Key Points
-
Google just started paying SpaceX for access to roughly 110,000 GPUs.
-
It’s saving a lot of time — but not necessarily money — with that massive deal.
-
Alphabet‘s (NASDAQ: GOOG) (NASDAQ: GOOGL) Google is one of the world’s largest cloud infrastructure companies, yet it just started paying SpaceX (NASDAQ: SPCX) $920 million per month — or $11 billion per year — to lease xAI’s cloud infrastructure through 2029. Let’s see why Google is renting that cloud compute from SpaceX rather than building more data centers — and how that unusual deal could expose it to some unpredictable concentration risks in the future.
Why does Google need SpaceX’s data centers?
Google controlled 15% of the global cloud infrastructure market in the second quarter of 2026, according to Synergy Research Group, putting it in third place after Amazon (NASDAQ: AMZN) Web Services (AWS) (28%) and Microsoft (NASDAQ: MSFT) Azure (20%).
Missed Nvidia in 2009? This Rare Signal Is Flashing Again. In 2009, a “Double Down” signal flashed for a little-known chipmaker called Nvidia. For the first time in years, that same “Total Conviction” signal is flashing for a company 1/100th the size of Nvidia. Continue »
From 2021 to 2025, Google Cloud’s annual revenue more than tripled from $19.2 billion to $58.7 billion as the cloud and AI markets expanded, and Google seems like a great vertically integrated play on those secular trends. The tech titan builds its own data centers, produces its own custom AI chips, and hosts a broad range of market-leading cloud-based services and apps. It also ties all of those pieces together with Gemini and its other AI models.
But over the past few years, Google couldn’t expand that infrastructure fast enough to keep pace with the AI market’s explosive growth. It produced enough of its custom tensor processing units (TPUs) for its own first-party cloud services, but most of its customers needed access to Nvidia‘s (NASDAQ: NVDA) graphics processing units (GPUs), which are better suited for training other AI models. Most of the top AI companies — including OpenAI, Anthropic, and Meta Platforms — use Nvidia’s data center GPUs to process their AI tasks.
Nvidia controls over 80% of the data center AI chip market, and it locks in those customers with its proprietary software and services. A large portion of Google’s servers already run on Nvidia’s GPUs, but it doesn’t have nearly enough chips to satisfy the market’s demand for Nvidia-powered cloud computing services. To fill that gap, Google is willing to pay SpaceX $11 billion per year for continuous access to a cluster of roughly 110,000 GPUs.
Is that a smart long-term decision?
Google’s deal with SpaceX instantly resolves a major bottleneck for its cloud infrastructure business. But if it had simply bought those 110,000 chips from Nvidia at $30,000-$40,000 apiece, it would have spent only $3.3 billion to $4.4 billion.
However, it would also have had to secure enough power to run those GPUs — and it can take three to five years to set up new grid connections for data centers. Google would also need to pay massive energy and labor expenses to continuously operate those centers, and the value of its existing GPUs would depreciate as soon as Nvidia launches newer chips.
Moreover, it could take a long time for Nvidia — which is still selling chips faster than it can produce them — to actually fulfill that massive order. Instead of dealing with all of that, it makes more sense for Google to just pay $11 billion (less than 2% of its projected 2027 revenue) to instantly increase its cloud computing power and keep pace with Amazon and Microsoft. Paying SpaceX, in which Alphabet owns a 4%-5% stake, also makes more sense than paying a direct competitor in which the company doesn’t own any equity.
But over the long term, this deal could expose Google to some major concentration risks. By paying SpaceX billions of dollars per year, it’s essentially funding xAI’s transformation into a future cloud infrastructure competitor in the generative AI market.
Google’s deal with SpaceX further reinforces Nvidia’s dominance of the data center market, rather than challenging it with its own TPUs. Moreover, that massive cluster is located at xAI’s centralized facilities — including its Colossus mega-clusters in Tennessee — so any power failures or disruptions could impact a large portion of its cloud-based tenants.
For now, investors shouldn’t fret too much over Google’s deal with SpaceX, since it’s a win-win situation for both companies. But it also exposes the limitations of Google’s cloud infrastructure, as well as some future catalysts for xAI’s growing cloud computing business.
Should you buy stock in Alphabet right now?
Before you buy stock in Alphabet, consider this:
The Motley Fool Stock Advisor analyst team just identified what they believe are the 10 best stocks for investors to buy now… and Alphabet wasn’t one of them. The 10 stocks that made the cut could produce monster returns in the coming years.
Consider when Netflix made this list on December 17, 2004… if you invested $1,000 at the time of our recommendation, you’d have $350,756!* Or when Nvidia made this list on April 15, 2005… if you invested $1,000 at the time of our recommendation, you’d have $1,363,905!*
Now, it’s worth noting Stock Advisor’s total average return is 922% — a market-crushing outperformance compared to 213% for the S&P 500. Don’t miss the latest top 10 list, available with Stock Advisor, and join an investing community built by individual investors for individual investors.
See the 10 stocks »
*Stock Advisor returns as of October 22, 2026.
Leo Sun has positions in Amazon and Meta Platforms. The Motley Fool has positions in and recommends Alphabet, Amazon, Meta Platforms, Microsoft, and Nvidia. The Motley Fool has a disclosure policy.
