- Meta’s $130B-$145B AI budget contrasts with just $784M in quarterly free cash flow.
- Meta shifted 80% of its $14B data center venture to BlackRock-managed funds.
- DePIN expands GPU access, although major deployments can remain physically centralized.
Meta’s artificial intelligence expansion is becoming a test of how much infrastructure one corporate balance sheet can absorb before financing pressures reshape the broader market.
According to Meta’s financial 2026 report, the company spent $31.08 billion on capital projects and finance leases during the second quarter, while free cash flow narrowed to $784 million.
Meta’s Capital Expansion Puts Free Cash Flow Under Pressure
Per the report, Meta now expects 2026 capital expenditures between $130 billion and $145 billion, even as revenue climbed 28% to $60.8 billion. However, expenses rose faster, increasing 55% to $42.03 billion, while long-term debt reached $83.66 billion by the end of June.
Spending could accelerate further in 2027. FactSet projections cited by The Wall Street Journal place 2027 spending near $174 billion. Deutsche Bank, on the other hand, estimated $215 billion, while Raymond James projected an even higher total of $280 billion.
Those figures reflect the cost of GPUs, electricity, cooling systems, networking equipment, land, and the construction required to operate large artificial intelligence campuses. Nevertheless, to reduce pressure on its balance sheet, Meta has started shifting that burden outside its balance sheet.
Under one arrangement, BlackRock-managed funds will own 80% of a $14 billion El Paso project. The company will retain a 20% stake and lease the computing capacity once the facility becomes operational.
The one-gigawatt campus will be financed with $12.5 billion in debt and is expected to begin operating in 2028. Meanwhile, the broader artificial intelligence infrastructure boom is driving companies deeper into debt markets. AI-related bond issuance reached $270 billion by early July, according to Bank of America data cited by Reuters.
DePIN Marketplaces Convert Idle GPUs Into Usable AI Capacity
Against that backdrop, DePIN networks offer a different way to organize computing infrastructure. Rather than relying on corporate ownership, they connect independent operators through blockchain-based marketplaces.
Render Network illustrates how this model can support distributed workloads. Its dashboard records more than 77 million frames rendered and 5,600 nodes since inception, showing how geographically dispersed hardware can provide computing capacity.
Similarly, Akash uses competitive bidding among providers, allowing customers to lease available CPU and GPU resources while improving the use of scattered or idle machines. The model is also expanding into larger deployments.
Axe Compute, for example, disclosed a 36-month, $260 million contract covering 2,304 Nvidia B300 GPUs, with Aethir provisioning the cluster through its distributed cloud platform.
However, the equipment will remain inside a single U.S. Tier 3 data center, meaning the deployment retains substantial physical concentration despite its decentralized coordination layer.
Tokenized Compute Financing Broadens Infrastructure Funding
Alongside distributed computing, blockchain-based financing is creating additional infrastructure options. Tokenized compute credits, GPU-backed assets, stablecoin loans, and revenue-linked claims can help operators finance equipment and sell future capacity.
Nevertheless, these structures do not eliminate depreciation, credit, custody, execution, or legal enforcement risks. In most cases, ownership rights and contractual obligations remain off-chain.
For now, DePIN provides wider access to computing supplies and more transparent pricing for inference and batch workloads. By contrast, tightly integrated data center campuses remain better suited to frontier-scale artificial intelligence training.
Related: Meta and BlackRock Commit $14B to One of the Largest AI Data Centers
Disclaimer: The information presented in this article is for informational and educational purposes only. The article does not constitute financial advice or advice of any kind. Coin Edition is not responsible for any losses incurred as a result of the utilization of content, products, or services mentioned. Readers are advised to exercise caution before taking any action related to the company.
