Nvidia (NVDA) Stock Rises as AI Hardware Demand Grows in Space and on Earth

TLDR

• Mistral AI raised $830M and plans to buy ~13,800 Nvidia GB300 GPUs for a new Paris data center, potentially worth ~$575M in chip sales
• Space startup Starcloud raised $170M at a $1.1B valuation, planning a more powerful satellite GPU cluster after its first H100-in-orbit success
• Nvidia unveiled its Space-1 Vera Rubin computing module on March 16 to tackle data bottleneck issues in orbital data centers
• Global cloud infrastructure spending hit $110.9B in Q4 2025, up 29% year-over-year, with 27% growth forecast for 2026
• Broadcom (AVGO) slipped 1.3% and AMD gained 1% in early Monday trading

Nvidia’s Monday session started on a quiet but positive note, with two fresh announcements giving the stock a nudge higher.

NVIDIA Corporation, NVDA

The bigger headline came from French AI company Mistral AI, which said it raised $830 million in debt financing — its first ever — to build a new data center near Paris. The facility will house 13,800 of Nvidia’s GB300 GPUs. Based on HSBC estimates that a GB300 NVL72 rack costs around $3 million, that single data center could represent roughly $575 million in chip purchases for Nvidia.

Mistral didn’t confirm exact pricing, and Nvidia doesn’t disclose individual chip costs. But the scale of the order reflects what analysts have been tracking for months: AI hardware demand isn’t slowing down.

Global cloud infrastructure spending reached $110.9 billion in Q4 2025, a 29% increase from the same period a year earlier, according to research firm Omdia. The firm forecasts another 27% growth rate for 2026 — the kind of tailwind Nvidia has been riding for the past two years.

Nvidia Eyes the Final Frontier

The more unusual story of the day comes from orbit. Starcloud, a startup building data centers in space, said it raised $170 million at a $1.1 billion valuation. The company already sent one of Nvidia’s H100 GPUs into space late last year aboard its Starcloud-1 satellite — the first time an AI model was trained in orbit.

Now Starcloud is planning a second satellite launch later this year, featuring a full GPU cluster and the largest commercial deployable radiator ever sent to space, generating 100 times the computing power of the first craft.

Space-based data centers are pitched as a potential solution to growing tension around terrestrial facilities — concerns over power grid strain, water use, and community pushback have made building on Earth increasingly complicated. Orbital systems could tap solar energy and skip traditional cooling altogether.

That said, the technical and cost hurdles remain enormous, and the market is still in its very earliest stages.

Nvidia moved proactively on this front two weeks ago. On March 16, it announced the Space-1 Vera Rubin computing module — hardware designed specifically to address one of the core problems with space-based systems: data bottlenecks. Because bandwidth between satellites and Earth is limited, Nvidia’s new module processes data where it’s generated rather than sending raw data back down for analysis.

Competition and Context

Nvidia isn’t the only company eyeing the orbital computing market. Elon Musk’s SpaceX has been linked to plans for solar-powered orbital data centers, expected to be funded through a potential public listing. Jeff Bezos’s Blue Origin, meanwhile, is seeking regulatory approval to put nearly 52,000 satellites capable of AI computing into orbit.

Nvidia currently trades at a forward price-to-earnings ratio of around 21.4, lower than in recent quarters but still pricing in growth. Its market cap sits above $4 trillion.

Nvidia has not yet said when it will begin shipping the Space-1 Vera Rubin module to customers.