Briefing
- Google has agreed to rent computing capacity from SpaceX at $920 million per month, generating a total contract value of approximately $30 billion over 32 months.
- The compute capacity is hosted at xAI data centres, linking SpaceX's infrastructure business directly to Elon Musk's AI venture.
- The deal is structured ahead of a planned SpaceX IPO, providing the company with a substantial recurring revenue stream that will feature prominently in any prospectus.
- At nearly $1 billion per month, the contract represents one of the largest single cloud-compute procurement commitments by a hyperscaler on record.
- For Alphabet, the arrangement signals a willingness to source AI infrastructure externally rather than rely solely on its own data centre build-out, a notable shift in capital allocation strategy.
Analysis
- The $30bn compute lease strengthens the SpaceX IPO valuation case by locking in roughly $920 million per month in contracted recurring revenue, the single most bankable line item any prospectus could display. Investors pricing a SpaceX float will now treat this as near-annuity cash flow, compressing discount rates applied to the satellite and launch segments and likely pulling the implied enterprise value materially higher than pre-deal estimates. xAI's data centre infrastructure is simultaneously monetised before it has demonstrated standalone AI product revenue.
- Alphabet sourcing external compute at $920 million per month directly contradicts the narrative embedded in its $85bn equity raise, which earmarked $40bn for internal AI infrastructure build-out. The two moves in combination suggest Alphabet's own data centre capacity is insufficient to meet demand it described as 'exceeding available supply', validating the supply constraint thesis. This is modestly positive for third-party AI infrastructure operators and server OEMs, and raises the question of whether Google's internal capex guidance is understated.
- The Google-SpaceX deal compounds the negative read-through from Broadcom's flat AI guidance. Broadcom's custom accelerator business depends on hyperscalers concentrating AI compute in their own facilities using Broadcom ASICs; Alphabet leasing capacity from xAI data centres instead of expanding in-house implies a portion of incremental AI workload will run on infrastructure Broadcom does not supply. If other hyperscalers adopt similar external leasing to bridge supply gaps, demand for Broadcom's custom silicon could grow more slowly than consensus capacity expansion plans imply.
Historical Context
- 2022-2023
Microsoft's multi-billion-dollar Azure capacity commitments to OpenAI established the precedent of a hyperscaler funding an AI lab's infrastructure in exchange for preferential compute access. The Google-SpaceX deal inverts the structure: Google is the buyer, not the funder, and the AI lab's infrastructure is already built, but the monetisation logic of locking in contracted recurring revenue ahead of a public offering is identical.
- 2021
AWS and Google Cloud competed aggressively for multi-year enterprise compute contracts worth $1-5bn over contract life. A $30bn single-vendor commitment at $920 million per month is an order of magnitude larger than those precedents, setting a new floor for what constitutes a material hyperscaler procurement event and resetting how investors will value contracted cloud and compute revenue streams.
Connected Events
Alphabet upsizes equity raise to $85bn, largest secondary share sale on record
Alphabet's $85bn equity raise earmarked $40bn specifically for AI infrastructure and compute capacity, with management citing demand 'exceeding available supply'. The SpaceX lease, signed in the same period, reveals that internal build cannot satisfy that demand fast enough, making the two capital allocation decisions contradictory on the surface and raising questions about the actual timeline of Alphabet's internal data centre expansion.
Broadcom sheds ~$300bn in market value after AI guidance disappoints
Broadcom's flat AI revenue guidance triggered a roughly $300bn market cap loss, with investors questioning hyperscaler AI capex commitment. Alphabet leasing $30bn of compute externally rather than deploying exclusively through its own internal chips and custom ASICs adds incremental evidence that hyperscaler demand is real but may not flow to established semiconductor suppliers in the proportions the market had assumed.
Dell Q1 FY2027 AI server revenue surges 757%, stock posts best day in two years
Dell's 757% year-on-year AI server revenue surge confirmed hyperscaler infrastructure spending is accelerating. The Google-SpaceX deal adds a novel channel for that spending, bypassing traditional OEM and cloud buildout cycles entirely, which could limit how much of the incremental hyperscaler AI budget flows through Dell and Super Micro's supply chains in future quarters.
Further reading
See Indexa more often on Google
Mark Indexa as a preferred source — your Top Stories will surface more Indexa coverage.