Observed Signal · Apr 24, 2026 · Partnership · Source: CNBC Technology · Impact: 4/5 · Sentiment: Positive
Meta to Use Hundreds of Thousands of AWS Graviton Chips
Meta Platforms has agreed to a multi-year deal to run large portions of its AI workloads on Amazon Web Services' Graviton ARM‑based CPUs. The arrangement, reported to run at least three years, will make Meta a top‑five Graviton customer and shifts part of Meta’s inference and agentic workloads onto AWS to reduce compute costs. CNBC notes the announcement helped lift Amazon shares and highlights AWS’s broader chip portfolio — Graviton CPUs, Trainium accelerators and the Nitro EC2 NIC — as an alternative to Nvidia GPUs for certain real‑world AI use cases. Amazon CEO Andy Jassy said the company’s chips business (Graviton, Trainium, Nitro) has an annual revenue run rate over $20 billion and is growing at triple‑digit percentages year‑over‑year. The deal reinforces hyperscaler competition to supply AI infrastructure and validates AWS’s strategy to compete as both cloud and chip provider.
Major cloud-infrastructure partnership between Meta and AWS affects AI compute demand, cloud vendor competition, chip procurement (CPU vs GPU), and energy/cost efficiency considerations for large-scale AI deployments.
Track Meta Signals & Market Shifts in Real-Time
Polaris7 autonomous intelligence agents track regulatory filings, primary sources, executive changes, and deal flow 24/7. Create your free Explorer workspace to monitor these entities.
Key Takeaways & Evidence Grounding
- Meta Platforms agreed to use Amazon Web Services' Graviton ARM‑based CPUs to run large portions of its AI workloads.
- The agreement runs at least three years and makes Meta a top‑five customer of Graviton.
- Graviton are ARM‑architecture CPUs; Amazon also offers Trainium accelerators and Nitro (EC2 NIC) as part of its chip portfolio.
- Amazon CEO Andy Jassy said the chips business (Graviton, Trainium, Nitro) has an annual revenue run rate exceeding $20 billion and is growing triple‑digit YoY.
- CNBC reported Amazon shares rose roughly 3% after the announcement.
Connected Companies & Entities
5 Entities mappedOntology Mapping & Concepts
Related Market Signals & Shifts
Recent verified developments and strategic activity across this market segment.
Meta Boosts AI Power with Nvidia Chip Partnership
Meta announced an expanded multiyear partnership with Nvidia to use millions of Nvidia AI chips across its data-center build-out, including Nvidia’s standalone Grace CPUs and next-generation Vera Rubin systems. Financial terms were not disclosed, though analysts say the agreement likely runs to the tens of billions and aligns with Meta’s planned AI capex (up to $135 billion in 2026) and a broader $600 billion U.S. data-center commitment through 2028. The deal includes Nvidia’s Spectrum‑X Ethernet switches and security features (to be used in products such as WhatsApp), and marks the first large-scale deployment of Grace CPUs as standalone data‑center components. Meta and Nvidia engineering teams will collaborate on co‑design and model optimization; Meta plans to deploy next‑generation Vera Rubin systems in 2027. The announcement affected markets: Meta and Nvidia shares rose in extended trading while AMD shares fell about 4%.
Anthropic in talks to lease compute from Meta
Anthropic is in very preliminary talks to lease computing power from Meta, according to a person familiar with the matter. The discussions follow Anthropic’s recent announcement of a similar arrangement with Elon Musk’s SpaceX to use capacity at the Colossus 1 data center. A New York Times report said a potential Meta-Anthropic deal being discussed could be worth about $10 billion. The article notes Anthropic’s continued need for access to AI chips made by Nvidia and references Meta’s broader push into cloud and AI infrastructure, including hiring Dave Brown (formerly at AWS) to lead infrastructure and potential capital expenditures of up to $145 billion in 2026.
Meta plans to sell excess AI compute
TechCrunch reports that Meta is planning a cloud infrastructure business to sell access to excess AI compute and hosted models, a move that would compete with major cloud providers such as AWS, Google Cloud and Microsoft Azure. Bloomberg first reported the plans, which mirror recent moves by SpaceX/xAI to lease data-center compute to third parties (including deals with Anthropic, Google and Reflection AI). Meta has committed large capital to AI infrastructure (reported at $182.9 billion in future spending) and is reportedly considering a business unit dubbed “Meta Compute,” led by Santosh Janardhan, Daniel Gross and Dina Powell McCormick. Bloomberg says Meta may copy CoreWeave’s raw-compute leasing model and also offer access to hosted models (including its closed-weight Muse Spark). The strategy aims to monetize excess capacity as direct demand for Meta’s own AI services remains limited and as industry debate continues over compute demand and chip depreciation.
Track Real-Time Market Signals & Shifts
Set up custom watchlists to receive automated, evidence-grounded executive digests whenever material signals or shifts occur across your tracked landscape.