TLDRs;
- Meta plans to offload $2 billion in data center assets to co-develop its AI infrastructure with partners.
- Reclassified assets mark a strategic pivot toward financial collaboration amid soaring AI-related infrastructure costs.
- CFO Susan Li confirms Meta is exploring partnerships, with no formal deals yet announced.
- Surging power demand and grid constraints push tech giants toward shared infrastructure models.
Meta Platforms has announced plans to sell approximately $2 billion worth of data center assets, signaling a major shift in how the tech giant approaches its AI infrastructure buildout.
According to a recent regulatory filing, Meta has reclassified certain land and construction assets as “held-for-sale” and aims to transfer them to external partners within the next year.
As of June 30, Meta reported $3.3 billion in assets tagged as “held-for-sale.” The move is part of a broader strategy to co-develop infrastructure with financial and development partners rather than relying solely on internal resources.
This trend is quickly gaining traction across the tech sector, as escalating costs and logistical hurdles surrounding AI data centers push even trillion-dollar firms to seek outside capital and collaboration.
The End of Self-Funding?
Meta’s decision reflects the growing realization that traditional self-funding models may no longer suffice in the age of AI. The company’s capital expenditure guidance has surged dramatically, with annual forecasts now ranging between $66 billion and $72 billion.
Meta CFO Susan Li confirmed that the company is actively exploring financial partnerships for its data center projects, although no deals have been finalized yet.
Global spending on AI-related infrastructure is expected to surpass $1.4 trillion by 2027. In 2025 alone, financing will be required for an estimated $170 billion in data center assets. These figures are stretching even the largest balance sheets, forcing a paradigm shift in how growth is financed.
Need for Collaborative Infrastructure
Beyond capital, energy is emerging as a major bottleneck. The power needs of AI data centers are skyrocketing, with some facilities requiring up to 50 megawatts of power on just five acres. U.S. demand for AI-related power is expected to balloon from 4 gigawatts in 2024 to 123 gigawatts by 2035.
Lengthy delays in securing grid access are pushing companies like Meta to pursue partnerships with firms that already have utility agreements in place.
By aligning with third-party developers, Meta can share the burden of both financing and energy sourcing, potentially speeding up timelines and minimizing solo infrastructure risk.
Infrastructure Remains Costly
Despite the high costs, Meta is already seeing returns from its AI investments in other areas. The company recently reported stronger-than-expected ad revenue, attributing the growth partly to AI-powered improvements in ad targeting and performance.
Still, the backend required to support these AI gains is becoming too expensive to shoulder alone. Meta’s planned $2 billion asset sale not only frees up balance sheet capacity but also opens the door to faster, more flexible deployment of critical infrastructure.
As the race to dominate AI accelerates, Meta’s strategic shift may serve as a blueprint for other tech firms grappling with the twin challenges of rising costs and limited grid capacity.
