Independent · Updated October 29, 2026 · Every claim traceable
Home › AI Infrastructure › Google Pays SpaceX $920M/Month for AI Compute: The $…
AI Infrastructure

Google Pays SpaceX $920M/Month for AI Compute: The $30B Bridge Deal

By the Pavitrams editors · October 29, 2026 · 8 min read
Google Pays SpaceX $920M/Month for AI Compute: The $30B Bridge Deal

SpaceX's IPO filing revealed a massive cloud services agreement: Google (Alphabet) will pay $920 million per month for access to approximately 110,000 Nvidia GPUs, CPUs, and memory housed in SpaceX data centers. The deal runs at full rate from October 2026 through June 2029, totaling roughly $30.4 billion in contracted compute revenue — the largest AI infrastructure lease ever disclosed.

The short answer
Update note — October 29, 2026

This article is based on CNBC's June 5, 2026 reporting on SpaceX's SEC Form S-1 Amendment No. 2 filing, Google Cloud's statement, and subsequent financial coverage. All figures are as disclosed in the filing or reported by the cited outlets. Market conditions and contract terms may change; verify against the live SEC filing before making investment decisions.

Days before a planned IPO expected to raise record sums, SpaceX disclosed in an SEC filing that Google (Alphabet) will pay $920 million per month for AI compute capacity at SpaceX data centers. The agreement spans from October 2026 through June 2029 at the full rate, with capacity ramping up through September 2026 at a reduced fee. If the deal runs to term, the contracted value is approximately $30.4 billion — the largest single AI infrastructure lease ever publicly documented.

What the filing actually says

SpaceX's Form S-1 Amendment No. 2, filed June 5, 2026, describes a "cloud services agreement" under which Google will use about 110,000 Nvidia GPUs, as well as central processors, memory, and other components housed in SpaceX's data centers. The filing states that if SpaceX fails to "deliver access to the committed amount of GPUs by September 30, 2026," Google can immediately end the agreement or accept the number of GPUs provided at a reduced fee after a one-month grace period. After the first year, either party can terminate with 90 days' notice.

The compute originates from SpaceX's Colossus 1 data center in Memphis, Tennessee — the facility originally built for xAI's Grok workloads. SpaceX merged with xAI in February 2026 in a transaction valuing the combined entity at $1.25 trillion. In Q1 2026, SpaceX's AI segment recorded an operating loss of ~$2.5 billion on just $818 million in revenue; capital expenditures hit $10.1 billion, with $7.7 billion committed to AI. This Google deal, plus a reported $1.25B/month Anthropic agreement, puts committed monthly compute revenue at ~$2.17B.

Why Google needs a bridge

Google Cloud told CNBC the deal was made "to ensure we have bridge capacity to meet surging customer demand for our agent platform, Gemini Enterprise, which has been even higher than we expected." Google introduced Gemini Enterprise subscriptions for large businesses in October 2025. Separately, Alphabet revised its 2026 capital expenditure forecast to $180–190 billion (up from $175–185B) and announced plans to sell $85 billion in stock — including a $10B investment from Berkshire Hathaway — to fund the AI buildout.

The irony: five years ago, Google Cloud won a deal to supply computing resources to SpaceX for Starlink. Now the roles are reversed — SpaceX is the compute landlord, Google the tenant. SpaceX's prospectus explicitly names Google as a competitor in both connectivity (Starlink vs Google Fiber) and AI (xAI/Grok vs Gemini).

What this signals for the neocloud market

The deal puts SpaceX in direct competition with neoclouds like CoreWeave and Nebius — companies that rent GPU capacity to AI labs. Those stocks initially sold off on the announcement. But SpaceX's entry as a lessor is also a hedge: the Colossus facility was built for xAI's own models (Grok), which have struggled to dent the market dominated by OpenAI, Anthropic, and Google. Leasing excess capacity to Google — a direct AI rival — converts a strategic liability into revenue.

For buyers, the deal reveals the true market-clearing price for large-scale GPU access in late 2026: roughly $8,360 per GPU per month ($920M / 110K GPUs), all-in with power, cooling, networking, and CPUs. That is a reference point for every enterprise negotiating with CoreWeave, Lambda, or the hyperscalers themselves.

Verdict For infrastructure buyers: the $30.4B Google–SpaceX contract is the new ceiling benchmark for dedicated AI compute leases. If you're signing a multi-year GPU deal, this is the number to anchor against.

For SpaceX investors: the deal de-risks the IPO by converting idle xAI capacity into contracted revenue from a blue-chip counterparty. But it also signals SpaceX is competing with its own AI customer (Google) in the infrastructure layer.

For the neocloud sector: a new, massively capitalized entrant just joined the leasing market — and its landlord is also your competitor's AI lab. Expect pricing pressure and contract restructuring through 2027.

Bottom line: Google is paying SpaceX $920M/month for 110,000 GPUs through June 2029 — a $30.4B bridge deal disclosed in SpaceX's IPO filing. The capacity comes from the xAI Colossus data center in Memphis. For Google, it's a stopgap until its own TPU and data center expansion catches up to Gemini Enterprise demand. For SpaceX, it's the revenue anchor for a $1.75T IPO. For the market, it's the largest AI infrastructure lease ever documented.

Sources

Prices and benchmark figures are standard published rates as of October 29, 2026 and can change without notice. Some outbound links may be affiliate links — see our disclosure. Found an error? Tell us and we'll correct it in public.