The U.S. technology sector is spending at unprecedented levels to secure the chips, land, water and electricity needed for artificial intelligence infrastructure. That backdrop is driving fresh scrutiny of Meta after a July 1 Bloomberg report, cited by Reuters, said the company is developing a business to sell excess AI computing capacity as Saudi Arabia expands its own data center ambitions.
Meta’s cloud move has put data center economics back in focus
Meta is developing plans for a cloud infrastructure business that would sell access to AI computing power and models, according to a July 1 Bloomberg report cited by Reuters. The reported effort would open a new line of business beyond advertising and place Meta into more direct competition with Amazon Web Services, Microsoft Azure and Google Cloud.
The timing matters because Meta is already committing enormous sums to AI infrastructure. Reuters reported on April 29 that Meta raised its 2026 capital expenditure forecast to between $125 billion and $145 billion, reflecting higher component costs and more data center construction. Earlier, in January, Mark Zuckerberg said Meta’s new “Meta Compute” initiative would build infrastructure consuming tens of gigawatts this decade and potentially hundreds of gigawatts over time.
That scale is why investors and industry analysts are increasingly focused on where AI capacity can be built most cheaply. Fortune reported on July 2 that media executive Mark Douglas argued U.S. data center projects may face long-term pressure from Gulf-backed infrastructure, especially in Saudi Arabia, where lower-cost energy and state-backed financing could change the economics of hyperscale AI computing.
What is confirmed is Saudi Arabia’s accelerating buildout of AI infrastructure, not a Meta commitment in the kingdom. Saudi-backed AI company Humain has begun construction on its first data centers in Riyadh and Dammam and expects the initial sites to come online in the second quarter of 2026 with up to 100 megawatts each, according to a Bloomberg report cited by Reuters.
Reuters also reported in May that Humain selected Goldman Sachs to advise on a financing package for Saudi data centers worth at least 20 billion riyals. In separate reporting, Reuters said Humain, backed by Saudi Arabia’s sovereign wealth fund, launched a joint venture with AMD and Cisco that starts with a 100-megawatt Saudi project. Reuters attributed Saudi Arabia’s appeal in part to abundant land and cheap power.
The company has not announced any Meta-owned data center in Saudi Arabia, and no public filing reviewed for this report confirms a Saudi Meta buildout. What the current reporting does show is that Saudi Arabia is positioning itself to host the kind of power-intensive AI workloads that U.S. tech companies, including Meta, may need as demand for training and inference capacity keeps rising.
The broader reason this story is gaining traction is straightforward: AI data centers need extraordinary amounts of electricity, capital and specialized hardware. Reuters reported that Meta’s latest spending increase is tied directly to expanding AI infrastructure, while Zuckerberg has framed compute capacity itself as a strategic advantage for the company’s long-term plans.
Saudi Arabia is trying to turn that demand into an industrial strategy. Reuters reported that Humain’s projects are backed by the Public Investment Fund and are being pitched around advantages including available land, lower-cost energy and rapid buildout potential. Nvidia said in May 2025 that it would send an initial tranche of 18,000 Blackwell chips to Humain, underscoring how quickly Saudi projects are moving to secure advanced hardware.
For residents and businesses, the practical takeaway is that the next phase of AI competition may be shaped as much by electricity and infrastructure financing as by software. Meta has not publicly confirmed a Saudi expansion, but its reported cloud plans and Saudi Arabia’s active buildout show that the location of future AI data centers is becoming a central business question, not just a technical one.