SpaceX Signs $920M Monthly Google AI Deal Ahead of IPO
Benzinga reported Friday that SpaceX has signed a staggering $920 million monthly compute-leasing agreement with Alphabet — the parent company of Google — granting access to approximately 110,000 Nvidia GPUs across SpaceX data centers through mid-2029. The deal was disclosed in a regulatory filing and comes mere days before SpaceX’s widely anticipated public market debut.
Terms of the SpaceX Google AI Deal
Under the agreement, SpaceX will ramp capacity to its full contracted rate by September 2026, with a reduced fee applying during the build-up period. Google holds an immediate exit clause if SpaceX fails to deliver the committed GPU count by that deadline. After September 2026, either side may exit the arrangement with 90 days of advance notice, giving both parties meaningful flexibility as AI infrastructure demands evolve.
Orbital Ambitions Run Alongside Terrestrial Compute
The terrestrial GPU lease is not the full picture of the SpaceX-Google relationship. The two companies are separately in discussions around orbital data centers. Google is reportedly planning prototype satellite launches by 2027 under an initiative called Project Suncatcher. That parallel track suggests the partnership could extend well beyond conventional cloud infrastructure into low-Earth orbit computing.
A Costly AI Division and a Monster Valuation
SpaceX’s IPO prospectus reveals the scale of its AI ambitions and their price tag. The company’s AI division posted an operating loss of $2.5 billion in the most recent quarter, even as it generated $818 million in revenue. Total capital expenditure for the period reached $10.1 billion, with $7.7 billion of that sum directed specifically toward AI infrastructure buildout. SpaceX is targeting a valuation above $1.75 trillion at its upcoming listing, a figure that would place it among the most valuable public companies in the world.
Background: The Neocloud Race Intensifies
The timing of the SpaceX Google AI deal is no coincidence. Demand for GPU compute capacity has exploded across the industry as hyperscalers and AI labs compete for scarce Nvidia hardware. Rivals in the so-called neocloud segment, including firms such as CoreWeave, face growing competitive pressure as well-capitalised players like SpaceX enter the market with both proprietary infrastructure and high-profile anchor customers. The sheer scale of the monthly fee underscores how critical secured compute access has become for frontier AI development.
Read Next: Nvidia Earnings Beat Sends Shares Higher as AI Demand Holds Firm
