Meta Platforms is in advanced discussions with artificial intelligence company Anthropic regarding a substantial computing power arrangement that could be valued at up to $10 billion across a two-year span, people with knowledge of the negotiations have indicated. The proposed structure would see Anthropic, the creator of the Claude Code platform, compensating Meta through monthly payments distributed over the contract period, though both parties retain the flexibility to terminate early if circumstances change. The talks remain in their preliminary phases, and sources caution that any agreement remains uncertain at this stage.
The potential partnership arrives amid a broader technology sector pullback that weighed on Meta's share performance on Friday, with the company's stock closing down more than two percent. However, the news of the computing arrangement provided marginal support to the broader decline, suggesting investor recognition of the strategic value in Meta's infrastructure capabilities. Extended trading hours saw the stock remain relatively flat, indicating cautious sentiment around the preliminary nature of the discussions.
For Meta, cultivating a computing services business represents a significant diversification opportunity beyond its traditional advertising revenue model. As artificial intelligence applications proliferate across industries, demand for specialized computational resources continues expanding at an accelerated pace. By monetizing its substantial data centre infrastructure and excess processing capacity, Meta could establish a revenue stream that insulates the company from fluctuations in the digital advertising marketplace, which has faced headwinds in recent years from regulatory pressures and platform algorithm changes.
The competitive landscape for computational infrastructure has intensified considerably, with emerging specialist providers such as CoreWeave and Nebius capturing growing market share by offering purpose-built solutions for AI training and deployment. These niche competitors have demonstrated that significant demand exists among AI developers and researchers seeking reliable, scalable computing resources. Meta's entry into this market would leverage its existing advantages in hardware design, data centre operations, and capital resources, potentially allowing the company to offer compelling alternatives to both established cloud providers and newer entrants.
Anthropic’s initiation of these discussions in June, preceding the company's anticipated initial public offering, suggests the AI startup recognises the strategic importance of securing stable access to computing infrastructure during rapid scaling. As Anthropic advances toward public markets, demonstrating secured relationships with major computing providers could strengthen its competitive positioning and improve investor confidence in its operational sustainability. The arrangement also aligns with Anthropic's broader strategy of partnering with established technology firms to support its infrastructure requirements.
The proposed compute lease structure mirrors strategies increasingly adopted across the technology sector. SpaceX notably completed a similar arrangement with Anthropic in May, granting the AI company access to the full computational capacity of its Colossus 1 data centre facility located in Memphis, Tennessee. This precedent demonstrates that major technology infrastructure operators view supplying computing power to AI firms as a viable business model. The success of the SpaceX-Anthropic arrangement has likely encouraged Meta to seriously pursue analogous opportunities with other AI companies.
Meta's leadership has signalled strategic openness toward cloud computing ventures. During the company's shareholder meeting in May, Chief Executive Officer Mark Zuckerberg indicated that entering the cloud computing sector represented a viable strategic option the company was actively contemplating. Zuckerberg noted that technology firms approach Meta with remarkable frequency, nearly weekly, requesting either access to Meta's artificial intelligence models or utilization of spare processing capacity. These overtures underscore the market demand Meta has identified for computational resources.
Recent reporting by financial media outlets has corroborated Meta's broader infrastructure ambitions. Bloomberg News disclosed earlier this month that Meta was constructing cloud computing infrastructure explicitly designed to sell surplus processing capacity and host third-party artificial intelligence models for external developers. This initiative would establish Meta as a legitimate player in the infrastructure-as-a-service market alongside Amazon Web Services, Google Cloud, and Microsoft Azure. For Malaysian technology companies and regional startups developing AI applications, Meta's emergence as a computing provider could offer additional options for sourcing computational infrastructure, potentially introducing competitive pricing dynamics and regional data residency solutions.
The structural complications noted by those familiar with the discussions centre on Meta's limited experience in commercial computing services businesses. Unlike specialists that have built their entire operational frameworks around managing external customer computing workloads, Meta would need to establish new business divisions, support structures, and contractual frameworks specifically designed for third-party infrastructure consumption. These organizational adjustments require both time and capital investment, explaining why Meta is proceeding methodically rather than rapidly committing to major infrastructure partnerships.
Neither Meta nor Anthropic offered immediate responses to requests for comment regarding the negotiations. Both companies typically maintain strategic silence around early-stage partnership discussions, particularly when formal agreements remain contingent upon regulatory approval or board authorization. The absence of official confirmation underscores that these discussions remain genuinely preliminary, and investors should exercise caution in extrapolating the current negotiations into definitive commercial arrangements.
From a regional perspective, Meta's expansion into cloud computing infrastructure carries implications for Southeast Asian technology development. Should Meta establish itself as a credible computing provider, Malaysian and regional artificial intelligence researchers and companies could benefit from enhanced access to large-scale processing capacity. Proximity to Meta's data centres and potential regional expansion could reduce latency issues currently affecting companies working with distant providers. Additionally, Meta's involvement in computing services could introduce competitive pressures that drive down infrastructure costs across the region, democratizing access to advanced computational resources among smaller enterprises and research institutions.
The broader significance of these negotiations extends beyond the immediate parties involved. They exemplify the intensifying competition within the infrastructure layer of the artificial intelligence ecosystem, where companies possessing existing computational assets recognize substantial revenue opportunities. As AI adoption accelerates globally, controlling access to computing power has become strategically equivalent to controlling access to data or algorithms. Meta's movement into this space signals that the company views infrastructure provision as fundamental to its long-term competitive positioning in the artificial intelligence-driven digital economy.
