Google is in talks with Elon Musk's SpaceX to launch orbital data centers for AI computing, the Wall Street Journal reported on May 12, 2026. According to the report, Google is also in discussions with other rocket-launch companies about similar arrangements. Both Google and SpaceX reportedly declined to comment, and the talks are described as preliminary.
The news lands at the intersection of two of the year's biggest stories: the scramble for AI compute capacity and SpaceX's planned public listing, which the company has pitched to investors at a roughly $1.75 trillion valuation later this year. Part of that pitch, per the WSJ, is the idea that space will become the cheapest place to run AI workloads within a few years — and a high-profile customer like Google would help make the case.
How this ties to Project Suncatcher
Google has already signaled interest in orbital compute through Project Suncatcher, its research initiative to network solar-powered satellites equipped with Tensor Processing Units into an in-orbit AI cloud. Google has said it intends to fly an initial prototype, reportedly with Planet Labs, around 2027. A launch deal with SpaceX — or another provider — would be the logistical backbone such a program needs.
The relationships here are already tangled. Google holds a stake in SpaceX (around 6% per recent filings, dating to a 2015 investment), and SpaceX acquired xAI earlier in 2026. SpaceX also struck a deal last week with Anthropic to tap computing resources at xAI's Memphis facility, with reported talk of future collaboration on orbital infrastructure. A Google arrangement would extend that pattern of cloud and AI players hedging into space-based capacity.
The skeptics' case
The economics remain the open question. Multiple analyses, including TechCrunch's, note that once you price in satellite manufacturing, launch costs, radiation hardening, and the difficulty of servicing hardware in orbit, terrestrial data centers are still far cheaper. The pitch leans heavily on a future in which launch costs keep falling — exactly the trajectory SpaceX is selling — and on non-cost advantages: abundant solar energy, passive cooling, and freedom from the local opposition increasingly slowing data-center construction on Earth.
Why it matters
If even a fraction of these talks turn into firm commitments, orbital compute moves from research-lab curiosity to a line item in hyperscaler capital plans. It would also deepen Google's dependence on SpaceX's launch cadence at a moment when the AI buildout is straining power grids, water supplies, and community patience worldwide. For now, it is a report of conversations — but it is the clearest signal yet that the largest players see space as a serious answer to the compute crunch, not a thought experiment.
