Meta expands AI partnership with Nvidia to bring confidential computing to WhatsApp
Meta deepens its partnership with Nvidia to bring confidential computing to WhatsApp and accelerate the expansion of AI infrastructure.
Meta has announced a major expansion of its long-term partnership with Nvidia, committing to purchase “millions” of the chipmaker’s Blackwell and Rubin graphics processing units as it ramps up its artificial intelligence ambitions. The deal includes plans to deploy Nvidia’s confidential computing technology across WhatsApp, signalling a push to integrate AI-driven features while protecting user data.
Table Of Content
The agreement marks one of the largest infrastructure commitments by Meta to date and underscores the growing reliance of big tech companies on Nvidia’s hardware and software ecosystem. Analysts estimate the deal could be worth tens of billions of US dollars, although neither company has disclosed financial terms.
Confidential computing to support AI features in WhatsApp
Meta said it will use Nvidia’s confidential computing technology to enable AI-powered capabilities in WhatsApp while maintaining privacy and data security. The approach is designed to protect data not only during transmission but also while it is being processed, a critical concern as AI systems handle increasingly sensitive information.
NVIDIA described the technology as a way for developers to keep both data and proprietary software safe. In a blog post, the company said confidential computing enables organisations “to preserve their intellectual property” while ensuring the confidentiality and integrity of user data during computation. This feature could be particularly important as Meta integrates AI assistants, automated messaging tools, and other advanced functions into its messaging platform.
The move reflects broader industry concerns about privacy as AI models become more deeply embedded in consumer services. By adopting confidential computing, Meta is positioning WhatsApp as a platform where AI features can be delivered without exposing personal conversations or data to internal or external risks.
New hardware deployments and infrastructure strategy
As part of the partnership, Meta will also become the first company to deploy Nvidia’s Grace central processing units as standalone processors rather than pairing them directly with GPUs. These CPUs are designed to handle inference and agentic workloads, enabling AI systems to act autonomously or assist users in real time.
Meta will additionally use Nvidia’s Spectrum-X Ethernet switches to connect its growing fleet of AI servers. These networking components are intended to improve data transfer speeds and efficiency within large-scale data centres, which are essential for training and running advanced AI models.
The deployment strategy highlights Meta’s intent to diversify its computing stack beyond traditional GPU-heavy architectures. By combining CPUs, GPUs, and high-speed networking, the company is building a more flexible infrastructure capable of supporting a range of AI tasks, from training large models to running real-time applications across its platforms.
Massive spending plans underline AI ambitions
Meta’s agreement with Nvidia comes amid aggressive investment plans in artificial intelligence. Earlier this year, the company said it could spend up to US$135 billion on AI in 2026 alone, with a substantial portion expected to go towards hardware, data centres, and related infrastructure.
The partnership with Nvidia is seen as a key component of this strategy. Analysts suggest the scale of the deal, potentially in the tens of billions of US dollars, represents a significant deepening of the relationship between the two companies. NVIDIA has become the dominant supplier of AI chips, and securing access to large volumes of hardware is crucial for companies competing in the AI race.
Beyond chip purchases, Meta plans to expand its global data centre footprint dramatically. The company has pledged up to US$600 billion to build as many as 30 data centres by 2028, with 26 of them located in the United States. These facilities are intended to support Meta’s AI models, social platforms, and future computing services, including virtual and augmented reality products.
Industry observers say Meta’s investment spree reflects intense competition among technology giants to build the infrastructure needed for next-generation AI services. With demand for AI chips outstripping supply, long-term partnerships such as this one provide Meta with a strategic advantage in securing critical components and deploying new technologies at scale.
The integration of confidential computing into WhatsApp also suggests Meta is seeking to balance innovation with privacy concerns, a challenge that has long shadowed the company’s products. If successful, the approach could set a precedent for how AI features are deployed in consumer messaging platforms while addressing regulatory and public scrutiny over data protection.
