Meta weighs potential multi-billion AI infrastructure deals, report says, even as it considers building its own capability

Meta Platforms is exploring large-scale arrangements for AI computing capacity with two specialized infrastructure providers, CoreWeave and Nebius, according to a report carried by Yahoo Finance. The piece says Meta could spend more than $62 billion across these relationships, reflecting how quickly demand for AI training and inference is pushing buyers toward suppliers that can deliver data center capacity and GPUs at speed.

The same report also suggests a longer-term strategy that goes beyond procurement. It says Meta could eventually offer customers a version of the same service it is buying today, effectively turning a customer relationship into a potential competitive offering.

CoreWeave (NASDAQ:CRWV) and Nebius Group (NASDAQ:NBIS) are positioned in the market as AI-focused compute providers. In broad terms, companies in this niche sell access to high-performance hardware, alongside the operational systems required to run machine-learning workloads. For large platform operators like Meta, such providers can reduce time-to-capacity when internal buildouts lag demand.

If Meta pursues the capability-building path described in the report, it would enter a sector where service differentiation is often tied to speed of deployment, reliability, and cost of delivering GPU performance. It would also put additional pressure on other AI infrastructure providers that already compete for enterprise and developer workloads.

Meta did not provide details in the Yahoo Finance report beyond the scenario described, including how the spending would be structured, which specific contracts are involved, or the timing of any potential transition from buyer to provider. The report also does not specify whether Meta’s future offering would be offered to all customers or limited to certain use cases or geographies.

The broader context is that AI infrastructure is becoming a strategic lever, not just a cost line. Large firms are trying to secure enough computing throughput to keep product development and research moving while maintaining enough flexibility to manage supply constraints. At the same time, platform companies have incentives to monetize internal capabilities by packaging infrastructure as a service for others.

For investors and customers, the key uncertainty is execution. Even if Meta can secure large commitments now, building and operating a comparable service typically requires substantial engineering, data center strategy, and long-term supply contracts. The report does not spell out those implementation elements.

What to watch next is whether Meta or the counterparties disclose contract terms, capex plans, or timelines tied to these potential arrangements. Any filings, earnings commentary, or infrastructure announcements that name contract values, duration, or service scope would clarify whether the scenario described by the report is moving from possibility to concrete plan.