Meta Platforms and AI firm Anthropic are engaged in negotiations over a substantial infrastructure partnership that could see the social media company lease computing power to the artificial intelligence developer in a deal potentially valued at $10 billion across two years. The discussions, still in their preliminary stages, underline Meta's emerging ambitions to transform itself from an advertising-dependent business into a diversified technology company capable of monetising its vast computational infrastructure.
The proposed arrangement would operate on a monthly payment schedule, with Anthropic assuming the financial obligations over the two-year window. However, both parties have incorporated exit clauses permitting early termination should circumstances warrant, and the financial terms remain fluid as negotiations continue. Industry sources indicate that Anthropic, which has been preparing for a public listing, originally introduced the proposal to Meta in June, and the social media giant is actively considering the proposition.
Shares of Meta experienced modest volatility following disclosure of the negotiations, with the stock closing down marginally on Friday amid broader technology sector declines. The company's foray into computing infrastructure sales represents a logical extension of existing capabilities, particularly given the astronomical infrastructure investments Meta has committed to over recent years. For Malaysia and the broader Southeast Asian technology sector, such corporate manoeuvres between global giants signal the intensifying competition for dominance in artificial intelligence infrastructure, a domain increasingly critical to regional digital development.
Meta's potential entry into the cloud computing market directly positions it against specialised infrastructure providers including CoreWeave and Nebius, companies that have built their entire business models around provisioning computational resources to AI developers. The explosive demand for computing capacity has been driven by widespread adoption of sophisticated artificial intelligence applications, creating substantial revenue opportunities for organisations possessing substantial surplus computational resources. Meta's vast data centres, originally constructed to support its social media platforms and advertising infrastructure, contain considerable excess capacity that has remained largely underutilised.
The negotiations between Meta and Anthropic mirror a comparable arrangement that Anthropic concluded with SpaceX in May, whereby the AI company gained access to the complete computational power of SpaceX's Colossus 1 facility located in Memphis, Tennessee. That partnership demonstrated Anthropic's strategic approach to securing dedicated infrastructure resources, potentially signalling a broader pattern wherein AI developers increasingly negotiate direct access to massive computing facilities rather than procuring capacity through conventional cloud service providers. This structural shift carries implications for traditional cloud infrastructure companies and signals investor appetite for alternative models.
Meta's chief executive Mark Zuckerberg addressed the company's cloud computing ambitions during the organisation's May shareholder meeting, explicitly characterising entry into this sector as "definitely on the table." He elaborated that Meta receives regular solicitations from technology firms seeking either access to proprietary AI models or surplus computing resources, underscoring consistent marketplace demand. Such statements from leadership suggest this is not merely exploratory but represents genuine strategic repositioning, with infrastructure monetisation forming a material component of Meta's long-term growth strategy beyond traditional advertising revenue streams.
Recent reporting has reinforced these strategic intentions, with prominent financial media outlets detailing Meta's efforts to construct a dedicated cloud computing business designed to commercialise excess computational capacity and provide hosting services for developer-created artificial intelligence applications. These developments align with Meta's broader digital transformation narrative and reflect the company's recognition that its substantial infrastructure investments represent underutilised assets capable of generating meaningful revenue in a computing-constrained market.
The complicated nature of these negotiations, according to sources, stems partly from Meta's historical positioning exclusively as a consumer technology platform rather than an enterprise infrastructure provider. The company lacks established infrastructure for conducting business transactions with corporate customers seeking computing resources, requiring operational and contractual frameworks that Meta has not previously developed at scale. Building such capabilities necessitates not merely possessing computational resources but constructing entire business divisions encompassing sales operations, technical support, contractual administration, and service level management.
For Southeast Asian technology companies and government agencies pursuing artificial intelligence initiatives, such arrangements among global technology corporations underscore the regional disparity in computational infrastructure access. Malaysia and neighbouring economies remain substantially reliant on international providers for industrial-scale computing capacity, creating both dependencies and opportunities. As global technology companies increasingly commercialise infrastructure resources, regional organisations may benefit from expanded access to computational power, though pricing structures and terms will likely reflect the underlying competitive dynamics among international providers.
The discussions remain preliminary, and sources have cautioned that negotiations may ultimately prove unsuccessful, a common occurrence in high-value technology transactions. Both Meta and Anthropic declined formal comment when approached by media organisations, a standard practice during sensitive commercial negotiations. The eventual resolution of these discussions will carry broader implications for how artificial intelligence infrastructure develops globally, particularly regarding whether specialised cloud providers or generalist technology companies dominate this emerging market segment.
