The opinions of authors and other contributors are their own and should not be taken as financial advice. If you require advice, HODL FM strongly recommends contacting a qualified industry professional.
